Professional Documents
Culture Documents
Pekelis - Cybernetics A-Z - Mir - 1974
Pekelis - Cybernetics A-Z - Mir - 1974
A ‘oZ
B hktop IleKejuic. MajieHBKaa aHniiKJioneaiiji o Soatmofi KiiSepHeTiiKe
Mir
Publishers
Moscow
First published 1974
A a a 3 3 Z n n p U u ts
B 6 b II H i p P r U q ch
B B V KK k c c s m m sh
Tr g JI JI 1 T T t m m sbch
R a d MM m y y u LI H y
E e e IT H n CD$ f 3 3 e
'K >K zh 0 0 0 X X kh IO K) yu
fl a ya
Ha amjiuucKOM naune
Foreword,
to the E nglish Edition
The appearance of the English tran slatio n of the book on cybernetics makes
the author feel additional responsibility. This is because literatu re on cyberne
tics w ritten in this language is quite volum inous. I t contains books by renowned
authorities on the subject such as N. W iener, W . R. Ashby, G. Shannon, G. W al
ter, and numerous popular-science books, to name the works of J. M urphy, D. Fink,
A. M. Andrew, A. Cote as the most notew orthy.
And still the author does not hesitate to present his book to the reader because
m any chapters of the book tell of Soviet achievements in cybernetics and com pu
ter technology, and they, as is well known, are form idable. People interested
in the achievements of Soviet specialists following their original ways in various
fields of cybernetics would do well 1o learn of them from this book.
The book, moreover, differs appreciably from numerous other books on this
subject. It was the intention of the author to write an easily understandable
popular-science book which would contain some am using, or even hum ouristic,
elem ents and, at the same tim e, abide by the principle of an encyclopaedia in
the alphabetical arrangem ent and in a serious and scientifically correct exposi
tion of the subject m atter.
This, natu rally , was no easy task.
Three books had to be combined in one so th a t one could be read, the other
looked over, the th ird used as a m anual.
The first consists of short stories about the wonderful and the unusual in cyber
netics. The second—of detailed drawings. Ju st look at them and y o u ’ll see eve
rything in a nut-shell. You will be amuzed, too, I hope, by the cartoons a t the
end of each story the ideas for which have been suggested by cartoonists from
all over the world.
The th ird book is the encyclopaedia from A to Z. Read it as you please: in the
alphabetical order, or any letter you choose first; whichever way you read it
the m ain concepts of cybernetics will be revealed to you. For each lette r is inde
pendent of the rest. Combined, they tell a concerted tale of the new science.
Only the most im portant “letters” of the vast “lette r store” of the cybernetics
“ABC” have been chosen for this book. It was absolutely im possible to include
all the “letters”.
The list of term s and objects described in the A utom atics and Electronics En
cyclopaedia for specialists alone takes 100 pages of prin t, the encyclopaedia i t
self consisting of four great volumes of 500 pages each. Even the word list for the
projected sm all cybernetics dictionary (note the word “sm all”) fills 30 pages of
compact tex t.
The principal words in this encyclopaedia are chosen so as to enable you to go
over its pages from the simple to the more sophisticated w ithout destroying the
order of cybernetic “lette rs” .
Victor Peltelis
a
STUDY CYBERNETICS
H alf a century spent in the cause of science gives me the rig h t to say some words
to you.
F irstly , I would wish you to stu d y cybernetics. Look, even a sm all encyclo
paedia on cybernetics contains a w ealth of treasures discovered by the intellect
of man. How much greater are the riches of full-scale science, of its v ast domains!
So this is my advice: develop your knowledge of cybernetics.
Secondly, I would like you to love the perfect creation of m an—the electronic
com puter. These good helpers of m an h a v e n ’t yet had their last word. Perhaps,
some of you will be able to make them reveal their new possibilities heretofore
unknown.
So learn about the electronic com puter, about its history, gain skills in electro
nics, dream about future electronic com puters.
My third wish is: rem em ber, th e re ’s top and bottom in every job. It is enti
cing to be able to make one leap to the top and, having once reached it, view the
boundless expanses around you. B ut d o n ’t forget: everything new, exciting,
dazzling, if only i t ’s the real thing, is always deeply rooted. Therefore, the know
ledge of the fundam entals is an absolute m ust. T here’s nothing so dangerous in
science (and in technology as well) as superficiality.
So, study fundam entals and do not forget th a t m athem atics, inform ation theo
ry, physics, electronics, m etallography and m any other sciences are the building
stones of which the foundation of the building of modern knowledge, modern
technology has been b u ilt.
And lastly . Every job profits from enthusiasm , from purposefulness, from the
a b ility to sort out the principal. Can these be attained w ithout love for the sub
ject, w ithout fidelity to it?
Time will pass, and, m ay be, you will remember the “le tte r” of this cyberne
tics encyclopaedia which prom pted you to enter full-scale science or the world
of exciting technology.
A happy journey to you all, cybernetic scientists, program m ists, operators
to be. All the best to you!
A. BERG,
Member of the Academy of Sciences of the USSR,
Hero of Socialist Labour,
Chairman of the Scientific Council
for the Complex Problem of Cybernetics
of the Academy of Sciences of the USSR.
11
A R ules for A l l
however, the secret reached Fiore. fore the contest he learned th a t Fiore
W ith a rule for solving a problem was in possession of the late Scipio
which had defied the Arabs, Greeks, F erro’s m ethod. U ndaunted, T artag
and scholars of M edieval Europe in lia got down to work and w ithin those
his hands, Fiore decided th a t he few days came up w ith a better m eth
could challenge the m athem aticians od which enabled him to trium ph
of the world. brilliantly at the contest.
His hopes, however, were dashed T a rta g lia ’s algorithm , later per
when T artaglia responded to his chal fected by the Italian m athem atician
lenge. T artaglia, a m athem atical ge Girolamo Cardano, survives to this
nius, was confident th a t he would day as a general solution of cubic
easily beat Fiore, but ten days be equations.
1 2 3 4 5 6 7 8 9
a b k a2 2 7 .5 a 2 1 7 4 .2 b (5 + 6) VK M U i )
1 .3 7 0.81 1 5 .6 1 .8 7 5 1 .6 1 4 1 4 1 .1 0 2 1 9 2 .7 1 6 3 .9 5 0 4 8 .7 8 9
1 .3 9 0 .9 4 1 2 .7 1 .9 3 5 3 .1 3 2 1 6 3 .7 4 8 2 1 6 .8 8 0 3 ,5 6 4 6 0 .8 5 3
M 2 7 . 5 a 2 -v 1 7 4 . 2 b
V"K
The solution of the m ost difficult problem can be broken down into a number
of sim ple operations, a sequence of elem entary steps. They are described by an
algorithm .
Thus, an algorithm is a precise instruction for solving a class of problems by
means of a series of sim ple operations. In other words, it is a m anual for problem
solving. It can be drawn up as a series of concise instructions to be carried out
exactly and to the dot. An algorithm is a faithful guide th at shows the road to
be followed to solve a problem.
A good exam ple is E u c lid ’s A lgorithm for determ ining the greatest common
divisor of any two numbers a and b. It consists of five instructions:
One. Inspect the two numbers a and b. Proceed to next instruction.
Two. Compare the numbers (a equals b, or a is less th an b, or a is greater than
b). Proceed to next instruction:
Three. If a equals b, this number is the greatest common divisor. If a is not
equal to b, proceed to next instruction.
Four. If a is sm aller th an b, change their places. Proceed to next instruction.
Five. Subtract b from a. Inspect the subtrahend and the rem ainder. Proceed
to instruction two.
Thus, after carrying out all the instructions, one m ust return to the second,
then the th ird , fourth, and so on, u n til the numbers are equal. Then the job is
done. Try and find the greatest common divisor of, say, 21 and 14, using these
instructions.
“First I m ust form ulate the pro and the problem ’s solution broken
blem. Given: a pot, a gas cooker, a down into an ordered sequence of
quart of w ater and a package of de precisely defined operations, all th at
hydrated soup. Required: to cook the rem ains is to faith fu lly carry out the
soup. The problem can be solved by instructions. Anyone can do this wor
means of a certain algorithm z: pour king alm ost m echanically.
the w ater into the pot; place the pot M echanically? But if t h a t ’s the
on the cooker; light the gas. W hen case, c a n ’t the work be entrusted to
the w ater boils, add the dehydrated a machine?
soup; ten m inutes later tu rn off the It can, and electronic com puters
gas.” are the answer. Nowadays scientists
“W hat if the pot is already on have learned to autom ate the solu
the cooker?” tion of any problem for which an
“This introduces a com plication, algorithm exists.
b u t the problem rem ains solvable.
In this case the algorithm z1 must The long history of m athem atics
be carried out: sw itch off the gas, is essentially a quest for algorithm s.
remove the pot from the cooker; pour Every new algorithm means new so
out the w ater. The new problem has lutions of problem s. The sim pler and
thus been reduced to the old one, shorter an algorithm , the shorter is
which I know how to solve. Hence, the the road to the solution of the m athe
new one can be solved, and I can m atical m ysteries concealed behind
cook the soup even if the pot is a l m any-tiered form ulae and equa
ready warming on the fire.” tions.
You m ay well ask whether i t ’s The elaboration of algorithm s spe
w orth w asting one’s tim e on such a cifically suited for problem -solving
cumbersome set of instructions to de by m achine is of prim ary im portance
term ine the greatest common divisor in our com puter age. A lgorithm s are
of two sim ple num bers. Perhaps not. of prim e im portance in com puter m a
B ut there are other, more complex them atics—in fact, they are com pu
problem s, and solving them requires ter m athem atics.
knowledge of an appropriate algo The greater the advances of com
rith m and how to use it. puter m athem atics and the more wi
As a guide to action every algorithm despread the use of com puters in all
m ust m eet certain requirem ents. Thus, spheres of life the more im p o rtan t is
it m ust be applicable not for the so the task of discovering algorithm s for
lu tio n of ju st one problem , b u t of solving large series of problems. W ith
all problems of a given type. Its such an algorithm a com puter can be
use as a guide to action would be programmed so th a t it can solve any
ju st about nill if it could be used or all of the problems of the series,
only for one pair of numbers such as as the case m ay require. The im por
21 and 14. tance of comprehensive algorithm s is
Discovering and form ulating an a l enhanced by the fact th a t com puters
gorithm requires extensive knowledge calculate very sw iftly and in tim e
and much hard creative work. But will work even faster. It is better
when the algorithm has been found to discover a general method of sol-
a l g o r it h m
ving a large num ber of sim ilar pro solution for every problem , to be
blems, leaving the actual calculations calculated w ith or w ithout a com pu
to com puters, than to work out a ter.
Men have endowed autom atic systems w ith alm ost lim itless capabilities. To
begin w ith, they have provided them w ith “sense organs” —sensors which regis
ter changes in physical or chemical state and tran sm it their findings to the auto
m atic device.
A lim it sw itch (1) “feels” a workpiece and when the piece is finished stops
the m achine tool. A photoelectric cell (2) lights a river buoy when it gets dark.
A sound sensor w ill slide open the doors of a fire departm ent at the sound of a
siren. There are sensors to indicate when the concentration of noxious gases in
a m ine approaches the dangerous level. There are “ta s te ” sensors which control
solution concentrations and signal if an acid or other chemical has to be added.
There are also sensors which register position relative to the force of gravity (4)
and ensure sta b ility at rest and in m otion. Other sensors determine tem perature
or pressure (3) with high accuracy, “see” invisible infrared, u ltrav io let, roent
gen or cosmic rays. Still others measure the intensity of electrom agnetic fields,
the concentration of ions, the slightest motion of the air, the im pact of dust mo
tes, the motion of electrons and m any other things to which the hum an senses of
sight, hearing, touch, taste, smell are insensitive. They are capable of penetra
tin g virtu ally everywhere to relay signals carrying precise inform ation about how
a machine is functioning or how a complex technological process is proceeding.
AUTOMATICS 17
Sensors are elements of prim ary im portance in all autom atic system s. They
are being steadily perfected and made evermore sensitive so th a t they can pick
up the weakest signals. B ut then, these signals m ay be so weak th a t the devices
are incapable of reacting to them .
So the next elem ent in m ost autom atic systems is the am plifier or booster,
of which there are m any types: electronic (1), m agnetic (2), hydraulic, pneum atic,
pneum oelectric, electrom echanical (3). Common to them all, irrespective of de
sign is their purpose of transform ing weak input signals into strong output sig
nals. A ll am plifiers, as a rule, are capable of handling energies m any tim es in
excess of the control energy. Some am plifiers, electronic, for instance, are more
sensitive, others are more reliab le —m agnetic, for example.
Amplifiers of different types m ay be coupled to operate together, say an electro
nic am plifier picks up a very weak signal, amplifies it somewhat and feeds it into
an electrom echanical am plifier, which converts it into a very strong signal indeed.
One note on term inology before proceeding further w ith some of the basic prin
ciples of autom atics: the output circuit of a sensor or am plifier is known as the
control circuit. It governs, regulates or controls the controlled circuit.
Often a smooth in p u t signal is required to produce a trigger-action response
in the controlled circuit. This is achieved w ith the help of relays. They can be of
different types: m echanical, electronic, electrom echanical, photoelectric, etc.
For instance, when you drop a coin in to a public telephone w hat you are doing
is sw itching on the telephone line by means of a grav itatio n al m echanical relay.
An im portant characteristic of any relay is its so-called response tim e: the
tim e it takes to produce a change in the circuit it controls. The slowest are mecha
nical relays, w ith speeds of tenths of a second. Electrom echanical relays are fas-
18 CYBERNETICS A TO Z
ro tatio n speed. A more sophisticated device is the autom atic voltage regulator
housed in a little black box under the bonnet of every car. An autom obile gene
rato r is driven by the engine whose speed varies over a wide range. W ithout the
regulator the generator would not recharge the storage b a tte ry at low speeds
while at high speeds the voltage would soar, damaging the electrical equipm ent.
Today, autom atic ind u strial installatio n s predom inantly employ electronic
regulators. In these devices the controlled param eter is expressed in term s of
electric current of corresponding voltage. The current is amplified and compared
in a m easuring u n it w ith the voltage of the program ming u n it. The required
adjustm ent is fed into the am plifier and from there to the effector un it.
A new device for regulating the tem perature of overheated steam in a high-
pressure boiler at a therm al electric sta tio n has considerably reduced the response
tim e and cut fluctuation by half, keeping the tem perature steady to a high degree
of accuracy.
F inally, the fourth “profession” of autom atics is control. Its im portance can
be shown on the example of the operation of a heavy-duty rolling m ill—a giant
m achine for m aking rails, beams, strip m etal and other rolled stock driven by
dozens of dozens of electric m otors w ith a to tal power of tens of thousands of kilo
w atts. The operation of all its m otors, big and sm all, m ust be coordinated w ith
split-second precision. Some five thousand switchings of the motors m ust be
performed in the course of a single run of the rolling m ill. This is done by th e
autom atic control system .
Offshoots of autom atic regulation and control are tracer control and tracking.
Tracer control is employed in autom atic copying m achines, tracking system s
are used to keep telescopes locked on to a certain sta r for continuous observation,
and various com binations of the two are b u ilt into electronic sim ulators.
22 CYBERNETICS A TO Z
A technological miracle of our age: automatic docking of two Earth satellites in space.
tes circling the Earth at eight kilo ters of the two vehicles and feeds
metres per second towards each other the relevant data into the active
and locked them together. one’s attitude control system. Gra
Of the two vehicles, one is “active”, dually the active vehicle aligns pro
the other is “passive”. Both are pro perly with the passive one and draws
vided with “sense organs” enabling slow ly closer. When they are 300
them to “see”, “hear” and “recogni metres apart the low-thrust docking
ze” each other. Their antennas emit motors fire, nosing the active vehicle
a steady stream of homing signals. up to the passive module’s docking
When the active vehicle picks up the collar. The two touch, triggering the
signal it approaches the passive one, circuit of the latching system, which
which is beaming back response clamps them tightly together.
signals. After a while, on command from
As soon as the radio link is estab the Earth, the two vehicles separate
lished the automatic homing system and are brought back to Earth indi
measures the relative orbital parame vidually.
24 AUTOMATIZATION
These pictures illustrate how man gradually freed himself of the functions of direct control: ►
(1) originally the operator had to provide the power to drive his machine tool; (2) then he in
vented power drives of different kinds; (3) later he added a tool holder; (4) automatic tracer-
controlled machines have freed man of direct control functions altogether; (5) at an automatic
factory the whole production takes place without direct participation of man.
26 CYBERNETICS A TO Z
W ith autom atization all this is done course, not. This has to be done by
by m achines. Moreover, autom atic autom atic devices. Thus, whenever
m achines which perform different ope hum an sense organs are incapable of
rations constituting a sequence are accurately determ ining a required pa
integrated into production lines, w ith ram eter or q u a n tity or where hum an
autom atic handling of the workpieces reaction is too slow to spot a change
between operations so th a t a hum an in a production sequence an auto
hand never touches them . m atic device is a m ust.
W hen it is tim e for autom atization, And there is a th ird consideration.
w hat are the considerations th a t fa Many industries are harm ful to m an.
vour it? A t chemical plants producing sulphu
One we know: mass production. It ric acid, tex tile dyes or fertilizers
is always much easier to m ake ten some operations present h ealth hazards
identical objects th an ten different and some interm ediate products m ay
ones, for the sim ple reason th a t they be toxic or explosive. I t is here th a t
involve the same operations, the same autom atic m achines are indispensable.
dim ensions, the same sequences. And Cybernetics plays a trem endous
w hat if not tens, but tens of thousands p art in the autom atization of pro
and m illions are needed? U niform ity duction processes. It enters autom a
of output is one of the m ain conside tizatio n not only in the form of in
rations in favour of autom ation. d u strial electronic com puters b u t also
There is another. in the form of instrum ents th a t do
Very often, and not only in aca not catch the eye b u t are, in fact,
demic laboratories b u t in factory shops extrem ely sophisticated.
as w ell, it is necessary to measure, We shall attem p t to exam ine how
regulate and control param eters lying two very different devices used in the
beyond the threshold of hum an sense autom ation of certain technological
organs. Indeed, can m an detect elec processes function. This is the dom ain
tric field in te n sity or radiation? Of of in d u stria l cybernetics.
M a g n e t i c d ru m
An automatic device for measuring the thickness of steel sheets, sorting and depositing them
into different bunkers.
A gauge continuously measures the thickness of the metal strip. Its readings
are converted into an electric signal which is duly encoded and recorded on the
magnetic drum which revolves in step with the roller conveyer. The recording
on the drum approaches the first pick-up head at precisely the same moment
as the corresponding sheet approaches the first bunker. If the sheet passes on,
it w ill approach the second bunker just as the recording on the drum reaches the
second pick-up head, and so on.
Each pick-up head is connected with a decoding unit which responds to only
one signal: the one corresponding to the sheet thickness for the given bunker.
When the signal matches, the device opens the bunker hatch, and the sheet sli
des in.
28 CYBERNETICS A TO Z
A nother example is one of the possible ways of autom ating the process of cu t
tin g dough for baking bread.
Every baked loaf m ust be of the same w eight. B ut the density of the dough
varies, and pieces of the same size m ay be of different weight. The sim plest thing,
it would appear, is to autom atically weigh every baked loaf and control the dough
cutter accordingly. This approach, however, overlooks the fact th a t a baked loaf
m ust necessarily be weighed quite some tim e after the chunk of dough th a t went
into its m aking was cut.
In other words, the cutter should be controlled not according to the weight of
every in dividual loaf but according to inform ation averaged over a large number
of loaves. Thus even such an apparently sim ple job involves the use of not-so-
sim ple statistics.
The autom atic system m ust be provided w ith a special com puter. As each loaf
is autom atically weighed the figures are fed into the com puter, which keeps up
a running com putation, controlling the work of the dough cutter accordingly.
The mechanism s described above step forward from autom atic m achi
were exam ples of autom ation of se nes to the autom atic transfer of m a
parate processes. In recent years these terials and pieces from one machine
ideas have been increasingly expan to another w ithin a shop or a section
ded into concepts of w hat could be of a shop.
called integrated and full-scale auto Full-scale autom ation envisages a
m ation. system of autom atic m achines and
Integrated autom ation represents a m easuring, handling, transfer and con-
AUTOMATIZATION 29
tro l units and devices which turn to m anagem ent and adm inistration?
out an item or product from s ta rt The answer is an em phatic yes.
to finish w ithout direct hum an par In fact, nowaday it is impossible to
ticipation. im agine the economy being run w ith
The difficulties involved in such out autom atization. Why? If only
an enterprise are enormous, w hat because in our country some years
w ith the volum e of m aterial th a t ago there were three m illion per
m ust be handled and m achined, the sons employed in the accounting ser
num ber of technological operations vices alone. A to ta l of 10 m illion
th a t m ust be carried out, the am ount people are employed in adm inistra
of inform ation th a t m ust be received tion and m anagem ent. Ten m illion
and processed. economists, planners, rate-setters,
Y et such an autom ated g ian t has bookkeepers, accountants, design
been designed at the A utom obile In ers!
dustry Technological Research In sti And the scale of the economy and
tu te in cooperation w ith the L ikha the rates of developm ent are expan
chov Motor W orks in Moscow. Their ding w ith each passing year. The
system involves a com puter centre to flow of inform ation—plans, progress
which is fed all p ertinent inform a reports, technical specifications, bills,
tion about the m anufacturing pro invoices, and w hat have you—de
cesses going on in the different shops, m ands th a t it be processed in the
about the m ovement of workpieces shortest possible tim e.
from shop to shop and section to The rise in the num ber of adm inist
section, about the functioning of the rativ e and m anagerial personnel could
m achines and transfer lines. Each well lead to a situ atio n in which
shop has its own operational control half the nation was employed in
u n it where all the elem ents of the m anagem ent, w ith a corresponding
production process are continuously reduction in the num ber of people
controlled, from the stocks of work- directly engaged in production. H en
pieces and tools to the rate of flow ce, w hat is needed is not greater
of finished item s to the next shop. num bers b u t a q u alitativ e leap. This
The tasks of the factory com puter is being achieved by the a u to m a ti
centre also include scheduling of pro zation of m anagerial processes.
duction plans for every shop and Let us try and define the tasks of
section and continuous inventory con m anagerial a ctiv ity . N orm ally these
trol . are the production, transm ission, sto
This autom ation system has pro rage and processing of inform ation.
ved its w orth, and sim ilar systems In other words, it is the job of kee
have been installed a t the Moscow ping track of production, q u a lity
Compact Car W orks, the Zavolzhsky control, draw ing up production spe
Engine W orks, the Y aroslavl Engine cifications, duplicating and d isp a t
W orks and other factories. ching documents, coordination of v a
A utom ation of production yields rious production sections, registra
an enormous growth in labour pro tion and sorting out of documents,
ductivity. This goes w ithout saying. com puter w ork—quite a job, in
But can autom ation be of any use short.
30 CYBERNETICS A TO Z
stage doors closed autom atically, the Of m ajor im portance was the in~
scenery changed, and the doors ope vention of the autom atic lath e scale.
ned again. The R ussian m echanic Andrei N artov
In the following two thousand years and then the Englishm an H enry
ingenious mechanics designed and bu M audslay made the cutter move au
ilt clever autom atic toys, m ostly for tom atically along the workpiece,
no better purpose than the enjoym ent which could now be m achined w ith
of the rich. out having to be directly handled
However, there was much more to by the operator.
them , if only because the m echani The shrill cry of w histles heralded
cal wonders of the past indicated the the advent of steam and the steam
road along which autom atic systems engine, which replaced draught ani
were later to develop. They were m als—horses, oxen, mules, asses—as
based on the achievem ents of techno a convenient source of energy.
logy of their tim e, the age of energy B ut the age of steam would not
of running w ater, wind or compressed have come if the demands of the
springs. tim e had not driven the talented
Springs were used to drive in tri R ussian mechanic Ivan Polzunov to
cate systems of cogwheels, levers, devise a float-type controller for his
connecting rods, cams, wormscrews “fire engine” so th a t “w ater, fire and
and other parts of autom ata. T hat is steam could sustain them selves in
why it can be said th a t spring-dri m otion” . Steam engines would hardly
ven m echanical toys were b u ilt on have ever found such universal ap
the clock principle. plication as they once enjoyed if
According to K arl Marx, clocks technological advance had not led
and watches, which rank among m a n ’s the English inventor Jam es W a tt to
best technological creations and were the invention of the centrifugal speed
the first autom atic systems b u ilt for governor.
practical purposes, suggested men the Man became stronger than ever
idea of introducing autom ata in before. The new m achines were qui
in d u strial m anufacture. I t is not ac ckly p u t to work and made to per
cidental th a t the Frenchm an Ja c form a wide range of jobs.
ques Vaucanson, celebrated m aker of The governors and regulators used
m echanical toys, was the one who in the first steam engines proved to
b u ilt a weaving loom which served be extrem ely versatile. They were
as a prototype for the autom atic adapted for use in m any different
loom designed by A ntoine Jacquard. kinds of mechanisms. G radually, w ith
V aucanson’s m achine replaced fifty the advent of the age of electricity
weavers. Less than tw enty years later w ith its high power, speeds and high
autom atic looms were introduced at precision, autom ation spread to all
m any silk m ills in France and Eng technological spheres. It became the
land. O ther industries were quick to m ainstream of technological advance.
follow suit. Bigger, better and more Engineers classify the v ast variety
sophisticated m achines were in tro d u of autom atic system s in use today ac
ced in m etal-working, m ining and cording to their purpose, designation
other fields. or sphere of application as tech nolo-
AUTOMATON 33
L et’s see how an autom atic system works, a bolt-m aking tool, for exam ple.
W ith a quick m otion a lever pushes out a brass rod from which a bolt is to be
machined to ju st the required length. A tu rn of the holder and the blank is brought
up to the cutting tool. A golden shaving curls away from the blank. W ith in se
conds it is m achined along the whole length. The cutting tool retreats, the hol
der is turned again and new tools cut the thread, champfer, finish the head. A few
seconds more and another cutter cuts the finished bolt from the rod. W ith a slight
clank it drops into the m etal bunker on top of a pile of other identical bolts w hi
le the m achine is already busy m aking another one.
Despite the trem endous v ariety of m odern autom ata, they all share in common
a peculiar superficial tra it: they all create an impression of intellig en tly per
form ing their jobs, going about them quite independently and on th eir own,
w ithout any outside interference.
W ell, you m ay say, this is, after all, a superficial im pression. But w hat makes
an autom atic system so independent, sophisticated in its performance and self-
contained in its behaviour?
One of the prim ary features of an autom atic system is its data-handling cha
racteristics.
W hatever an autom aton does its performance is controlled by a programme
prepared well in advance. In a m etal-cutting tool, for exam ple, autom atic con-
3 -6 1 6
34 CYBERNETICS A TO Z
This automatic system is capable of self-regulation to meet the working requirements in the
course of operation.
involves w hat is known as closed-loop, feedback, control system w ith two data
input routes.
The age of cybernetics has seen the appearance of third-generation autom ata
capable of mem orizing past performance and experience, analysing previous
and current operation and working out an optim um programme which can be
changed and adjusted in operation to m eet changing conditions. The principle
of operation of such systems is shown schem atically in the drawing given
above.
The u ltim a te purpose of the operation—the expected o u tp u t—is defined in
the programme originally fed into the system . The relevant inform ation is carried
along the first data route. The second data route carries the working programme
issued by the working m emory block which controls the system ’s operation
3*
36 CYBERNETICS A TO Z
The th ird data route, as the diagram shows, carries the inform ation fed back
from the m easurem ent block.
There are autom atic systems w ith a greater num ber of data routes for carrying
relevant inform ation, and their performance capacity is, of course, correspondingly
greater.
The world of autom ata is great A utom ation theory m ust, for exam
indeed, and it is continuously expan ple, show how to determ ine the s ta
ding. This, understandably, has cal b ility of systems so as to prevent
led for the creation of a com prehen deviations in operation. It m ust in
sive scientific theory to define the vestigate the sen sitiv ity of autom ata,
laws according to which it functions. since the properties of a control sy
This theory rests on the firm foun stem vary in the course of operation.
dation of the old classical theory of There are m any other complex pro
autom atic control. Now it is called blems which the theory of autom atic
upon, however, to solve numerous control attacks and solves, thereby
problems of extrem e com plexity nee prom oting the advance of autom ation.
ded to design and build reliab ly
functioning autom atic systems.
37
Now take the binary number 1111. The first num eral at the extreme rig h t is 1.
The next position is only two tim es the first, hence it represents 2; the th ird po
sition is two tim es the second, and it represents 4; and the fourth position is,
accordingly, 8.
Let us try to represent, say, the number 1017 in the binary system . For this,
as in decim al notation, we count off the positions. We s ta rt w ith the lowest, 7.
The number 7 comprises four plus two plus one: 7 = 4 + 2 + l . This can be w ritten
down as follows: 1 X22+ 1 X2x+ 1 X2°. Accordingly, in each of the positions
we w rite down a 1, which gives 111.
The next number is ten, which is made up of eight plus two, viz., 10 = 8 —2 =
= 1 X23+ 0 X 22+ 1 X21+ 0 x 2°. You see th a t there are no ones or fours, and ac
cordingly we w rite down zeros, obtaining 1010.
Continuing the operation of reducing decim al to binary, we can w rite down
the num ber 1017 as 5 1 2 + 2 5 6 + 128+ 64+ 32 + 1 6 + 8 + 1 = 1 X29+ 1 X28+ 1 X 27---
1 X26+ l X25+ 1 X24+1 x 2 3+ 0 x 22+ 0 x 21+ l X2°. Expressing eacli position of
a num ber correspondingly in terms of 0 or 1, we obtain the binary for 1017, viz.,
1111111001.
You m ay wonder, of course, how one can perform the basic a rith m etical opera
tions w ith such unw ieldy numbers.
The rules are essentially the same as in the decim al system , though w ith some
peculiarities due to the fact th a t b inary notation is based on only two d igital
values.
Thus, the rule for carrying o u t,b in a ry calculations can be stated as follows:
“As in binary notation there are only two num erals to denote the coefficients
of corresponding powers of 2, in any position 1 added to 1 yields 0, w ith 1 carried
to the next position to the left. W hen a column of l ’s is added, the coefficients 0
and 1 alternate, w ith 1 being carried to the left for each change to 0 .”
Let us carry out a sim ple addition operation according to these rules, nam ely,
1 + 1. In binary we have:
, 0001
^0001
0010
Now let us add 8 + 3 in binary:
1000
0011
1011
Addition M ultiplication
0+ 0= 0 0+ 0= 0
0+1 = 1 0X 1 = 0
1+ 0 = 1 1X 0 = 0
1+ 1= 10 1X 1 = 1
38 CYBERNETICS A TO Z
began to look for ways of reducing tion of all numbers to the sim plest
the num ber of switches. elem ents of 0 and 1."
The engineers and m athem aticians A t L e ib n itz ’s request a medal was
who tackled the job proceeded from struck to honour the “dyadic system ”,
the tw o-state—b in a ry —nature of the as it was then called. It depicted a
basic elem ents of com puter hardw are. table of numbers and sim ple opera
Take, for exam ple, a neon lam p. tions w ith them . Around the rim
It can only be in one of two states: ran the inscription: “U nity can suf
it is either “open”, and conducts fice to derive all from m inuteness.”
electricity, or it is “closed”, and Subsequently the binary system
carries no current. The flip-flop was all but forgotten and for alm ost
sw itch, as its very name im plies, 200 years h ard ly a work dealt w ith
can also be only in one of two stable the subject. It was only in 1931 th a t
states. Memory elem ents, loo, are several possible practical applications
binary. of binary notation were suggested.
So why not employ the binary sys Like decim al counting, the binary
tem of num erical notation, which system obeys rigid laws. B ut in the
uses only two num erals 0 and 1, former each place of a num ber cor
m aking it extrem ely suitable for elect responds to a power of ten, while in
ronic machine operation. the la tte r it corresponds to a power
Accordingly, new machines were of two. Accordingly, in decim al no
b u ilt to perform their calculating tatio n there are ten digits from 0
operations w ith only zeros and ones. to 9, while in binary there are only-
Not th a t binary notation is a con two, 0 and 1.
tem porary of electronic computers. Num bers in binary n o tation a t first
In fact, people had been dabbling seem queer indeed to the unaccusto
in binary notation from as far back med eye, as suggested by such a
as the la tte r p a rt of the 16th century brief biographical statem ent:
to as recently as the beginning of “A t school he was extrem ely cle
the 19th century. ver w ith num bers. Problem s which
The celebrated G ottfried W ilhelm his schoolm ates solved in half an
Leibnitz considered the binary sys hour took him no more th an 101 or
tem sim ple, convenient and b e a u ti 110 (5 or 6) m inutes to solve. Thanks
ful. “In dyadic com putation” , he to his wits and his energy he finish
declared, “the rewards for its pro ed his college course in 11 (3) years,
lix ity are new discoveries of funda and he was only 1,010,000 (20) when
m ental value to science__ R em ar he was placed in charge of a research
kable order is achieved by the reduc lab o rato ry .”
M athem aticians claim th a t the arithm etical operations are much sim pler in
the b inary than in the decim al system . Even extracting a square root is sim pler.
If you practise for a w hile in binary arithm etic you w ill surely agree w ith them.
There is one stickler, though: subtraction is performed by means of addition.
This is done by adding the complement of the subtrahend (i.e. the num ber th a t
complements it to the next order of m agnitude) to the m inuend.
Suppose we w ant to subtract 101 from 10011 (19 — 5). First, add two 0 ’s to
the left side of the subtrahend to obtain the same num ber of digits as in the m i
nuend. The complement of 00101 to 100000 is 11011. Adding th is to the minu
end, we obtain:
, 10011
+ 11011
101110
BINARY SYSTEM OF NOTATION 41
D iscarding the 1 on the extreme left, we obtain OHIO, i.e. 14. And, as we know,
in decim al notation 19—5 = 14.
M ultiplication is reduced to w riting down the m ultiplicand in a staggered
column and then adding up the column. Thus,
v 11011
x 1101
11011
11011
11011
101011111
D ivision is achieved hy a process of consecutive subtraction, replaced, as we
know, by addition.
As you see, all m athem atical operations in binary are reduced to the sim p
lest arithm etical operatio n of addition.
Thus, we now know the advantages they are not intim idated by the long,
of the binary system , which at first dreary rows and columns of zeros and
glance seems so strange because of ones th a t obey the tedious rules of
our decim al trad itio n . It only re binary arithm etic. Machines race
m ains to sta te the sy stem ’s draw through endless ranks of zeros and u n i
back: the binary system is used only ties w ith incredible speed and easily
in com puters, only they reign u n overcome this shortcom ing of the b i
disputed in its dom ain. To be sure, nary system.
42 BIONICS
Visual perception: the frog’s eye perceiving the object takes in its various qualitative features
separately: contours, curvature, brightness.
Im age on the
re tin a
BIONICS 45
a certain level, the threshold. The process involves accum ulation and build-up
of signals in tim e and in space. The num ber of signals a neuron can receive may
vary over a wide range from only a few to several thousand.
The sim ulated neuron has told researchers m any things. To begin w ith, the
engineer in building a control system will like to have the flexibility, reliab ility
and efficiency of a neuron. B ut, on the other hand, an engineering system has
no use for such a property of the neuron as its need for a relatively prolonged
‘'rest period” to “restore its stre n g th ” . A technological system does not need it,
and this is its advantage.
W hat practical results has bionics stop-over. And every year they fol
yielded, w hat are its contributions low the same routes.
to technology? flow do they take their bearings
Bionics experts are am azed, for in space? How do they find their
exam ple, at the rem arkable naviga invisible routes in the sky? W hat
tional ab ilities of birds. A common are the highly accurate and sensitive
carrier pigeon, for exam ple, returns “navigational instrum ents” th a t fu n
home from practically any place. It ction inside these cham pion naviga
has been shown th a t the unassum ing tors?
little bird golden plover can cross So far scientists are unable to ex
the A tlan tic from New Scotland to plain how th is highly sophisticated
South America (a distance of some orientation system operates. Howe
4000 kilom etres) w ithout a single ver, we are en titled to expect th a t
the question w ill not rem ain unans cally w ith the greatest of ease, which
wered. This is confirmed by the in i interest the transport engineers.
tial results of some extrem ely in te Japanese engineers and biologists
resting experim ents being conducted established after m any experim ents
w ith birds. th a t whales had a better shape than
On the other hand, the rem arkable the best-designed modern ship, and
a b ility of bats to find their way about used their newly found knowledge to
the darkest caves or flit among tree build a “w hale-like” vessel. The ad
branches on moonless nights is no vantages of the new design were rea
longer a secret to bionics experts. dily apparent: w ith the same speed
They know th a t the apparently ran and load capacity as its conventional
dom, helter-skelter flitting of a bat equivalent it required only three-
out foraging for foot) is in fact an quarters of the engine power.
extrem ely accurate m ethod of sonar Bionics principles have been incor
(sound navigation and range finding), porated in the design of the “Pen
enabling the anim al to measure the guin” Soviet snowm obile. Its name
distance to its prey w ith utm ost is fu lly justified. How does a pen
precision. W hen hunting, bats are guin travel over soft snow? Sliding
extrem ely “garulous” “shooting” on its belly and propelling itself
bursts of ultrasonic rad iatio n at in w ith its flippers much like a skier
sects and picking up the reflected uses sticks. The m echanical Penguin
waves. also slides on its belly, pushing off
The study of sonar techniques of w ith spoked wheels. W eighing 1300
anim als, notably bats, has been use kilogram s, the m achine can travel at
ful in investigating the direction- 50 km per hour.
finding system blind people develop Bionics is helping chem ists. Orga
in them selves. I t has been found nic chem ists studying and creating
th a t a blind m an can use the sound polymer m aterials are interested in
of his voice to take his bearings, the “technologies” employed by na
find his way w ith o u t the help of a ture in producing complex chemical
stick and even distinguish wood, me compounds.
tal or fabrics “by ear”. The sym bol of bionics is a crossed
For m any years bionics experts scalpel, soldering iron and an in te
have been studying the speed w ith gral sign. This alliance of biologist,
which various prairie anim als, birds, technician and m athem atician offers
insects, fish and m arine anim als tra hope th a t the science of bionics will
vel. As is known, m an has long prove capable of penetrating where
since surpassed the speed records of none have penetrated before and of
the blue shark, which can swim at discovering w hat none has ever seen
70 km per hour, and the most before.
fleetfooted grasshoppers which leap The tim e is not far off when the
at speeds of 10 to 60 km per advances of bionics w ill cause m any
hour. rem arkable changes in the field of
I t ’s the problem s of m anoeuvrabi technology. We can look forward to
lity and flexible steering which birds, new breakthroughs in m ethods and
fishes and insects handle a u to m a ti tools for detecting and extracting mi-
BIONICS 47
nerals and m anufacturing m aterials. lous beauty on the sea bottom . Men will
In engineering, control system s will learn to orientate as freely in outer
incorporate w hat could be called bio space as birds do in the air, lo fore
logical m achines. cast the w eather, earthquakes and
Looking farther ahead, scientists volcanic eruptions, to litera lly grow
predict the advent of the age of bio various electronic devices, rem arkab
nics. Its contours can be surmised le biom echanism s and artificial neu
from the first advances of this dis rons, to build protein com puters....
cipline. One book describing the new The hook envisages direct transfor
science declares th a t in the future, m ation of sunlight into clothes and
following the exam ple of living na food along the lines of photosynthesis
ture, men w ill design and build orni- th a t takes place in every green leaf....
thopters, fast subm arine liners, vehic A rtificial muscles instead of cum ber
les for negotiating the rugged terrains some m achines__ A ircraft, m achines,
of the Moon, Mars, Venus and other autom obiles, rockets guided by w ill
planets, sparkling cities of dendroid power alone, w ithout steering wheels
dwellings and com m unities of fabu or levers....
48 BLACK BOX
ther it should be returned for repairs electronics, of autom atic control, me
or sim ply throw n out. dicine and economics, and the list
A sim ilar task confronts a telephone could be continued. True, the E ng
engineer whose installations are out lish scientist W alter Ashby, who in
of order but cannot be disassembled vestigated the “black box” concept,
w ithout good reason. declares th a t the purpose of his theo
Or take a physician, who in his ry is m erely to investigate the rela
practice is confronted w ith only the tionship between the experim enter
external m anifestations of a disease and the environm ent w ith special
and knows nothing of the actual state a tte n tio n to the flow of inform ation.
of the p a tie n t’s organism . He has Thus, we know th a t a “black box”
before him a “black-box” problem. is a system of the internal structure
Specialists studying “black-box” of which we are ignorant. How then
theory claim th a t its application can can it be discussed, studied, in v esti
be useful in coping w ith the trem en gated? The “black box” m ethod in
dous diversity of problems involved vestigates a system from only one
in ind u strial planning. For even an aspect: the relationship or connection
enterprise w ith only ten production between its in p u t and o u tput. The
processes has a choice of alm ost ten purpose is to com prehend the sy ste m ’s
m illion different plans. behaviour. Only its behaviour: stru c
As you see, the “black box” is a ture and m aterial are not taken into
problem of electrical engineering and account.
W illiam Ashby describes “black b o x ” investigations as follows!
A person cannot enter the same stream twice; sim ilarly, he cannot perform
the same experim ent twice. All he can do is to perform another experim ent dif
fering from the first in elem ents which can, by general agreement, be ignored.
The same is true of “black box” investigation. The basic data about the ob
served states of different parts of the “b o x ” (its “in p u t” and “o u tp u t”) are tab u la
ted in chronological order. Thus, suppose you are studying a “box” th a t has
fallen from an unidentified flying object. You could w rite down:
Time (hours) State
11.18 Did nothing—box em itted steady buzzing noise at 240 hertz
frequency
11.19 Pushed b u tto n marked “K ” —buzzing rose to 480 hertz and
continued at th a t level
11.20 Accidentally pushed b u tto n m arked tem perature of the
box rose by 20 °C
Thus, every system is investigated by chronologically tab u la tin g the observed
sequences of the states of the “in p u t” and “o u tp u t” . The record tells w hat input
leads the experim enter m anipulated w ith and w hat happened as a result at the
“o u tp u t” . By gradually expanding the “in p u t’’- “o u tp u t” scope the experim enter
gets to know the behaviour of the object. And as he gets to know more and more of
its behaviour he attem pts to reveal the inner connections of the “box” , its stru
cture.
4 -6 1 6
50 CYBERNETICS A TO Z
This is a “black box”. By repeatedly pressing the buttons the lamps on top can be made to
flash on in different combinations, making it possible to establish the laws of “input”- “out
put” relationships without knowing what goes on inside the box.
hum an nervous system w ith the same it, he is swamped by the incom ing
speed. A sim ple calculation shows inform ation and cannot grasp it
th a t this would take 40 000 years! all.
Detailed investigation of a very big W hat is the conclusion? There are
system is im possible. several. One of them is v iv id ly s ta
A nother exam ple. A utom ation is ted by the cybernetician Ashby. If an
being carried out at a chem ical works engineer building a bridge had to
in shops where h ealth hazards are consider every atom of the building
present. For a w hile everything is in m aterials his job would be im prac
order: the operator pushes the neces tical, if only because of its scope.
sary b uttons, the autom atic m achi Therefore, a builder ignores the fact
nes function flawlessly, the processes th a t beams and panels are made up
run norm ally. B ut here is a hitch: of atom s. He considers them as in
the operator is hum an and it is dan divisible u nits, because th eir nature
gerous for him to be in the shop! allows of such a sim plification, and
He, too, has to be replaced and his the engineer’s task becomes im ple-
duties handed over to an autom atic m entable.
device. B ut are sim plifications always pos
In the dry language of cybernetics sible? Not by a long shot.
this problem is defined as “the need A p sychiatrist try in g to determ ine
to replace one system by another, w hat is going on in a disordered
sim ilarly functioning one”. brain has to deal w ith a very big
This is where the “black box” system . M oreover, this is a system
comes in. We know th a t some as yet w hich dem ands a very cautious ap
undesigned system m ust perform the proach to sim plifications. They are
functions of the hum an operator. We not alw ays perm issible and can gros
ignore how a hum an being performs sly d isto rt the overall picture of the
his actions, concentrating only on disease.
w hat he does (input) and w hat comes W hen an econom ist engages in his
out of it (output). W e regard the complex calculations a t an enterprise,
hum an operator as a “black box” and the raw m aterials prices, manpower
build its model. The im portant thing requirem ents, finished goods prices,
is th a t both the box and its electro expenditure of labour, dem and and
nic replica can be described by the other indices are buried in an ocean
same m athem atical form ulae. of m inor things which m ust be taken
Here, as you can see, we again have into account. He m ust approach sim
a very big system . Such systems may plifications w ith the utm ost care, and
be so complex th a t even when comp weigh them in his calculations w ith
lete inform ation concerning the state the greatest precision. Such exam ples
of individual elem ents is available, are lite ra lly countless.
the num ber of elem ents is so great Therefore, the “black box” is in d is
th a t it is very hard to synthesize pensable in stu dying very big sys
this inform ation, to bring it toge tem s. It is the magic key of a p rac ti
ther, so as to judge of the behaviour cal m ethod which can be employed
of the whole system . A hum an inves for controlling the enormous diversity
tigator is incapable of coping w ith of very big system s.
4*
52 CYBERNETIC? A TO Z
F lip-flop type
electronic circuits Valves Semiconductors
B E G I N N I N G OF T H E
DESI GN WORK ON M O D E R N FI R ST GE N F R A T I O N S E C O N D G E N E R A T I ON
COMP U T E R S
Uniac 1946
w ith plain nails. The use of counting ductive forces the im portance of cal
boards was obviously widespread in culation grew apace. M echanical cal
R ussia. Special instructions were even culators were b u ilt. Man made them
printed, such as “Convenient Calcula count faster, gradually reducing the
tions for Use by Buyers and Sellers” degree of his own particip atio n in the
printed in Moscow in 1682, which work. L ater he added motors and
contained a table of products of whole “tau g h t” the machines to “read”,
num bers from 1 to 100. “memorize” num bers and “record”
The abacus is still used in some interm ediate results.
countries, having hardly changed It is only n atu ral th a t in the age
since ancient tim es. A high degree of autom ation calculating machines
of skill can be attained in using the have also gone autom atic. Nowadays
abacus, so much so th a t some accoun they control and regulate the whole
tan ts can compete w ith a sim ple com puting procedure operating at tre
desk calculator. mendous speeds. Their spheres of ap
W ith the developm ent of the pro plication have expanded enormously.
Great is the arsenal of modern calculating machines, numerous are the ways
of m echanizing calculations. From the desk calculator to the high-speed electro
nic com puter. From the sim plest planim eter to the most complex electronic ana
logue com puter. From a sm all accounting office to a huge computer centre.
All this has been placed at m a n ’s service, enabling him to perform m athem a
tical operations w ith huge numbers w ith great speed, accuracy and re lia b ility ,
to solve the most complex problems of higher m athem atics, to study very fast
processes.
Today a vast range of calculating m achines of all conceivable sizes and types
are m anufactured, hut they all fall into several distin ct classes.
The oldest and sim plest calculating appliances gave rise to the adding m achine.
The evolution of this branch led to the appearance of all kinds of desk keyboard
machines of the type th a t can he found in book-keeping departm ents, planning
departm ents, etc.
Another branch in calculating technology includes punched card m achines.
They employ oblong cards w ith holes punched in them . A com plete set of such
machines forms a computer flow line em ploying punched cards at all stages of
the process. Punched card machines are capable of handling tens of thousands
of cards per hour.
Complete sets of punched card m achines are the basic equipm ent of com puter
stations and bureaus. They are used at factories, offices, in stitu tes, collective
farms.
Our designers are continuously striv in g to improve designs of punched card
m achines, supplem enting and com plem enting them w ith new types and system s.
Sorting, tab u latin g and m ultiplying system s have been b u ilt which incorporate
electronic elem ents, all the efforts are aimed at expanding the data-handling
capacity of computer flow lines and increasing their productivity.
Another branch of computer hardw are is analogue machines and systems in
CALCULATING HARDWARE 59
m anufactured goods by 546 per cent in W est Germany, 505 per cent in Italy ,
and 303 per cent in Great B ritain . Only three computers were b u ilt in Japan in
1957, by 1970 their number rose to over four thousand. In 1960 there were
some five thousand computers throughout the w orld. By the end of 1967 their
number rose to forty thousand. By the end of 1975 their number in the largest
c a p italist country, the USA, is expected to reach 200 thousand, the forecast
for 1980 being 355 thousand.
The same situ a tio n exists in other industrial countries. In 1959 France had
20 electronic com puters, and by 1969 the number was five thousand. Japan had
only three computers in 1957, nowadays it has thousands.
The to ta l number of computers in 1949 could be counted w ith the aid of fingers;
by 1966 it reached 30 thousand, by 1969—75 thousand, and now it is presumed
to go into hundreds of thousands.
In the Soviet Union the m anufacture of m athem atical machines is soaring.
Taking the production for the year 1950 as 100, five years later the figure was
737, and only one year after th a t, 1171. By 1965, the output had increased 4.7-
fold over 1956, and in 1970 it exceeded th a t of 1965 by 4.8 tim es. In the next
seven or ten years the USSR w ill reach one of the first places in the world by the
production of electronic computers.
Several m ajor computer fam ilies have appeared in our country. One is the
“U ral” type, ranging from “U ral-1” , “U ral-2” , “U ral-3” , through “U ral-10” and
on to “U ral-16” . A lthough belonging to the same fam ily they serve different pur
poses and can solve a wide circle of m athem atical and logical problems.
Not long ago a new series of universal d igital computers appeared: “M insk” ,
“M insk-1” , “M insk-2” ... “M insk-22” ... “M insk-32” . This is a fam ily of small
m achines designed m ainly for solving scientific and engineering problems and
also for production planning and control. The “M insk-22” is a semiconductor
u n it intended m ainly for data processing in handling economic planning pro
blem s. W hat can it do? Calculate wages, keep warehouse tabs, compile balance
sheets, process s ta tistic a l d ata, draw up finished products specifications and do
m any other things listed in the accom panying instructions as “other economic
problem s” . Furtherm ore, the com puter can solve sim ultaneous algebraic equa
tions and perform m ultiple arithm etical operations “and other m athem atical
operations involved in solving scientific and complex engineering pro
blem s” .
As you see, this is n ’t just a m achine, but an all-purpose m athem atical combine,
a jack-of-all-trades capable of perform ing any task.
The “M insk-32” is another step forward beyond its brother 22. It calculates
five or six tim es faster and can handle eight tim es more inform ation. Furtherm ore,
it can function w ith 136 external appliances. It is operated by one person w ith
the aid of a typew riter.
Our review of sm all machines would be incom plete w ithout the universal “N airi”
designed for scientific research organizations and design offices, w ithout the
“Prom in” , a sm all desk w ith the control console on the top and the hardw are under
neath capable of solving engineering problems of average com plexity and per
form ing sm all com putations. Another interesting m achine is the “Setun” which is
CALCULATING HARDWARE 61
housed in two cabinets. Special mention should be made of the “Mir" which, in
spite of its sm all size, is capable of solving m any types of sim ultaneous equa
tions, linear program m ing problems, of calculating tim e-tables and of performing
a wide range of complex m athem atical operations, including analytical tran s
form ations of form ulae and solving equations in letter form.
Academician V. Glushkov, Hero of Socialist Labour, head of the team which
built the “Mir” , lists the characteristics of his offspring as follows. It is capable
of m emorizing 12 000 sym bols—six or seven pages of tex t. Innate to it are all
the m ain form ulae we were taught at school plus some item s from the college
course. Called the “electronic engineer” , it is an indispensable tool in the hands
of specialists employing m athem atics in their investigations.
Alas, so m any m athem atical machines are produced in our country th a t it is
impossible to tell of them all.
Before we forget, we should note one more q u a lity of the “M ir” fam ily of compu
ters: the problems are fed into them w ith the help of conventional m athem atical
signs, symbols and numbers. No translation into a m achine language is required,
which is extrem ely convenient.
Now let us proceed to the most im portant fam ily of m achines, the “generals”
of Soviet computer hardw are, the “E9CM” fam ily. They are large in stallations,
extremely reliable, fast, w ith a vast data-handling capacity, tremendous
universality, and capable of solving the most complex m athem atical pro
blems.
The first-born of this fam ily was the “E9CM-1” created by a team of scientists
and engineers under Academ ician S. Lebedev, Hero of Socialist Labour. O rigi
nally it was the best in Europe, capable of performing 8 000 and, after some im
provem ent, 10 000 operations per second. It also possessed an excellent memory
w ith swift retrieval.
Today this m achine is a great-great-grandfather. I t was replaced by a new ge
neration, the “B9CM-2” and “E9CM-3”, capable of performing 20 000 operations
per second and w ith a vast memory of four m illion words.
The next generation was represented by the “B9CM-4”, awarded a diplom a of
the International E xh ib itio n of Engineering, A dm inistration and Management
M echanization held in Moscow in 1966.
Life goes on, and we became witnesses of the creation of “B9CM-6”. This was a
veritable g ia n t—not in size but in capabilities. Though several tim es sm aller
than its great-great-grandfather, the “B9CM-1” , it performs a m illion operations
per second. It is designed for solving a wide range of complex scientific, technolo
gical and economic problems requiring a vast am ount of calculations.
The m achine incorporates all of the latest achievements of computer techno
logy. Its designers were awarded the 1969 State Prize. The “B9CM-6” is capable
of carrying out several operations and solving several problems sim ultaneously.
Input inform ation can be fed to it by means of punched cards or tape, m agnetic
drums or tape, and even by telegraph. A very high-speed memory device additio
nally enhances the m achine’s already high productivity. I t can be run by several
operators sim ultaneously working from several control panels, which need not
be in the same room.
62 CYBERNETICS A TO Z
One of the flag-ships of Soviet dern physics, which entail the pro
com puter electronics has already as cessing of numerous experim ental da
sumed its duties a t the head of the ta in the shortest possible tim e.
electronic m achines of the sophisti As you see, m any excellent compu
cated com puter complex a t the Jo in t ting machines have been created in
Nuclear Research In stitu te a t D ubna, this country. Their rate of production
north of Moscow. The “E9CM-6” is is such th a t specialists confidently
employed in solving the most diffi declare: “In the near future w e’ll
cult m athem atical problem s of mo move to one of the first places.’"
CAPACITY OF MEMORY 63
If we define m em ory as the ab ility the num ber of elem entary signs will
of the brain to store inform ation ac be 100 0 0 0 x 1 0 000 = 1 0 billion. This
cum ulated in the past and produce is equivalent to 50 000 pages of the
it on request, in our century of elect Grand Encyclopaedia.
ric ity we unw ittin g ly tend to assume B ut this is not the final word.
th a t memory is a com bination of If we go on, it will become clear to
pulses travelling along a circuit in us th a t w ith an “id eal” code the
the same way as does electric current. tex t w ritten w ith the aid of the two
B ut, alas, as yet no “memory cur signs of our chromosomes would amo
ren ts” have been found in the brain. u n t to hundreds of thousands or
The question looms large in the even m illions of encyclopaedia pages.
m inds of scientists: how does the And w hat if nature managed to pro
brain hold the fantastic am ount of duce some special signs by m aking
inform ation accum ulated by m an du sm all shifts in the molecules? Then
ring his lifetim e? Scientists are labou the inform ation capacity w ill defy
ring unceasingly to solve the problem all im agination. T h a t’s w hat can be
of the origin of memory. done w ith the aid of atom ic and mo
The opinion holds th a t our brain lecular ABC. E xperim ents have shown
is made up of 12-14 b illion neurons. th a t not everything and not always
It can be presumed th a t each neuron is retained in m a n ’s memory for a
is capable of storing more th an one long tim e. Frequently inform ation is
u n it of inform ation. The presum ption kept stored in his head only “u ntil
th a t this m inute particle of the brain tom orrow”. This leads us to suppose
is able to assume ten or hundred th a t m a n ’s memory is of several
states to record inform ation seems types.
less justified. Even if th is was the Inform ation reaching the brain con
case, it would be impossible to im a nects its neurons into electric cir
gine how the brain is able to hold cuits and circulates in these circuits.
the gigantic reserve of intelligence. This is the short-tim e memory.
We cannot b u t presum e th a t memo The currents circulating in the neu
ries are recorded on the m olecular ron circuits act upon the molecules
level, th a t molecules of memory are of the nerve cells, and this leads to
active in the brain. Those are enor proteins being synthesized in them .
mous, very com plicated molecules, W hen the signal is repeated, the pro
which look like rope ladders w ith tein reacts to it and adm its infor
cross-beams of two types. These mo m ation to the long-tim e m em ory—
lecules, placed in a definite order, the brain. If the signal is not repea
like the dots and dashes in the Morse ted, the circuits w ill fall apart, the
code, constitute a peculiar atom ic short-tim e m em ory w ill be switched
and m olecular ABC. off. It is presumed th a t there is also
How much can be recorded w ith a working memory. Ju st w atch a
the aid of it? L e t’s make some cal ty p ist. W hen typing a tex t, she keeps
culations. Each molecule has 10 000 in her memory only short pieces of
cross-beams of two types. The chro it for a very short tim e u n til she
mosomes of hum an cells contain finishes typing them .
some 100 000 genes. This means th a t People were quick to see th a t all
CAPACITY OF MEMORY 65
records in the brain are short-lived, People learned to memorize, not only
and th a t extraordinary m emory is, w hat they have seen, but also w hat
un h ap p ily , the gift of very few per th ey have h e a rd —due to the appea
sons. P ractically , to memorize m an rance of sound recording of the m e
had to seek the aid of artificial means chanical, photographic and m agnetic
of memory, capable of supplem enting types.
the natu ral memory of the hum an A m ethod of recording various com
brain. m ands has been devised as far back
The oldest practical m ethod is the as the M iddle Ages. P lates were
use of inscriptions and signs. It “sensed” by pins fixed onto drum s—
dates back to rock-wall pictures. Then b ells rang, m usical boxes played.
came the turn of characters, and Inform ation was also memorized by
after them of the letters of ABC’s. holes of perforated drums of looms
The invention of book-printing has in the m iddle of the 18th century.
increased hum an m em ory to the d i G reat prolification of artificial me
mensions capable of engulfing all m ory aids is characteristic of our
sciences, all in tellectual heritage of tim e. E lectrom echanical and elect
m ankind. ron-tube recording devices em ploying
L ater photography came to the aid two stable states (on-off) were suc
of graphics, and it was followed by ceeded by novel devices using m ag
cinem atography. This enabled events nets (m agnetized-dem agnetized), se
to be memorized in action, in tim e. m iconductors, etc.
L e t’s take a look at the computer m em ory. Its electronic “m em ory” is used
as a depot to store numbers and instructions. From this depot they are sent for
processing to the “m athem atical m ill” . The results obtained are returned to the
“m em ory” . An absolute order m ust be m aintained here to fac ilita te the extrac
tion of the necessary numbers or of a definite result.
The computer “m em ory” , its recording device, can be compared to the head-
gear wardrobe. I t ’s full of hats, but each occupies its own section. E ach section
has its number. Produce your check and take your hat (yours, and nobody else’s).
The sections of the w ardrobe, its cells, all have th eir addresses—th eir num bers.
Take, for instance, the instru ctio n + 475. I t m eans th a t th e num ber occupying
cell 475 should be added.
The com puter, like m an, has “m em ories” of different kinds.
Y ou’ll sm ile recalling how you say aloud, when m u ltiplying: “Seven tim es
five is th irty five. W e’ll w rite five and keep th re e ” . The com puter has to do the
same. I t, too, w rites five and keeps three in its m in d —memorizes.
Sometimes an address or a telephone number is needed only once. In this case
it is w ritten down on any scrap of paper th a t comes handy. H aving telephoned,
you throw it away. If you expect to use the address or the telephone once or twice
more, y o u ’ll, probably, copy it into your note-book or address book. This con
tains data the need for which will arise in due tim e. This is a fundam ental, long
tim e “m em ory” .
5-616
66 CYBERNETICS A TO Z
*♦£+***+
*+*+*+*+1
*>*+► :<■***!
* + * * * •* * + !
The com puter, too, has all these kinds of “m em ory” . Y o u ’ll find here the
“scrap of paper” , together w ith the “address book” of long-tim e “m em ory” and
the note-book.
The “scrap of paper” is the w orking “m em ory” d irectly connected w ith the
arith m etical blocks of the com puter. The “note-book” is the long-tim e recording
devices. They serve to m em orize all inform ation introduced into the computer
th a t is needed in the course of solving the problem . The “address book” of the
com puter contains constant reference d ata, tables, coefficients—all th a t, like the
m u ltip licatio n tab le , m ay be needed in calculations. Hence, its nam e—the per
m anent recording block.
The capacity and operation speed of the computer “m em ory” are determined
by its construction. It is interesting to note th a t these two characteristics are
closely interrelated . The sim plest type of “m em ory” is the m agnetic tape. I t ’s
CAPACITY OF MEMORY 67
easy to increase its capacity just by increasing its length. But sim ultaneously
the tim e needed to find a number on this tape w ill be increased, too. Tens
or even hundreds of m etres of tape m ust be turned over to find the required
data.
Therefore, the device was adopted of increasing the capacity of the “m em ory”
while a t the same tim e decreasing the tim e needed to produce d a ta , the so-called
circulation tim e.
M agnetic tape drums enable recording blocks of practically lim itless capacity
to be b u ilt. W ith the recording density of one number per m illim etre, 500 m etres
of tape can record 500 000 numbers; one hundred tapes of this length w ill hold
5-107 num bers. Compare th is number w ith the number of signs in L. T o lsto y ’s
novel W ar and Peace, which is only several m illions. But in this case it w ill take
C o n tro l w ir e
N um bers
S u p e rco n d u ctin g
f ilm
O rd e rs R eading w ire
C IR C U L A R C U R R E N T
TW ISTORS M AG NETIC F IL M S
ELEM ENT
Reading w ire
MINIATURE
MEMORY P rin te d
c ir c u it
ELEMENTS
T R A N S F L UX OR S SE M IC O N D U C TO R E L E M E N T S
O)
M em ory
e le m e n ts
a t jo in t s
s &
4~e g
. a:
P =6
o CO
0O C
0C
A d d re s s
T U N N E L D IO D E S M IN IA T U R E F E R R IT E
CRY OSAR
P L AT E S
5*
68 CYBERNETICS A TO Z
several m inutes to find the necessary cells w ith numbers. Most ingenious cont
raptions enabled this tim e to be reduced to half a second. However, th is tim e is
etern ity for m odern high-speed computers.
So m agnetic drum s were invented. The m agnetic drum is, in essence, a very
wide m agnetic tape closed to form a loop. It contains m any—up to eig h ty —
recording paths. This requires a corresponding number of recording and reading
heads.
The drum holds up to 30 thousand num bers, which are recorded, as on th e tape,
by m agnetizing sm all areas of the surface.
The drum revolves a t a high speed th a t sometimes reaches 12 thousand revo
lutions per m inute. The necessary num ber, or group of num bers, is read or recor
ded in the tim e of one revolution.
sum ption does not exceed ten w atts; difficulty in reproducing lost scores
a hydro-electric sta tio n would be of long m usical com positions. The
needed to power a corresponding com b rillia n t p ian ist R achm aninov had to
puter “m em ory”. However, much b e t listen to a piano concert only once
ter results are now being obtained to reproduce it correctly. T he pro
w ith integrated circuits (IC). m inent chess-player A. A lekhine held
And, of course, there rem ains the in his m em ory all the chess games
param ount advantage th a t every one played and could sim ultaneously play
of us feels every m inute. D uring our w ith 30-40 partn ers w ithout looking
lifetim e we register m illions of events a t the chess-boards.
and im pressions which accum ulate to Our m em ory is defined as asso
make our in tellectu al w ealth bound ciative.
less. This w ealth is a powerful source Rem iniscences follow each other in
of supply for m a n ’s em otional life. a logical process th a t guards the brain
Man retain s in ta c t in his m emory against excessive work by providing
the colours of the settin g sun of it w ith necessary inform ation a t the
some evening in his youth, as well as rig h t m om ent, im parting to it a won
h ardly perceptible events from his derful inventiveness.
private life. The “m em ory” of the electronic
We are tem pted a t th is place to b rain looks quite hum ble alongside
cite some exam ples from a book on th is wonder of nature. Inferior capa
bionics of how m an memorizes im c ity (as com pared to the capacity of
pressions once received. The French our brain) is not the m ain draw back
graphic G ustave Dore (the author of of the com puter “m em ory”. Its tr a
well-known illu stratio n s of Francois gedy is its im potence in searching
R a b e lais’ book Gargantua and Pan- for inform ation on the “shelves”
tagruel) once received an order from of the electronic or m agnetic “d e
his publisher for a copy of a photo p o ts”.
graph of an A lpine landscape. Dore To find a definite “rem iniscence”
went home and forgot to take the the com puter has to look through
photograph w ith him . On the follo tens of thousands of inform ation b its,
wing day he brought an exact copy ju st as a negligent store-keeper would
of the photograph. I t ’s an estab look through his store to find a lost
lished fact th a t the best p o rtra it of item .
President Lincoln was m ade by his True, scientists foresee the possibi
provincial adm irer who saw him only lity of increasing the com pactness of
once. com puter “m em ory” to th a t of the
The contem poraries of Ju liu s Cae brain. If th is m aterialized, th e com
sar and A lexander the G reat cre puter would be able to store in a
d it them w ith knowing all th eir sol volum e of 1 cu m all in form ation
diers by name and by sig h t though accum ulated by m ankind in all its
they num bered as great as 30 000. history! This is equivalent to 100 000
I t ’s rum oured th a t M ozart managed b illion b its of inform ation. T his gives
to w rite down accurately a long in tri us reason to hope for a practically
cate sym phony which he heard only unlim ited w orking m em ory in the
once. The composer Glazunov had no com puters of the future.
70 CYBERNETICS A TO Z
F irst we come to the key-board ted w ith the aid of sim ple equations
counting m achines departm ent. I t re w ith o u t tak in g into account the afo
m inds one of a big typing office. resaid factors. Otherwise a situ atio n
There are autom atic arithm om eters on m ay arise of the sort in w hich the per
special tables. They look like ty p e sonages of Jules Verne found them sel
w riters. The fingers of the operator ves when they failed to reach the Moon.
ru n fluently over m ulti-coloured keys. Of course, com putations of this kind
He types num bers—item s, m u ltip li- were not m ade in our presence. We
cants, dividends and devisors. By saw only some au x iliary w ork—the
pressing definite keys the operator checks of operation stages, of com pu
makes the m achine au to m atically per ta tio n accuracy. The m athem aticians
form some arith m etical operation. can be said to follow the long path
A short buzz of the electric m otor, a leading from the equations and for
lig h t c la tte r of the gearwheels, and m ulae to tables and num bers. The
the resu lt is there, on the counter flow-line production of all this gi
of the arithm om eter. The operator gantic mass w ill be done by power
enters the result on a b lank. And his ful electronic com puters.
fingers are on the keys again. They occupy the m ain shop of the
We are told th a t it is no easy job com puter centre. One wonders a l i t
to prepare for com puting operations. tle seeing such vast work being done
There are some very com plicated ca so silen tly and unostentatiously. Only
ses. For exam ple, there are num erous the lig h t hum of the com puter and
factors th a t have to be taken into the ra ttle of the typing device filling
account in constructing a space rock in long paper tapes w ith colum ns of
et. I t m ust be controlled from the num bers te ll of the u n tirin g work of
E arth . The rocket has liq uid-propel the “m athem atical m ill”.
la n t rocket engines which enable it The com puter centre m ay be con
to rise to a great height. Its body nected by special com m unications
houses extrem ely sophisticated devi channels w ith the organizations it
ces, control system equipm ent, fuel serves: w ith plants and factories, in
and oxidizer tanks. stitu te s and offices. This is the more
Before s ta rtin g th e design work, economical and advantageous method
m any calculations have to be m ade. th a t provides for c o n tin u ity of work
I t m ust be taken in to account th a t and for standardization of m aintenance.
such a rocket is launched v e rtica lly One com puter centre can do the
and a t the s ta rt, during the first work of an arm y of calculators tens
period of its flight, it passes through of thousands strong. C om putation
the densest layers of the atm os w ith the aid of com puters saves m il
phere. The m ain p a rt of its trajecto ry , lions of roubles. Even a sm all com
however, passes in vacuum . The puter, say, “M insk-22”, brings 50
E a rth ’s air jack et intensively decelera thousand roubles profit per year. F or
tes th e rocket. As a resu lt of air m erly, a com putation requiring one
frictio n the rocket’s body is intensively y e a r’s w ork of calculators was unac
heated. T his, too, has to be taken ceptable to m any design bureaus. And
into account. now? Now a problem requiring a
T rajectories should not be com pu hundred years of ca lc u la to rs’ work is
COM PUTER CENTRE 73
Here’s a first-class computer centre. The area of its premises amounts to 4000 ma.
Some 500 people work here, of them 150 are highly qualified m athem aticians.
The centre has several big electronic digital computers. One of them works
at a rate of up to a m illion operations per second. There are, as well, several sets
of autom atic calculating machines of simpler types and up to a hundred table
key-board calculating machines.
The CC, like a big plant, works in three shifts. Computers are expensive. It is
unprofitable to let such valuable equipment stand idle. Once a week “the number
factory” stands still: there is no rattle of the typing devices, no monotonous
hum of the computers. Expert engineers test, tune and carry out maintenance
repairs—the computers must always be in a working condition. Should anything
go wrong, and there w ill be a flood of mistakes on the tapes where the compu
tation results are printed. The computers demand precision.
There are hundreds of computer centres in this country. Some of them are the
so-called principal centres, for instance, the Computer Centre of the Academy of
Sciences of the USSR, the Principal Computer Centre of the State Planning Com
m ittee of the USSR. There are computer centres working a t the Academy of Sci
ences of the Ukraine, of Ryelorussia, of Georgia, of other Union Republics of the
USSR, in many of the larger cities. Nowadays powerful factories for the proces
sing of numbers ceased to be a rarity.
Computer centres may serve various purposes: the general-purpose CC, CC for
computations of plans and for economic studies, CC for branches of industry.
In the drawing you saw a general-purpose computer centre. Specialized CC
are somewhat different. They may have a sm aller number of low-power m achi
nes. Research work of a general nature, such as the principal problems of com
puter m athematics, is not done at a branch or a specialized computer centre,
general methods of solving scientific problems are not developed here either.
These centres carry out concrete computations in the field of planning of natio
nal economy, in economics. There is a lot of accounting and statistical work to
be done here and of calculations for administrative purposes.
Recently a new trend in the use puter systems for collective use. These
of electronic computers and of com will be perm anent systems to which
puter centres has developed. A ten several subscribers would be able to
dency may be observed to “collecti apply at any time for any reference,
vize” them, to unite them into com for any computation, because they
Computer employed in the divided time mode to serve multiple subscribers. This makes a dia- ►
logue between the subscriber and the computer possible. This faculty of information exchange
exists for several subscribers as well. Tens of distant subscribers, too, may be connected to the
computer.
Several types of such connections are shown: a scientist in his flat, an engineer in a design of
fice, a research worker in his laboratory, a scientist at an institute, a secretary preparing matter
for print, a mathematician, a newspaper editor, an historian, a physicist.
The automatic distribution of computer time is much more efficient than the usual mode of
employing computers; it saves time and money.
76 CYBERNETICS A TO Z
E longating as it is being rolled the steel moves between the rolls of the last
stand at a speed of 900 m etres per m inute or nearly 60 kilom etres per hour. And
despite this speed, control gauges installed along the rolling path m anage to
report to the control device the thickness and the stress of steel at various stages
of the rolling process. Zonal regulators w ill measure the tem perature in the fur
nace and correct it, special gauges w ill m ark the changes in m etal.
The cybernetic commander w ill re-arrange all incoming d ata, compare it w ith
the standard recorded in its ‘“m em ory” and in sta n tly decide w hat corrections
are needed in the speed and pressure of individual rolls to ensure the required
thickness of the rolled sheet.
A deviation of 0.13 mm from the specified thickness of the sheet is considered
norm al. The controlling device managed to reduce it to 0.076 m m —nearly twofold!
It was not overnight th a t experts in autom ation succeeded in building such
perfect com puters for the control of technological processes. During the first
experim ents in the use of computers operators introduced into them d ata they
read off the instrum ents. On the basis of this inform ation the electronic com puter
evolved directions for the tuning of control devices. N atu rally , the operator
retarded the control process. Moreover, he was not a very reliable p art of such
a system . Then it was decided to connect the instrum ents directly to the com
puter. Now it could itself read off data needed for com putations. But still the
operator had to tune the control devices in accordance w ith computer directions.
Only the th ird stage of control im provem ent resulted in full autom ation. It
involved the direct connection of m easuring and control instrum ents and control
mechanisms w ith the com puter. And w hat did this lead to?
L e t’s take a specialized paper on the subject: “The com puter connected w ith
hundreds of gauges through its own com m unication system quickly collects in
form ation about deviations from the standard technological process and ana
lyses these d ata. It can find out w ith fan tastic speed w hat takes place, say, in a
rolling m ill or in a chemical plant, where accurate determ ination of the course
of the process involves thousands of calculations. H aving analysed the situation
the com puter decides quickly how valves, injectors, rollers and sundry mechanisms
should be regulated. Then it takes appropriate action. Through the comm uni
cations line it sends instructions to organs controlling individual mechanisms
of the aggregate, ordering them to carry out the necessary tuning.
“L iterally w ithout the slightest interference by m an the computer can repeat
the operation several hundred tim es a day, week after week, using free m inutes
to type concise reports of its w ork.”
F irst results of com puter applica To autom ate power production the
tion were the most im pressive in readings of thousands of control in
four m ajor industries: power, m etal strum ents m ust be taken into acco
lurgical, oil-processing and chemical. unt.
W hy? They all belong to the so-cal The control system m ust carry out
led mass production type w ith num e some 1000 control operations in se
rous control param eters. quence only to put a therm al power
CONTROL COMPUTER 81
statio n tu rbine into full operation: parativ ely sim ple highly reliable com
it m ust w atch the cooling, the water puters are usually chosen. T heir task
level in the boilers, the flame, etc. is to work w ith the object of control
B ut this turns out to be only the on the real tim e scale. Each of the
beginning of a difficult work. H aving lower-level com puters can do the job
put the aggregate into operation, the of several tens or even hundreds of
com puter m ust control the tem pera autom atic regulators. A higher-level
ture and the pressure of the steam , com puter is a m uch more in tric a te
and the speed of the tu rb in e accor m achine w ith a m uch more com pli
ding to the readings of 1560 in stru cated structure. Such a com puter
ments. works w ith up to several tens of lower-
And th is is its everyday work. I t level com puters, and is engaged, as
has also to be on a look-out for emer a rule, in optim izing calculations,
gency situ a tio n s—everything can h ap planning and organizing the work of
pen in a pow er-station. Should gas the system as a w hole.”
(the usual fuel) be cut off the com pu Now you see th a t the step-by-step,
ter w ill carry out fifty operations to hierarchical stru ctu re of control sys
change the burner over from gas to tem s autom atically leads to a na
oil (the reserve fuel). Even when a tu ra l change over to system s control
turbine bearing fails, the com puter ling entire factories, firms and even
com m ander is not terrified. I t is industries.
capable of sw itching off the colossal This to a greater degree obviates
power p lan t in the spell of several the need for continuous inspection of
m inutes. the autom ata by m an. All control
In addition, the com puter sim u lta and regulating functions are concen
neously calculates technical and eco tra ted in one block.
nomic data, sums up report data, Man ru lin g over the production
calculates the efficiency coefficient of process plays the p a rt of a strateg ist,
the plant. because now he is free from the ne
There is one more property left to cessity of solving m inor problem s of
com plete the description of the con operation, from the routine everyday
trol com puter—it makes no m istakes. factory work.
Now is the tim e to revert to the Meet the control system of the
hierarchy of control, to the higher Lvov television receiver p lant. Its
and lower levels of it. One of the in i name is “Lvov”. I t ’s a system ty p i
tiato rs of control system s develop cal of mass production plants. It
m ent in the USSR, A cadem ician can be connected to the country-w ide
V. G lushkov, defines th eir tasks in system of economics control. In one
these words: year “Lvov” helped to achieve a
“For the lower control level com 20-per cent increase in production
.4 The control computer in control of a rolling m ill. Steel slabs pass through the furnace (1), pass
at great speed through a row of mills: cutting-off (2), rolling (3), grinding mill (3), and
finally leave the rollers of the coder in the shape of thin sheets. The dashed line shows the
route of input data to the control computer; the solid line shows the route of instructions
to control centres and thence to individual instruments.
6 -6 1 6
82 CYBERNETICS A TO Z
APPLIED
A s tro n o m y ,
: y b e r n e t ics T ra n s la tio n ,
a s tro n a u tic s I in g u is tic s
E n g in e e rin g
T ra n s p o rt In fo rm a tio n
p ro b le m s, P la n n in g
c o n tro l s e rv ic e
c o n s tru c tio n
C o m m u n ic a M odel 1ing
th in k in g p r o -
tio n c h a n n e ls
T h e o ry o f cesses
a u to m a tic
c o n tro l and
re g u la tio n
In fo rm a tio n S ta tis tic a l Machines
th e o ry A n im a l models
m a c h in e s that learn
\CYBERNETICS,
86 CYBERNETICS A TO Z
of inform ation about the controlled processes of tran sm ittin g it obey cer
object, process, environm ental condi ta in common q u a n tita tiv e laws.
tions, work program m e, etc. A characteristic feature of the whole
O bviously, the nature of the in diversity of such system s is the
form ation carriers m ay v ary w idely feedback. Through it they receive in
from system to system : sound, lig h t, form ation on the effects or results of
m echanical, electrical or chem ical sig th eir control operations.
nals, punched or m agnetic tape, m as F in ally , the control system s—l i
ter pattern s, etc., m ay be employed. ving and m an-m ade—incorporate ele
The im p o rtan t th ing is, as m entio m ents perform ing sim ilar jobs: the
ned before, th a t, irrespective of the receiving, classification, m em oriza
m aterial carrier of inform ation, the tio n , etc., of inform ation.
V a c u o le s
M em brane
C h ro m a tin
of physicists, chem ists, engineers. The d a ta obtained in experim ents can be rep
resented on them in consecutive series of dots which reveal to the specialist the
laws governing the phenomenon under consideration.
It is extrem ely difficult to grasp, describe and represent graphically the comp
lexity of living system s. This is the complex, tangled world which cybernetics
is invading through two m ain gates a t the term ini of its two m ain directions,
theoretical and applied.
This subdivision as it were delineates the spheres of activ ity of cybernetics
in biology: the purpose of theoretical research is to elaborate problems connected
w ith the investigation and description of various control functions, control “mo
tio n s” in the body. This is “in te rn a l” cybernetics. It seeks to understand how
various control and controlled system s function w ithin the organism, w hat com
m unication channels tran sm it the signals about various actions.
b u ilt a “mouse in the m aze” device. its cupped paws, and a m etal ta il
P lastic lam ina on the cover of a black which drags over the floor behind it.
valise form a fanciful p a tte rn like Here is how th is little “a n im a l”
leafless branches and tw igs. A sim i operates. E lectric lights are burning
lar p a tte rn can be observed on the in a big room. W hite beads are scat
glowing w all panel of an autom atic tered over the floor. In one corner
blocking system a t which a railw ay is a m etal sheet l it by a daylight
dispatcher works. lam p, the “sq u irre l’s n e s t” . I t is
The p lastic strip s represent the brought in and placed on the floor.
tracks, the lights between them are The “sq u irre l” begins to move about
the statio n s. There are 25 of them . the room u n til it spots—w ith the
Each one has its own sw itch. W hen help of its photoelectric cells—one
it is flipped over one of the sta tio n of the w hite beads. I t tu rn s to the
lights flicks on. bead, scoops it up, “ta s te s ” it w ith
Now the th irte e n th flicks on. We its “tongue” and then tu rn s to look
press a b u tto n and the strip s repre for its “n e st” . The electric filter
senting tracks flash on one after enables it to steer tow ards the day
another. This is the lig h t signal, lig h t lam p.
which passes along the sidings on its W hen the “sq u irrel” m ounts the
way to the d estination. m etal sheet its ta il closes a circu it,
T ry and press the b u tto n a second i t opens its paws and drops the bead
tim e. All the lig h ts on the panel go into its “n e st” , after which it is
out, then the beam of lig h t, bypas free to continue its search for food.
sing the now unnecessary sidings and To be sure, all these “to rto ise s” ,
spare tracks, takes the shortest route “m ice” , “sq u irrels” and other “an i
to the th irte e n th statio n . m a ls” are very approxim ate m odels
A red lig h t glows a t the left-hand of conditioned reflex form ation. B ut
corner of the panel. This is the me they have helped men approach a
mory in d icator. I t m easures the tim e new stage in studying nature: inves
i t takes the u n it to m em orize the tig a tin g organism s by the m ethod of
route. models.
Here is another representative of In biology not only organism s,
th is unusual “zoo” : an electronic b u t processes, too, are sim ulated.
“s q u irre l”. I t rides a platform d ri By building analogues of biological
ven by a m otor and is provided w ith processes we can verify in practice
two sensitive photoelectric cells, a the correctness of our theoretical con
filter th a t can distinguish between structions and postulate hypotheses
direct and a lte rn a tin g current, and for experim ental verification.
other gadgets. Im agine a colony of living orga
The “sq u irre l” has two paws which nism s living in favourable conditions.
it can cup together. I t also has a T heir num bers w ill change depending
little tongue w ith which it can “lic k ” on th eir b irth and death rate. B ut
◄ Electronic “animals” are used in the search for methods of studying organisms with the help
of models.
94 CYBERNETICS A TO Z
how, according to w hat law? The cal processes helps men gain an u n
b irth ra te is affected by the available derstanding of the complex system
food supply, suppression by other of links between species and forecast
species, life expectancy and m any the consequences of hum an in terfe
other factors. Scientists have b u ilt rence in the affairs of n ature.
so-called ab stract models and used Nowadays biology, one of the ol
them to establish the precise laws dest sciences which originated as a
of developm ent of organism s in c ir descriptive and experim ental science,
cum stances of u nlim ited food supply is boldly ap p ro p riatin g all of the
and living space, in the absence of late st from the arsenal of scientific
harm ful species, and. in conditions means: cybernetics, m athem atics,
of hunger, lim ited living space and com puters, m ethods of a b stra ct ana
harassm ent by predators. lysis, speculative reasoning and, a t
Such a model was used, for exam the same tim e, precise calculation.
ple, to help in growing penicillin I t can be said th a t we now w it
fungi. They were fed abundantly, ness the emergence of a new biology,
provided w ith ample space for sw iftly advancing thanks to num erous
grow th and sheltered from predatory highly prom ising discoveries in mo
species. And the future harvest was lecular biology, biochem istry, bio
predicted accurately by m eans of a physics, cytology and genetics, thanks
special form ula. to the appearance of new and unex
In terestin g facts are provided by a pected branches such as b io statistics,
model of the sta b iliz atio n of the biom athem atics and system s theory.
num ber of two species, one of which Besides theoretical investigations,
is a predator w ith respect to the other. cybernetics in biology engages in im
R unning ahead, we can say th a t the m ediate p rac tic a l work. T his is the
“prey-predator” m odel offers m athe o th er—ap p lied —branch, which abuts
m atical proof of the usefulness of on cybernetics in m edicine.
predators. N a tu ra lly , it is hard to cover all
Suppose we decide to destroy all the spheres of applied cybernetics in
the wolves so as to increase the num biology and we sh all have to single
ber of th eir prey—herbivorous an i out the prin cip al targ ets of th e “cy
m als. The model revealed th a t an n i bernetic a tta c k ” , the m ain directions
h ila tio n of the predators m ay result of the “offensive” .
in a brief burgeoning of the popula L et us retu rn once again to m a
tio n of th e ir prey, followed by a th em atical sim ulation. W ith the help
sharp drop and alm ost to ta l e x tin of special equations one can describe
ction. And in fact biologists have the process of photosynthesis—the ab
observed a m arked deterioration of sorption of solar energy—in plants.
herbivorous stock wherever wolves S cientists are w orking on m athem a
have been elim inated: ailing anim als tic a l m odels of such processes as the
abounded, the offspring grew sm aller. evolution of species, blood circula
The wolves, it was found, performed tio n , and others.
a very useful function by destroying There is another approach to m a
unviable anim als. th em atical sim ulation. W ith the help
As you see, sim u latio n of biologi of a su itab le analogue of the biolo-
CYBERNETICS IN BIOLOGY 95
then consult th is central m edical strid e longer, its accom plishm ents
“m em ory” . greater. I t is well to recall here the
Cybernetics has entered the surgi words of our celebrated physiologist,
cal operating room. I t particip ates Ivan Pavlov: “The tim e w ill come—
d irectly in operations and controls even though it m ay be a long way
v ita l functions of the organism : looks off—when m athem atical analysis in
after the work of the h eart, regulates com bination w ith n a tu ra l science w ill
a rte ria l blood pressure, controls ana illum ine all these e q u ilib ria [life
esthetic sleep. from the sim plest to the m ost comp
Cybernetics is confidently invading lex m echanism s—V .P . ] w ith the m a
biology and m edicine. In tim e its jestic form ulae of eq u a tio n s.”
pace w ill become even firmer, its This time has now come.
97
of fam iliarizing him self w ith one-tenth of the publications appearing in perio
dicals on his specific field. And “sw allow ing” is not enough: one m ust also “digest”
w hat one has read.
The situ atio n is hardly any better in m athem atics, physics, biology and other
m ajor spheres of scientific progress. Experts have calculated th a t in the near
future computers w ill be handling such a q u a n tity of inform ation th a t it will
be equivalent to 7000 pages of reading m atter per head for every m an, woman
and child in Europe.
As for fiction, some are thinking of adopting the A m erican way out: novels
by Tolstoy, Stendahl and Dickens are compressed into 20- or 30-page digests
th a t can be carried around in a coat pocket and read in two instalm ents when
there is nothing better to do. How convenient, it would seem, yet ....
Here is w hat Stephan Zweig had to say on this score: “Try and read a Dostoy
evsky novel in an abridged French edition. E verything seems to be there: the
sequence of events unfolds faster, the figures seem more m obile, integral, impas
sioned. Yet they are somehow em asculated: th eir souls lack the same finish, the
specific sparkle rad iatin g all the colours of the rainbow , the atm osphere of glit
tering electricity, the oppressiveness of in tensity which only a discharge makes
so terrible and so salu tary . Som ething is irretrievably destroyed, the m agic cir-
cle has been broken. In these experim ents in abridgem ent one realizes the m ean
ing of D ostoyevsky’s breadth and the necessity for his apparent long-w indedness.”
It is better to read a thousand novels in the original or a good tran slatio n th an
to gain a distorted idea of a hundred thousand. B ut in science and technology
the condensation of inform ation into digests, abstracts or extracts is a m ust.
In our country in 1952 an In stitu te of Scientific Inform ation was set up under
the Academy of Sciences. I t has now expanded into a vast “concentration” m ill
for processing and enriching scientific and technical literature.
like if it is to serve its purpose? and other hybrid sciences, even nar
F irstly , it m ust be brief and to the row specialists cannot afford to miss
point: better to read the original pa parts of papers dealing w ith adjacent
per th an a verbose exposition of it. sciences. F requently the same paper
A t the same tim e it shouldn’t be is of interest to researchers in diffe
overabridged because, by its very na ren t fields, for instance in chem istry,
ture, it m ust contain inform ation m athem atics and biology. This means
th a t does not lend itself to abridging, m aking three extracts from a single
such as form ulae, dates, geographical paper. B ut extracts are convenient
names or names of investigators. Fi as a source of inform ation only so
nally, in our age of m athem atical long as there are not too m any of
linguistics, bionics, space chem istry them . Y et already the annual o u tp u t
A n a ly s is
1APBp n
fE3 J
Q 1
C a ta lo g u e
P u b lis h e d
Id e as and fa c ts
da ta
DATA HANDLING 101
of the Soviet in stitu te alone runs to Ivan Fedorov, spent eight m onths
a thousand th ick volumes. and nine days setting and printing
In cidentally, in 1965 some 1500 his In stru ctive Scriptures. Four hund
journals of extracts were published red years late r the 11th volume of
all over the w orld. Y et even so they the com plete works of Charles Dick
are no longer capable of encompas ens, which required no editing, took
sing the steadily growing avalanche eight m onths and tw enty days to be
of inform ation. published, and volume 30 took more
We are often fascinated by the th an a year. For articles published
w ealth of the m ajor world book reposi in 1916 in the Journal of the Russian
tories. For exam ple, the Lenin State Physico-Chemical Society, the tim e
L ibrary in Moscow has 22 m illion lag between in itia l com position and
volumes of books, journals and news final readership was two and a half
papers. Its to ta l shelf length exceeds m onths; in 1966 the tim e lag in the
300 kilom etres, and each year ano Journal of Physical Chemistry was
th er 15 kilom etres of shelves is added. 19 m onths. And th is at a tim e when
B ut a large proportion of th is w ealth the tempo of scientific progress is so
is no more th an so m uch deadweight: fast th a t m anuscripts can grow old
about half the m aterial in the Lenin in a few months!
L ibrary has never been requested by Such is the problem of inform ation
readers. The reason? The same one: processing in only one departm ent of
the files and catalogues are engulfed data handling, the im portance of
by the sea of printed m atter. which has not yet been fully reali
No wonder m any scientists m ain zed. The difficulties facing the in
tain th a t if an investigation is not form ation ind u stry are m u ltiplying
too costly i t is cheaper to carry it w ith every passing year.
out rath er th an h unt for a needle in W hat are the prospects?
a haystack in an effort to establish It is cybernetics and electronic com
whether someone has or has not car puterization th a t hold promise of a
ried it out before. Indeed, “excava real revolution in the inform ation
tin g ” the paper deposits of Am erican industry.
libraries, for instance, costs 300 m il One of the problem s is sw ift loca
lion dollars a year. It has been esti tion and retrieval of all pertin en t
m ated th a t one-fifth of the invest data on a given subject. D ata re t
m ent in scientific research through rieval systems are being devised to
out the world is spent on gathering ensure the m ost ratio n al organization
and d istrib u tin g inform ation. Y et and com plete autom ation of the pro
m ost scientists spend perhaps one- cess. They employ m icrofilm s, pun
th ird of th eir tim e on painstaking ched cards and electronic com puters.
research in the quest of results al Such a system , for exam ple, was
ready obtained by someone else. developed in Moscow a t the V ish
We adm ire m odern prin tin g m achi nevsky In stitu te of Surgery. In K iev,
nes, forgetting th a t w hilst they have at the clinic of thoracic surgery hea
helped boost p rinting, they have not ded by Professor N. Amosov, an archi
made pre-publication preparation any ve has been organized in w hich case
faster. The first R ussian p rinter, histories are recorded on punched
102 CYBERNETICS A TO Z
cards. The inform ation thus stored books and journals to be compressed
is intended for com puter processing. into the size of a spool of film. Thus,
W orkers of the Cybernetics In stitu te a whole lib rary or archive can be
of the U krainian Academy of Scien stored away in a single cupboard.
ces are studying ways of tran sm ittin g Tom orrow’s m icroduplicating m a
m edical data over distances in a di chines w ill be capable of compres
g ital and diagram m atic form. sing the contents of a 30-volume en
An “electronic encyclopaedia” is cyclopaedia into the volume of a
being assembled for chem ists, and a w riting pad or even, some day, a
data retrieval system on the m athe pinhead!
m atical theory of experim entation is In response to telephone requests
being designed. autom atic m achines w ill locate, re t
Machines w ill make work easier rieve and tran sm it inform ation to be
for bibliographers. A m achine was represented as an image on your home
used to compile the index for the TV set. The first such experim ent
com plete works of V. 1. L enin—55 was carried out by Soviet scientists
volumes. The job took several hours. in 1957. The contents of several
I t would have taken m onths if not printed pages stored in the m achine’s
years for a staff of m anual special memory were tran sm itted through the
ists. c ity autom atic telephone exchange.
A nother im portant problem is th a t At the receiving end a screen presen
of autom ating the process of tra n s ted a clear impression of the p rin
lation from one language into ano ted message.
ther. A t present “m an u al” tran sla W hile progress in electronic m icro
tion from R ussian into English em b m in iatu rizatio n holds promise of m ar
races only one-tenth of R ussian scien kedly increasing the capacity of data
tific litera tu re . O bviously, even une repositories, achievem ents in laser
dited word-for-word texts prepared techniques w ill help expand the handl
by electronic translators would help ing capacity of com m unication chan
solve the tran slatio n problem . nels for tran sm ittin g those d ata. The
However, even if m achines are visible lig h t w avelength range in
able to translate all, or alm ost all, which laser tran sm itters operate, has
the printed m atter appearing in the a capacity m illions of tim es greater
w orld, no hundred-eyed Argus could than the to ta l range of w avelengths
hope to read it all. T h at is why, in which the sound and television
from the practical point of view, the tran sm itters operate today. E xperi
thing is to autom atize not only tra n s m entally, light rays are being used
latio n b u t the m aking of extracts as substitutes for telephone cables.
too. This requires m achines capable A t the Econom ic Achievem ent E x h i
of not ju st choosing the equivalent b ition in Moscow there is on view a
foreign words, but of “u n d erstanding” TV in sta lla tio n which tran sm its both
the m eaning of phrases as well. This image and sound not by ultra-high
is no sim ple task, b u t it is already frequency waves b u t lig h t rays.
being tackled. Com m unications satellites w ill make
Today libraries m ake wide use of it possible to establish a fast
m icrophotography, which enables operating system of d ata exchange
DATA HANDLING 103
between libraries and publishers all nes connected through com m unication
over the globe. channels to telephones, television sets
It is not accidental th a t already and phototelegraph a p p a ra tu s.”
today men realizing the im portance All these technological innovations
of coping w ith the torrents of printed will help the specialist of the new
m atter and the flood of inform ation, era of cybernetics—the data hand
predict th a t soon “the day w ill come le r—to chart routes in the boundless
when paper w ill be replaced by elect ocean of inform ation. His m ain task,
ronic im pulses spanning vast d istan however, will be to curb inform ation
ces and processed easily by m achines. about inform ation, to help scientists
The inform ation service w ill develop find their w ay in floods of news and
into a stupendous system of m achi avalanches of documents.
104
The “dem on” of the famous English scientist Clerk Maxwell made his home in
dozens of specialized and popular-science books. The essence of the “dem on”
is, of course, the same in all books, only it is expressed in different words.
To avoid retelling this example anew le t’s take it in the form it was cited by
the “father of cybernetics” N orbert W iener.
“Suppose we have a container filled w ith gas, whose tem perature is everywhere
the same, and suppose th a t some molecules move faster th a n the others. Suppose
further th a t there is a sm all door in the container through which gas flows into
a tube leading to a heat engine, and th a t the exhaust of this heat engine is by
means of another tube connected through another door w ith the gas chamber.
Each door is attended by a tin y being who watches the approaching molecules
and, depending on their m otion, opens or shuts the door.
“The ‘dem on’ lets through the first door only the molecules w ith a high velo
city and stops those w ith a low velocity. The task of the ‘dem on’ at the second
door is opposite—he opens the door only to molecules leaving the container w ith
low velocity and stops molecules w ith high velocity. This results in the tem pera
ture on one end of the container rising, and on the other fa llin g .”
To fac ilita te the understanding of the example w ith M axw ell’s “dem on” W ie
ner proposes his own analogy:
“Perhaps I w ill be able to clarify th is idea w ith the aid of the exam ple of a
throng of people fighting their way into the subway through two turnstiles, one
of which lets through people th a t move sufficiently fast, and the other those who
move sufficiently slow ly. Random m otion of people in the subway w ill then
assume the form of two stream s—of fast-m oving people from the first door and
of slowly-moving people from the second.
A simplified diagram of Maxwell’s “demon”. The molecules with long arrows move faster.
ENTROPY 107
“If both doors lead to a single passage w ith a treadm ill on the floor, the stream
ox fast-m oving people w ill ro ta te the platform in one direction faster th an the
stream of slowly-m oving people w ill ro ta te it in the opposite direction (the flu
xes of “fa s t” and “slow ” people being assumed equal). Thus we w ill be able to
extract useful energy from random m o tio n .”
In other words the ro tatin g door in the subway and M axw ell’s “demon” both
are in a position to reduce the entropy of their “parishes” . So there is a contra
diction to be overcome: the “demon” reduces the entropy, which, according to the
laws of physics, can only rise. A paradox, isn ’t it? Yes, it is. And it rem ained
so for a long tim e, u n til the H ungarian physicist L. Szilard published a paper
under a very learned heading: “On the Decrease of the E ntropy of a Therm ody
namic System Due to the Interference of an Intelligent B eing” .
Here the scientist again m entions M axw ell’s “dem on” .
W hy does he arouse interest? F irstly, because the “dem on” can only act if
he gets additional energy for his w ork—the opening and closing of doors alterna
tely to fast and slow molecules. Only by spending th is energy can the “dem on”
keep his subjects “locked” , so th a t fast molecules stay on one side and the slow
on the other. In other words, only then can the “dem on” reduce the entropy of
the system . B ut w hat does the “dem on” spend the received energy on?
On organizing activ ity , says Szilard — and proves i t —trading energy for infor
m ation as to where should definite molecules go.
The proof presented by the scientist is very com plicated and specialized. There
is no sense in repeating it here. So l e t ’s take his results for granted, all the
more so, since they have been recognized by the scientists the world over.
A tower built by man, a “purposeful physical system”, displays a high degree of order. As years
go by, under the influence of random non-directional forces the tower will tumble and turn
into a stone-heap—the entropy of the system will increase.
Does it make any difference w hat vented the word “entropy” for the
to call inform ation—just inform ation new concept, about B oltzm ann’s theo
or negentropy? I t does. The negen- ry, linking entropy w ith the pro
tropy principle of inform ation unites bable m icrostates of the body, about
the concepts of entropy and infor the ideas of W iener who identified
m ation on a new basis and points entropy w ith the measure of chaos.
out th a t they cannot be treated se Our road has been h ard —from diffe
parately b u t m ust always be consi rence to u n ity . We learned about
dered together. And th is rule is a l the responsible p a rt played by ent
ways true for various infinitely dis ropy in science, how it helped in
ta n t spheres of app licatio n —from the understanding of num erous laws
theoretical physics to everyday life. of nature, and how recently theory
Thus, the circle has been closed. of inform ation “tw iste d ” entropy and
W e’ve read about Clausius who in made science take a new look at it.
Ill
How great is the num ber of in te r m ay be said not only to take p art
connected positive and negative feed in physiological phenom ena, but to
backs! W hat exceedingly difficult be absolutely necessary for the con
tasks the organism has to perform: tin u a tio n of life. W ithout it life is
distrib u te muscle strain in order to impossible!
uphold the required body posture; Thus, feedback is at work in quite
react to the m inutest changes of pres different fields. This enabled scien
sure in blood vessels; respond to the tists in cybernetics to talk about the
heat and the cold; w atch the contents universal, the general character of
of acid, alkalies and m any other the “feedback” concept, to insist th a t
substances; control the incessant work it operates everywhere where in te r
of the h eart, the kidneys, the liver, connected m achines and system s form
the lungs! some new com bination. These com
And the more com plicated is the binations, these new system s m ust be
biology of the organism , the greater harm onious, m ust retain former in
is the com plexity and the v ariety of dividual features and at the same
its “ autom atic regulators” . Feedback tim e gain new qualities.
tie FORMALIZATION
of the waves, and the foam on the chemical elem ents it consists of and
beach are all ignored. Then the sea the physical equations describing the
is described by the symbols of the in ten sity of the waves.
Scientists themselves stress this aspect of the scientific approach when they
say th a t science writes a th rillin g novel about the hidden secrets of nature not
in a colourful language th a t re-creates live association and bright images, but in
its own language, in which everything t h a t ’s individual, subjective, is sacrifi
ced to the advantage of the abstract, the objective, the general.
Every one of you, probably, paid a tte n tio n to a newly b u ilt house. Storey ri
ses above storey, identical staircases, corridors, doors, identically planned, iden
tically arranged flats, one above the other—everything clear-cut, common, iden
tical. And now the tenants arrive. They arrange their dwellings each in his own
way, and the form erly identical flats become different. A difference in furniture
gives in d iv id u ality to the flats.
Roughly speaking, science deals w ith the “house” of nature itself, in its “unin
h a b ite d ” s ta te —w ith only such general regularities, such objective features,
which unite objects perceived by us as different into classes and groups.
Dry and rigorous schem atic diagram s, graphs, form ulae, tables, equations
and symbols help to “u n v eil” essential features of the real w orld, to describe the
“stru c tu re ” of real life, to m ark interconnections in nature.
The difference between the scientific m ethod of studying life and the m ethod
employed by a rt is very accurately expressed in the following words: while art
makes us cry and laugh, science makes us understand and calculate. The pain
ter, the composer, the poet tell us about colours and sounds. The scientist “shuts
his eyes” to the beauty of colours and to the sym phony of sounds. From the colo
urs and sounds he extracts their m ain features, only th a t which makes a colour
a colour, a sound a sound; the scientist reduces both colour and sound to waves
(electrom agnetic in the first case and elastic in the second) of appropriate length
and studies their regularities.
How does science study nature? W hat methods and w hat tools does it use?
T here’s a real arsenal of means of approach to the phenomena of reality .
Rem ember the phrase you often used to hear from your teacher during biology
lessons. “Tomorrow w e’ll s ta rt a new subject. W e’ll study the constitution of,
say, the black cockroach” (or of the bom byx m oth, or the nervous system of the
frog). The example is unim portant, the im portant thing is th a t here we are
up against the logical m ethod of identification: “the study of the black cock
ro ach ” , and not of black cockroaches, variously coloured, probably, w ith feelers
of different lengths or w ith some other individual characteristics. In other
words, you are going to separate the general, common to all, ignoring in d iv i
dual features.
There are other ways of approach, as well. Idealization is one exam ple. In
this case scientists build ideal models of the objects being studied and deal w ith
their general, essential features and properties. E xactly for this purpose the
“absolutely black b o d y ” , the “absolutely hard body” , the “ideal gas” , the
118 CYBERNETICS A TO Z
A closed line L drawn on the surface of a thoroid (a life buoy) will not necessarily divide this
surface into inner and outer parts.
V A >=OR AND
—
FOLLOWS IDENTICAL
tin g situations. This happens so often the sake of sim p licity and c la rity and
th a t conflicts, collisions of interest, perhaps by force of h a b it the m a
have been accepted as one of the m ain th em atical theory of games adopted
subjects of literatu re. the term inology of ordinary games.
We overcome a conflict when we The partners in the game are called
play chess w ith a friend. players, the resu lt—the gain or pay
A conflict is experienced by a child m ent.
who in spite of his p aren ts’ wishes True, the context of term s is some
stubbornly refuses to go to sleep. w hat different here. In the theory
Conflicting is the situation of the of games several people w ith a spe
vendor who, n a tu ra lly , wants to sell cified interest confronting one or ma
dearer and the buyer who w ants to ny adversaries m ay be term ed a play
buy cheaper. er, the same term being applied to
The confrontation of rival political the adversaries. Thus a player is
parties during an election campaign just one in terest group. A football
is an example of a political conflict m atch from the standpoint of the
in cap italist countries. games theory would he “calcu lated ”
The conflict between the hare and as a game of one player against the
the fox as biological species in the other. In th is respect it is not diffe
struggle for existence. rent from a chess game.
Conflict ... we im agine it to be a The prom inent French m athem ati
com plicated, sometim es personal, of cian Louis Borelle already at the
ten em otional and always a difficult beginning of our century published a
affair. I t is never easy to resolve a great m any-volum e Course of Theory
conflicting situ atio n . Y et, m odern of P robability and I t s A pplications.
m athem atical science considers it fea The la st b u t one volume contained
sible not only to analyse a conflic “supplem ents to games of chance” .
tin g situ atio n , b u t even to “calcu Here the scientist summed up the
la te ” w hat line each of the rivals results of his extensive studies of
should take to a tta in his ends. games of chance which interested him
M athem atics has its own approach from the m athem atical point of view.
to collisions of interests, and the Borelle introduced into the theory of
men to exercise it are specialists in games audacious and original ideas.
the theory of games. His predecessors considered only ca
The first thing to do before the ses where the course of the game was
m athem atical analysis of a conflict determ ined by chance and not by
can be carried out is to clarify, to players. Borelle endeavoured to find
unveil the conflict of interests m ost a m athem atical form ula for the games
rigorously, make them so clear and which would take account of the
doubtless as in a game where even proficiency of the players. In the
a laym an sees who is on whose side. course of tim e m any scientists deve
T hat is ju st w hat m athem aticians loped the theory of games to such
do: they build a simplified model of an extent th a t it became m uch wider
the conflicting situ a tio n and call it than the theory of games of chance,
a game. This model-game is played and its results found extensive ap
in accordance w ith certain rules. For plication.
GAMES THEORY 125
As is usually the case, the game is the m ain concepts of the games theo-
very difficult to explain, it is much ry w ith the aid of one of numerous
easier to show how it should be played. exam ples cited by specialists.
And it w ill be easier for us to grasp
Imagine two players A and B. Each of them independently of the other writes
on a piece of paper one of the numbers 0, 1 and 2. N ext they show the numbers
to each other and add them up. If the sum is even, B pays A the sum of money
equal to the sum of the num bers. If the sum is odd, the paym ent is made by A
to B.
The w riting of the numbers and showing them to each other is term ed a move.
The prem editated choice of a system of moves is term ed strategy. A player makes
the decision: “In such conditions I shall act in this w ay .” And the gain of one
or the other riv a l is, as you already know, term ed paym ent.
L e t’s try to follow the course of the game.
The results of all possible moves can be envisaged beforehand. A writes 0 (zero),
B writes 0, too. The sum is zero, and nobody receives anything. A w rites 0, B
writes 1. The sum is an odd num ber. A loses and B wins one rouble. A writes 0
again, B w rites 2. The sum is even. Now B loses and A wins two roubles.
The same m ay be done w ith numbers: (1, 1), (1, 2), (2, 1) and (2, 2). It is easy
to see th a t the number of possible com binations from which A and B have to
choose independently is 3 x 3 = 9 .
L e t’s denote by A i the strategy of A when he writes 0, by A 2 when he writes
1, and by A 3 when he w rites 2. In the same way we shall denote B ’s strategy
by B v B 2, B 3. All possible gains and losses (gains w ith the m inus sign)
can be summed up in a table.
The table containing gains and losses of th e rivals resulting from the applica
tio n of all possible strategies is termed paym ent m atrix , and the game itself a
m atrix game.
W hat does an analysis of the paym ent m atrix yield? To begin w ith, player
A m ust expect player B to be sufficiently clever to evolve a strategy which would
m inim ize the gain of A irrespective of the la tte r ’s strategy. N ext, A m ust choose
such a strategy th a t will guarantee him the m axim um from the m inim um
gains. Such a strategy for A is to continue w riting 0. In this case his loss will
never exceed 1 rouble.
Likewise, player B will also choose a strategy which promises him m inim um
losses. To arrive at it we should look through the columns of the paym ent m at
rix and find the m inim um gain of A .
The gains of A and B are w ritten underneath the table.
The goal of the theory of games is to evolve recom m endations for each of the
rivals as to th eir tactics in the game. W ith the aid of these m athem atical recom
m endations the rivals can choose the best (optim um ) strategy th a t would guaran
tee them best results in the game. True, one strategy is practically never used
throughout. The partners frequently change strategies, and they become mixed.
B ut notw ithstanding the course of the game or the changes in A ’s and B ’s stra
tegies the rivals always strive to act against one another; th eir interests are al-
126 CYBERNETICS A TO Z
min-max = 2 max-min = -1
ways clashing. A strives a t the m axim um from the expected gains (this is ter
med m axim im ); B , in his turn, w ants to m inimize the maximum sum he will
have to pay A (this is term ed m inim ax). 5=3
As you can see, the m ain aim of the player is to get m axim um gain at the ex
pense of his artful riv al, who is determ ined to beat him . This means th a t the
player w ants to win b u t at the same tim e cannot count on his rival to make an
error. The player knows th a t no m atte r how good his strategy is his adversary
cannot fail to make the best response. For this reason, when calculating paym ents
the player counts on the worst he can get.
B ut how to assess q u a n tita tiv e ly how much one of the partners (both so accu
rate, so free from em otions, incapable of m istakes and capable of logical reaso
ning) is going to win? This is fairly easy. For every game there is a definite pay
m ent th a t a good player receives from another good player. This paym ent is cal-
GAMES THEORY 127
led the price of the game. This price lim its the gain: we can never win more than
the price of the game, if our rival makes no m istakes.
Correspondingly, if we do not make m istakes, our loss will not exceed the price
of the game. W hich one of the rivals is going to receive the price of the game,
and which one to pay it, depends on chance. Such chances in most games of chance
for two perfect players are fifty-fifty. There are also games where nothing is left
to chance, i.e. the in itia l conditions are strictly identical for both players. There
the theoretical result is always a draw.
W ith this aim in view l e t ’s analyse in the superior position, and 100 if
another sim ple game w ith two adver it is not attacked a t all.
saries and two strategies. The sym L e t’s w rite it out as it is usually
bolic term for it is “the bombing done in the theory of gam es—in the
m ission” . Suppose th a t two “B lue” form of a m atrix.
bombers are sent on a m ission. One N aturally, the “R eds” and “B lues”
carries bombs, the other rad ar jam w ill have to change th eir stra te
m ing equipm ent, various equipm ent gies. I t has been calculated th a t for
for the assessment of damage, etc. every 20 cases of the first strategy
The bombers fly in such a form ation there would be 40 cases of the second.
th a t the first is better covered by Hence, the “B lues” should in the
the guns of the second than the se ratio 40 : 20 favour the covered po
cond by the guns of the first. There sition of the bom b-carrier. The same
is the danger of the bom b-carrier ratio (40 : 20) should hold for the
being brought down by a “R ed ” fight “R eds” in attacking the covered bom
er—th ere ’s only one fighter, and it ber.
can make only one a ttack before the Now we can calculate the price of
bombers reach th eir target. The pro the game for the “R eds” , i.e. the
blem is which plane should carry the chances of the bom b-carrier to reach
bom bs—the first or the second, and its target:
which one should the “R e d ” attack,
20 X 6 0 + '4 0 X 100 o e 2 c,
i.e. w hat should both adversaries do
20 4 - 4 0 0 3 /o
to achieve best results.
The following strategies are open. And now after tiresom e logical rea
“B lues”-1—-the bom b-carrier in in soning read about a scandal th a t
ferior position. took place some years ago in E urope’s
“B lues”-2—the bom b-carrier in su casinos.
perior position. A t first no one paid any atten tio n
“R eds”-1—attack the bom ber in to them . These young men strolled
inferior position. into casinos to watch roulette being
“R eds”-2—attack the bomber in played. They stopped at green ta b
superior position. les and entered w inning num bers in
Suppose, the chances of the bomb- th eir notebooks. W hen asked for their
carrier to survive are 60 out of 100 m otives they answered: “Ju st so, for
if it is attacked in the inferior posi the fun of i t . ”
tio n, 80 out of 100 if it is attacked Two m onths late r a storm swept
128 CYBERNETICS A TO L
over M onte-Carlo, this cap ital of ca and sent them over to th eir com pa
sinos. The same young men came nion in London. The la tte r fed them
back. B ut now they ceased taking into a com puter. E v id en tly , the com
down num bers and started playing puter m anaged to do som ething th a t
them selves. And they did it wi was beyond the power of m an: ha
th o u t losing. M arvellous! U nexplainab ving processed some m illions of num
le! bers it guessed several accurate w in
The journalists started looking for ning com binations.
an explanation. And they th in k they True, m any specialists m aintain
found it. I t appears the young men th a t th is could happen only if the
wrote down w inning num bers, these roulette had some constant defect.
strange arrays of num bers, not for B ut even if th is was the case, it was
fun. Moreover, they encoded them not known either to the owners of
GAMES THEORY 120
the casino or to the people who took W hat do we mean by the struggle
p a rt in the calculations. of one system against another? E xam
W hat was the am ount of calcula ples are m anifold in different fields.
tions the com puter had to perform? For instance, the theory of games can
This is not known. B ut some other be adapted for m ilita ry com m unica
facts are. One American m athem ati tio n s’ purposes, to a n ti-aircraft de
cian calculated th a t in order to find fence (of course, the problem s there
w inning com binations in the pip card are much more com plicated than the
game one would have to analyse 34 problem of the bombing mission th a t
m illion card com binations. True, this has been solved above), to problems
is beyond m an ’s faculties, b u t the facing a commander in action.
eager-to-win m athem atician got hold The work of an experimenter who
of a com puter. He set it to work on draws up a programme of action—a
a 10 thousand m an-hour program me, plan of experim ents—too, can be con
and the com puter calculated some sidered from the viewpoint of the
“guaranteed to w in ” com binations. theory of games.
In the “Am erican case” the compu The experim ents m ay be conside
ter program me (algorithm for the red to be a game between the scien
search for combinations) was aimed tis t and the nervous system of the
a t increasing the chances of a favou anim al which he studies.
rable outcome of the game, and to The economist planning the work
achieve it the computer tested moun of a producing p lan t “play s” against
tains of num bers basing its play the moves of his “adversary” —the
against an im aginary rival on the consumer.
rules of the theory of games. It will In some respects the relations of
be reasonable for you to rem ark th at the sides (for exam ple, in the court)
there w asn’t much sense in creating can be considered to be a game in
a new branch of m athem atics just which the rivals strive to a tta in op
- to help people win in the games of posite aim s.
chance. Of course, the theory of And in m ost cases in the games
games w asn’t created for the sake of we have to operate w ith numbers
this (le t’s note, by the way, th a t for and numbers again, to perform a
such games as roulette and lotto there m addening am ount of calcula
doesn’t exist a programme of ac tions.
tion which would always give a s tric t So i t ’s not just by way of chance
ly definite w inning strategy). th a t computers are being tau g h t to
Games of chance, models of games play different games: dominoes, che
play the p art of guinea pigs which quers, fifteen and, of course, chess.
serve to test much more im portant Chess, boasting of an astronom ical
problems of the theory of number of variants of the gam e
games. some 2 -10116!—open up a wide scope
Specialists in cybernetics try to for research.
evolve w ith the aid of the theory of The com puter m atch between the
games a rational line of conduct for American and the Soviet chess pro
all sorts of systems struggling against grammes continued for a whole year.
some other system . The scientists attached great im por-
9 — G16
130 CYBERNETICS A TO Z
Here is a diagram of a lab y rin th and its crossings. I t will serve as an introduc
tion to one of heuristic programmes called “the universal problem so lv er”.
The solution of the problem of exit from the lab y rin th by the “try -o u t” pro
gramme would involve the testing of all v a ria n ts —even those th a t are known to
be senseless, such as w alking around a crossing. In heuristic programmes chance
elem ents are also present, such as arrival a t a “good” crossing by chance.
To cut the num ber of crossings (or ways) being tried out “the universal pro
blem solver” makes use of the term inal crossing of the labyrinth-problem and
of the distance to this term inal crossing.
The com puter begins its work according to programme by testing variants
issuing from the entry. The testing continues u n til the programme reaches a cross
ing nearest to the term inal. In our diagram we denote such a crossing by B„.
Then testing begins anew and results in a new jum p—nearer to the term inal.
Now our programme has reached the crossing C3. New tests, followed by a new
134 C Y B E R N E T IC S A TO Z
jum p. And so on u n til in the course of some test the programme reaches the ter
m inal D e. T hus the last jum p leads to a successful solution of the problem.
Have you noticed th a t “the universal problem solver” divides a problem into
several sim p ler problems? T h a t’s w hat makes it different.
But have you noticed as well th a t notw ithstanding the difference of heuristic
programmes from sim ple “try -o u t” programmes the test m ethod still plays a
big p art in th e “universal problem solver”? So it tu rn s out th a t the com puter
can tackle su c h problem s only because it works very fast. A m an working much
more slow ly (a hundred, or a thousand tim es slower) can, on the other hand,
successfully solve problems of the “chess ty p e ”.
This conclusion makes some specialists cast doubt on the universal v alidity
of heuristic programmes. In th eir opinion creative heuristic activ ity of m an can-
HEURISTICS 135
I t is true th a t up to now heuristic pected situ a tio n . This was the idea
program m ing has, too, failed to d i behind the fam ous “electro n ic” chess
vulge com pletely the mechanism s of m atch between the Soviet and
hum an creative a c tiv ity . B ut a great A m erican com puters which took place
stride forward in th is direction has in 1966-67.
been made: new program m ing p rin A m erican scientists in all games
ciples have led to the idea of study adhered to a single heuristic pro
ing the work of the brain on a new, gramme. Soviet scientists used two pro
interm ediate level—on the level of gram m es—one sim plified, another, in
inform ation processes. This w ill make th eir opinion, more sophisticated. This
i t possible to link inform ation pro opinion proved rig h t—the second So
cessing w ith the physiology of the viet program me won.
brain. This m atch rig h t from the sta rt
This m ethod has been compared took on the character not of a sports
to the m ethod, used in chem istry, game, b u t of com paring scientific
of decomposing complex chem ical ideas. As is often the case in cyber
compounds into sim ple elem ents. netics, chess was used to test the
H euristic program m ing and heu principles of heuristic program m ing
ristics as a whole are of special im w ith the aim of increasing the capa
portance for the progress in electro b ilitie s of com puters in a wide field
nic com puters. Before the advent of of th eir application.
heuristic program mes electronic com Look a t a m odern com puter centre.
puters were able to solve only pro Even in such a highly autom ated
blem s rigorously lim ited by m athe p lan t m any prelim inary operations
m atical description. Now it appears are m ade “by h a n d ”. H euristic pro
possible to solve problem s devoid of grammes w ill p u t an end to this.
such a description. The possibility w ill become real of
From the point of view of heuris going over from p a rtia l autom ation of
tics m odern electronic com puters have “w hite c o lla r” work to full autom ation.
m any deficiences: they are s tra ig h t In the opinion of specialists, the
forw ard, u n in tellig en t, inflexible, not use of heuristic program mes in me
clever, etc. dicine, transport, astronautics, phy
The role assigned to heuristic pro siology and neurophysiology, produc
gram m es is to find m eans of m aking tion control and m any other im por
the com puter sharp and clever, ca ta n t fields of science and technology
pable of finding its way in an unex w ill be very effective.
I W hat Is W hat?
We call all the figures shown at the top the letter “A ”, despite the great difference of shape.
Any sign of the written language—a letter or a digit—may experience all sorts of distortion
in the course of writing.
M an' has the faculty of recognition cognition and revelation of its secrets.
from tim e im m em orial. And yet, up And here again we encounter a
to now scientists have not found paradox—one of m any th a t spring
out how he does it. How does he m a up when we deal w ith the riddles
nage on the basis of barely percep of hum an b rain , of hum an m entality:
tib le signs and frequently incom plete scientists d o n ’t know how m an builds
ch aracteristics to construct in his up an image, b u t they know the va
b rain the concept of an image? Of lue of th is faculty. They m aintain
the im age th a t plays a m ajor p a rt th a t the perception of actual pheno
in his perception of the surrounding m ena in the form of images enables
world and in the processes of its the memory to be used more sparin-
IDENTIFICATION OF IMAGES 139
gly. This is because the image makes and other living organisms to iden
it unnecessary for us to rem ember tify images. The a b ility to classify
innum erable concrete objects and phe complex situ atio n s in anim ate nature
nomena. The image, in p articu lar, is acquired through education. It
enables us to make use of accum ula is, therefore, advisable to apply the
ted experience. principle of education to the crea
Scientists sta te a u tb o rita tiv e ty th at tion of classifying autom ata. The
w ithout the ab ility to group objects la tte r is possible even if the designer
into images we would be puzzled is in itia lly ignorant of the features
by every new phenomenon (just like co n stituting the basis of classifica
the electronic com puter), since no tion, provided he disposes of an ade
object, no phenomenon is an exact quate num ber of exam ples of situ a
rep etitio n , a precise copy of those tions being referred to a certain class. ”
we m et before. Since it was deemed possible in p rin
How do we acquire th is faculty, ciple to create an identifying m achine,
the value of which can h ard ly be scientists all over the world began
overestim ated? looking for ways to im plem ent the
Through education, of course. principle.
In the course of education, as his The Am erican cybernetician F. Ro-
experience accum ulates, m an learns senblueth b u ilt one of the first m a
to classify w hat he sees, to recognize chines of th is kind and called it the
images. “perceptron” . The name stuck.
At th is stage scientists come face The “perceptrons” are reading m a
to face w ith the question which is chines. The scientists worked long
quite unexpected to the laym an but and hard to educate the com puters.
quite legitim ate for a cybernetician: A t last, these “lite ra te ” autom ata
can the electronic com puter be were born. To tell the tru th , their
taught to model the process of image looks can hardly be distinguished
identification? from those of th eir “illite ra te ” b reth
We are tem pted to answer this ren—the same fam iliar narrow m etal
question w ith a citatio n from a spe cabinet. Only w ith “eyes” —screens.
cialized encyclopaedia publication: This optical eye of the photo-voltaic
“The so lv ab ility in principle of the cell scrutinizes the tex t and enables
image identification problem follows the m achine to identify the images
from the a b ility of hum an beings of letters.
Random c o n n e c tio n s
R e c e iv in g d e v ic e N e u ro n s w ith O u tp u t u n its
am pl if ic a t io n
This is how a perceptron works. After the process of learning the letter “A ” has been repeated
many times to a satisfactory result, the system is “turned over” to the letter “B ”, and so on.
IG N O R A N G E
C O N T R A D IC T IO N
142 CYBERNETICS A TO Z
To this end he tells him about the features by which this letter m ay be recogni
zed, for instance, by two inclined sticks w ith a cross-bar in between.
However, th e re ’s another way open to the teacher. He m ay lay out before the
pupil 20 different letters A and tell him th a t they all are A ’s. After th at he will
show him 20 letters B , 20 letters C, etc. And the pupil w ill begin to recognize
the letters unfailingly. He has evolved for himself the visual image of each le t
ter and will not fail to recognize them , no m atter how they are w ritten.
The hypothesis of com pact sets is based on the assum ption th a t when man
sees^the lette r A a p o in t—the image of the le tte r —is in sta n tly fixed in his m ind.
W hen next tim e he is shown the same letter, h u t in a different handw riting, ano
ther point is fixed in the v icinity of the first. The different transcriptions of the
le tte r —the th ird , the ten th , the hundredth are all reflected by corresponding
new points fixed in his m ind. B ut all these points are arranged in a com pact lot.
Lots of points th a t reflect the different transcriptions of B form another com
pact set. The same happens w ith the images of other letters.
And each set is separated from the other by clearly defined boundaries—they
are divided by a sort of fence.
“ Is there a need for an identifying m achine?” y o u ’re going to ask.
Sincet>the “electronic calcu lato rs” work w ith the speed of lightning, it will
not be difficult for them to compare hundreds or even thousands of features of
various images w ith the standards th a t can be introduced into the computer
“m em o ry ” .
This is all true. B ut one sh o u ld n ’t forget th a t the capacity of the computer
“m em ory” is always lim ited. Moreover, there are no two images exactly alike
in the m inutest detail. I t turns out th a t, if all features are strictly taken into
account as should be done for a com puter, it is even impossible to find two iden
tical type letters. W hat then rem ains to be done? To describe each image w ith
scrupulous precision? This is a to ta lly im practical task.
And there is still another obstacle: com puter operation according to “sta n
d a rd s” is impossible w ithout the collaboration of m an. I t ’s m an who has to des
cribe the standard and provide a whole set of features by which the comparison
w ith the stan d ard should be m ade. The scientists, however, have set the goal
to teach the com puter to identify images by itself. This is no idle task aimed
only at solving an interesting theoretical problem, but one th a t w ill enable iden
tifying machines to be used in practice.
The need for such m achines in relating to the properties of the stra
practice is very great. H e re ’s a vivid tum , assesses its electrical, rad iativ e
exam ple of the way in which a per- and geom etrical characteristics.
ceptron could be useful. Im agine a T here’s lite ra lly an ocean of com
geologist who has to decide on the binations of characteristics th a t cor
evidence of geological survey whe respond to an oil-bearing stratum .
ther a given stratu m is oil-bearing. A w ater-bearing stratu m , from which
The interpreter-geologist (there is the interp reter has to distinguish the
such a profession) analyses the data oil-bearing stratu m , has just as m any.
IDENTIFICATION OF IMAGES 143
“ Mushrooming."
<44 IDENTIFICATION OF SPEECH SOUNDS
tu rally , everybody would like to W hat takes place inside the m achi
know how it can identify sounds. ne?
The answer is twofold: it is, a t the The processes are sim ilar to those
same tim e, easy and difficult to iden taking place when you ta lk over the
tify sounds. At first l e t ’s find out telephone or radio: sound oscillations
about the easy side—the acoustical, are transform ed into electric oscilla
physical aspect of the problem . tions. Special filters filter them ac
Sounds are, by n ature, vibrations cording to th eir frequencies. Then
of air, waves of varying length. Every th eir “p a tte rn ” is com pared w ith
sound is characterized by a corres th a t of the standards stored in the
ponding frequency. In consequence, m achine’s mem ory. This “p a tte rn ”—
sounds m ay be produced not neces the image of the sound—is the ave
sarily w ith the aid of vocal chords, rage sound th a t the m achine has been
they m ay be synthesized. tau g h t to identify.
For th is purpose the m achine is The production of the p a tte rn is a
m ade to listen to words pronoun rath er difficult and m onotonous job.
ced m any tim es by the same person For instance, in one experim ent the
and by different persons. N atu rally , m an under test pronounced the sound
everybody pronounces the same word “a ” 100 tim es. The sound was des
in his own m anner: the tim bre of cribed 100 tim es by 14 v arian ts of
the voice, the intonations, the p u rity pronunciation, also called images.
of pronunciation are all different. N ext the frequencies of appearance
The m achine has to “average” over of the resulting images were compa
ind iv id u al pronunciations, exclude red. O ut of 14 v arian ts one was en
individual hues, so th a t when it countered more often th an the others.
hears the fam iliar word in future it This image was recorded in the m a
makes no m istake. ch in e’s memory.
This is how speech is made “visible”. The microphone converts the sound into electric current
of corresponding frequencies. This current is passed through the frequency filter. Each filter cor
responds to a definite sound pitch range. Small electric lamps are connected to the filters, and
their brightness changes with the changes in the current intensity.
The brightness changes are recorded on the light-sensitive film. Below are fourteen different
pictures obtained when the sound “A ” was pronounced 100 times.
The computer “Brick” identifies words by the alteration of voiced sounds and sibilants.
The dictation typing machine identifies over fifty single-syllable words. Amplified sound is ►
passed through the filters where it is separated by frequencies and transmitted to the compari
son block. Here the sound is re-coded into numbers. Next the digital cede is compared with the
codes of the syllables to be identified previously recorded in the “memory”. If the recordings
coincide, this means that the appropriate syllable has been found, and the machine prints it.
DC
£
LU
CL
>-
Qco
<8 ro m
< C C LU
oooooooo £
r rL U H oooooooo-
lEz^
QQLU>^-
OOOOOOOO
OOOOOOOO
0)
c
OOOOOOOO.-
'X
S I0 U U
S A B |0 y
ra m ti
ow
DC •2
cu 8a)o
<
p +-■
cl
f f f f f < s
< 0-0
o“
z
DC
m u m
o o o o o o o o
jo s s a jd d n s / < jv
0 S jO U p u B J 0 1 |iu i| ( \ )
o ,jo s s a jd u jo o pun0 $ V
U
DC
D
<
150 CYBERNETICS A TO Z
in sta n tly hand you a precise and only the com puter program m ists who
comprehensive reference. have to conduct dialogues w ith the
Everyw here would there be room com puters, but inexperienced people,
for talk in g and understanding m achi as well: economists, who use them
nes. “E lectronic arithm om eters” wo in industry, controllers, supervising
uld perform arith m etical operations autom atic control system s. And in
on voice orders. Control system s wo the near future these w ill be supple
uld receive oral inform ation, process m ented by m any other professions.
it and issue necessary commands. Today electronic com puters can per
Such m achines could be employed in ceive dozens of spoken words. They
scientific research centres, in indus are like a sm all child th a t builds
try , in transport. up its vocabulary, learns to pronounce
And how great is the need for such words in order to s ta rt speaking “like
machines! A lready now it is not everybody” .
The greater the num ber of chance I t turns out th a t the am ount of
outcomes of an event, the more va inform ation is a m easure of the dec
luable w ill be the news of its re rease in the un certain ty of some si
sult, the more information will it tu atio n . To calculate it special for
carry. mulae are used.
News of an event w ith two equally Different am ounts of inform ation
probable outcomes contains a u n it of are tran sm itted through com m unica
inform ation called the b it. You have, tions channels. The am ount of in
probably, guessed th a t the choice of form ation passing through a channel
the u n it of inform ation was not ca cannot exceed its capacity. Vice ver
sual. W ell, indeed, it is linked w ith sa, the capacity of a channel is de
the m ost w idely used binary m ethod term ined by the am ount of infor
of coding inform ation. m ation it can tra n sm it in a u n it of
L e t’s try to grasp the notion, a l tim e.
beit in the most elem entary form, You m ay remember how one of
of th is general principle of q u a n tita Jules V erne’s personages, the journa
tive evaluation of inform ation th a t lis t Gedeon S pillet was tra n sm ittin g
is the cornerstone of the entire th e over the telephone a chapter from
ory of inform ation. the Bible to prevent his rivals from
We know already th a t the am ount using the telephone. In this case the
of inform ation depends on the proba channel was fully loaded, b u t the
b ility of specific outcomes. If an am ount of inform ation passed was
event has, in the words of scientists, zero, since the recipient received in
two equally probable outcomes, this telligence th a t had already been
means th a t the pro b ab ility of each known to him . This means th a t the
outcome is equal to 1/2. Such is channel was “id lin g ” , passing a
the p robability of heads or tails fal definite num ber of pulses th a t
ling out, when a coin is throw n. were not loaded w ith any inform a
If an event has three equally probable tion.
outcomes, as in our exam ple w ith the
chess tournam ent, the p robability of A t the same tim e, the greater the
each is equal to 1/3. Notice th a t the am ount of inform ation carried by
sum of the probabilities of all out each of the definite num ber of pul
comes is always u n ity , since one of ses, the more efficient is the use of
all the possible outcomes w ill cer the channel’s capacity. B ut in order
ta in ly m aterialize. to achieve such results inform ation
An event, as you understand, can should be coded ratio n ally , and an
have outcomes th a t are not equally economical, concise language should
probable. Thus, in a football m atch be devised for the transm ission of
between a stronger and a weaker intelligence.
team the p robability of the victory For th is purpose inform ation is
of a stronger team is g rea t—say, 4/5. thoroughly “ filtered” . To cite an
The p robability of a draw is much exam ple, in the telegraph code let
less, say, 3/20. The pro b ab ility of ters, le tte r com binations and even
defeat is, on the other hand, quite entire phrases in frequent use are
sm all. represented by a shorter sequence of
INFORMATION 155
The fourth suggested th a t the word the end, only the h at was retained
“h a tte r ” be deleted, too, since the on the signboard.
h at on the signboard leaves no doubt Of course, if people were to use
as to the progression of John Thom p only the economical codes w ithout
son. redundancies in intelligence, all “in
F inally, the fifth stated th a t it form ational form s”—books, reports,
was none of the buyer’s business w heth papers—would be extrem ely concise.
er the h a tte r’s name was John Thom p B ut they would lose in c la rity and
son, and advised to drop it. So, in elegance.
To th is end the signals are sent through a decoding device, after which they
become intelligence to the recipient.
The com m unications system has worked, the aim has been attained.
A ll the tim e we are speaking about com m unications channels using for the most
p art the telegraph as an example. B ut com m unications channels is a very wide
concept th a t includes numerous quite different system s, th is difference being
sometim es too great to be expected. To m ake the abundance of meanings of the
“com m unications channel” concept clear a few additional examples w ill suffice.
L e t’s look from th is point of view at the telephone. The speaker is the source
of intelligence in telephone com m unications. The coding device th a t turns the
sounds of words in to electrical pulses is the m icrophone. The inform ation trans
mission channel is the telephone cable. The earphone works as the decoding devi
ce. Here the electrical signals are again turned into sounds. And finally, the in
form ation reaches the “receiving device” —m a n ’s ear at the other end of the
cable.
Here is a com m unications channel of a quite different n a tu re —the live nerve.
Here, too, as in the technical system , the process of transm ission is the same for
all intelligence. True, th e re ’s a difference, since in the technical system the direcr
tion of inform ation transm ission can be changed, while in the nervous system such
a change is impossible.
And yet another exam ple—the com puter. The m ain features are again sim ilar.
The inform ation from one individual part of the com puter to another is tran s
m itted w ith the aid of signals. The com puter is an autom atic device for processing
CT3
C h annel
aD
O
D e c o d in g d e v i c e O u tp u t m e s s a g e
158 CYBERNETICS A TO Z
inform ation in the same way as a m achine tool is a device for processing m etal.
The com puter does not create any new inform ation, it only transform s inform ation
fed into it.
i 5 | m 5 S 5 5 S S ! i S S 5 S l S S S S ! S 5 S S S 5 ! S 5 S 5 5 5 S S 5 5 S S S S S ! S i i i 5 5 5 S551SlS5S55 555SS5SiSSSSSS5
i i u u u m m m t m m i i i m m u i m m m u m i i u u u i x i m m i i a i i u i m
n in |] i n m i i ] |i i m m i i i i i m m ? i i m i i m i i i m i ? n i i i i m i / i i i m m ||i i |
n m m m itm sitim 'tm tiim iM iiiiitttiiiH iitt m u i i H s m m m m m s m
i t i i i t y i i i t i i n i i i i y i i N « i i ) i t r » v * i * i i i t n i i t i i i i i i i q i t u » V f f i i K »
53 i ) j > 5 j > i m j 8 j ! » > n i j > > u m > j j > 3 » 8 i u i s n j i j m > » j s i u » » » ! j i j n j j > > j > U 3 ) 5
go to another m achine, the controller, th a t checks whether the cards were punched
correctly.
Suppose the holes on the punch-card in our example show the data of the work
ing order of a lathe operator. In a m onth in a large plant hundreds of thousands
of such cards w ill be accum ulated. W hen the tim e comes to calculate w orkers’
pay, to collect data about the execution of the plan or about production costs,
the electrical sorting m achine sets to work. It groups cards in to separate blocks
by various features w ith the speed of tens of thousands of cards per hour. Then
the cards are sent to the tab ulator.
Here, to begin w ith, the cards are read. This is done by a special block of 80
brushes—one for each colum n of the card. The brushes are connected by m eans
ii*
In this way the computer compiles a summary inventory-bill card for goods sent to the custo
mer from a set of cards. The summary includes the goods, their price, deduction, transport char
ges, etc.
166 CYBERNETICS A TO Z
of wires to the counters and the printing m echanism s. The punch-card moves
w ith the tens first. The brushes s ta rt by sensing every n in th position, then the
eighth, etc.
H ere’s a card w ith a hole in the eighth position of the th irte e n th colum n. The
brush w ill close the electric circuit, the electrom agnet of the num ber disc w ill be
energized, and the disc w ill s ta rt to tu rn . The card w ill move one position for
ward to the seventh position. The disc w ill tu rn the angle of the num ber and
w ill show 1. The card w ill move yet to another position—the s ix th —and the disc
w ill show 2. The tu rn of the disc is completed as the brush reaches the zero posi
tion. The disc has turned through each of the eight positions and now shows 8.
The eight from the punch-card has been thereby transported to counter file th a t
corresponds to the th irte e n th colum n. A ddition is done in the same way.
And who controls a big complex computer?
Ju st these punch-cards and brushes. To enable the electric pulse to tra v e l in
side the m achine all electric circuits connected w ith the brushes, counters, p rin t
ing mechanisms are m ade to term inate on the com m utating panel which w ith the
aid of switches distributes the electric pulses over the entire m achine.
A calculating and perforating set. The perforator punches holes in the card. The controller
checks and sorts them. The tabulator—the main machine of the set—counts the numbers: 1—
magazine; 2—contact brushes; 3—contact shaft; 4—card-guiding rollers; 5—contact; 6—coun-
►
JACQUARD’S METHOD 167
num bers can be summed up sim u lta sed. B ut the punch-card and the p rin
neously—70 thousand additions per ciple of organization of the calculat
hour! ing process in the com bination of
A hundred accountants w ill be able punch-card ca lc u la tin g m achines set
to perform during th is tim e only 25 a lim it to the productivity. A point
thousand arithm etical operations. to note is th a t the calculating speed
According to their principle of ope of the ta b u la to r is some 15 tim es
ratio n the punch-card calculating m a higher th an the speed of key-board
chines are subdivided into m echani sum m ing m achines, while the produc
cal, electrom echanical and electro tiv ity of the whole com bination is
nic types. only 3 to 4 tim es higher. This is the
Y o u ’ve ju st read about the elec resu lt of the, as yet, great p a rt
tric m achines. In m echanical m achi played by m anual labour: the cards
nes the holes are sensed— “read ” —by are controlled and punched by operator
special ten tacles—the pins. Such m a and transported from m achine to m a
chines perform only 100 operations chine by m an. And even if the lo t of
per m inute. punch-cards carried a t a tim e is much
The change over from m echanical greater in th is case th an in the case
and electrom echanical devices to elec of key-board m achines, the slow speed
tronic devices resulted in the speed of card processing in m achine’s
of punch-card m achines being increa mechanisms greatly influences the
overall pro d u ctiv ity of calcula p art in the calculating process only
tion. by th eir concerted operation.
The supply of punch-cards is th a t Only after a century of progress in
num ber “depot” th a t Babbage tried technology were the engineers able,
to build w ith the aid of counters work using the m odern counting and pun
ing a t the same ra te as the “factory” . ching set as the basis, to s ta rt work
He understood th a t m an could be on advanced calculators and la te r on
liberated from the necessity to take electronic com puters.
pherer in this case you'll have to arm I t ’s a general rule that a solution for
yourself with paper and scissors. any ciphering system may, in prin
Draw 64 chess squares on a piece ciple, be found by the simple trying
of paper. Cut out holes precisely to out of all the keys possible in each
the drawing, and you’ll get a grid. case. But this trying out is to be
Lay the grid over the disordered continued until the unique key is
set of letters with number 1 directed found that w ill make the cipher talk.
upwards. The American scientist Claude Shan
Look, a text has appeared in the non carried out very informative
holes: electronic com put— Now turn calculations to the effect. He set him
the grid clockwise by a quarter of a self the task to find the key to a ci
revolution. Y ou’ll get another part pher consisting of only 26 possible
of the phrase—ers are able to so lv.... key combinations. Twenty-six is a
Another such turn yields e compli very small number. One has to use
cated m a th .... And the last: ematical these 26 key variants, only one of
problems. In the case of this cipher which is the correct variant, substitu
the grid serves as the key that enab ting in turn all the 26 letters of the
les the message to be read. Every English ABC. The scientist obtained a
secret inscription, every cipher nor formidable number—1012! T h at’s the
m ally has only one correct solution, number of years that w ill have to
a unique key, the secret of which is be spent in search of the key to the
to be guarded. Even when a message cipher.
not destined to be seen or heard by And this in conditions extremely
strangers falls into the hands of the favourable for the imaginary oppo
enemy, it remains silent until the nent: Shannon assumed that the op
key to it is found. ponent invented an electronic device
KEY TO THE CIPHER 171
With the aid of this grid you’ll be able to read what’s written here.
for trying out • the keys at a Here’s a vivid proof of the immense
speed of one key per microse discrepancy existing between the “so
cond. The opponent will find the lution of the deciphering problem in
right key after trying out, approxi principle by the method of trying out
m ately, one half of the possible com the keys” and its practical realiza
binations! tion.
Enem y c ip h e re r
There are always two terminals: the transmitting and the receiving.
The transmitting terminal is always connected to two sources of information:
one is the source of intelligence that has to be transmitted, the other is the source
of keys that determines the key for ciphering by choosing one definite key out of
all the keys of the system.
The completed cryptogram is transmitted along the communications channel.
The communications channels may be of different kinds: messenger, post, telegraph,
radio. At the receiving end the other cipherer with the aid of the key reproduces
the intelligence from the cryptogram, i.e. deciphers it.
It is natural to presume that the enemy w ill, certainly, try to intercept the
message. Therefore, another factor w ill act on the cryptogram during its transmis
sion along the communications channel—the cipherer of the enemy.
Such is the general lay-out of a secret system proposed by Shannon.
If you take a look at this lay The monument of the written lan
out, you w ill not fail to conclude guage, itself, may be regarded as a
that its bounds are much wider than “cryptogram”, consisting of the source
the secret system in its “pure” form. of intelligence (in this case i t ’s
L et’s take, for example, a message what the author of the text wanted
in some unknown “dead” language. to transmit) and the source of keys.
KEY TO THE CIPHER 173
'{in our example i t ’s the ABC used my investigations once and for all,
by the author). The scientist who since I lost all hope of ever achieving
tries to find his way about in the sec any satisfactory results.” It took
rets of the unknown language and years to understand the Babylonian
read the text of interest to him, will inscription of Xerxes: “Xerxes, the
assume the role of the cipherer ope great king, the king of kings, the
rating at the receiving terminal. True, son of Darius, the king, Achaemenid”.
in our case the intercepting cipherer Couldn’t electronic computers be
is totally out of the question. used to decipher ancient manuscripts?
What an enormous number of keys The statistical method is of great
the scientists deciphering forgotten help in the case. The essence of the
written languages have to try out! method entails the precise knowladge
W hat toil, what patience and, fre of the signs contained in manuscripts
quently, despair accompany their gi that have yet to be read and of the
gantic work! regularity of the appearance of these
It’s a known fact that the famous signs. For instance; the ancient Egyp
Rawlinson who succeeded in reading tian writing contains up to 800 differ
the clay tables of the Babylonians ent characters, the H ett—some 500.
wrote in 1850: “I must openly admit The syllable systems of languages
that when I, after having identified contain from 50 to 80 sounds. The
every Babylonian sign and every Ba European languages usually contain
bylonian word, for which I could about 30 sounds, the Polynesian
find support in the tri-lingual inscrip only 10-12; some Caucasian langua
tions, tried to apply the information ges, on the other hand, 70-80.
to interpret Assyrian inscriptions, I Accordingly, the experiments aimed
was frequently tempted to give up at deciphering the written language
T ‘{ V -
v <3<m
7$
’ S W Yf >7^7 Tf
of the Maya carried out in Novosi out a computer algorithm for deciphe
birsk at once lead to the conclusion ring forgotten written languages.
that Maya signs can be neither of the This algorithm has already been
purely character type, nor of the tried: it was used to read control
purely ABC type. The language con texts, texts which had been read
tains 340 signs. No ABC w ill hold as before and had been offered to the
many. On the other hand, 340 signs computer as “examination papers”.
are insufficient for character writing. W ell, the computer, adhering to the
In all, seven methods were used “rules for action”, dealt effectively
to decipher Maya texts. And each even with short texts that are very
proved useful: it either supported difficult to decipher.
the results or controvened them. This computer algorithm was also
The computer was at work for two used for independent research. The
days. It performed a billion opera electronic computer was instructed to
tions. Deciphered 40 per cent of the clarify obscure texts relating to the
text. To decipher all known Maya ll-1 2 th centuries found on the terri
texts 200 computer hours w ill be nee tory of modern Mongolia—the Ki-
ded additionally. The computer w ill dan inscriptions.
have to perform billions of operations. In the course of previous studies
W ell, it can be agreed that the the Kidan written language was
first step in the computerized deci classified as relating to the Turk
phering of forgotten languages has or Tungus-Mongolian groups of lan
been made. It is to be hoped that guages. The computer “swallowed”
in due tim e the scientists decipherers large texts, studied the regularities
aided by a mathematical cryptogra of the language and confirmed that
phic system and electronic computers the language of the once m ighty state
w ill be able to vitalize numerous of the Kidans was akin to Mongolian.
ancient inscriptions that up to now W ell, le t ’s bid the computer-deci
have remained silent, and those “dead”, pherer a happy journey, le t ’s wish it
“forgotten” written languages w ill new successes. The more so because
tell stories of peoples of by-gone centu specialists consider deciphering his
ries, of their life, of their culture. toric written systems to be a particu
And those are not idle hopes. There lar case of the general problem which
is a sound basis for them: the they have termed “the problem of
Soviet specialist M. Probst has worked formal research of the language”.
175
L Chocolate and “A lg o l”
of instructions th a t the com puter th a t every class of com puters has its
m ust carry out to solve the problem own “d ialect” , which only it can
is recorded in a special (binary—0 understand. Years of existence of
and 1) code. This lis t of instructions high-speed electronic “counters”
together w ith th eir sequence is called brought into being alm ost 5000 artifi
the program m e. I t d istributes all the cial languages! “Cobol” , “ F o rtran ” ,
operations of the com puter, describes “J o v ia l” , “A lpha” , “APS” , “Alco-
a ll calculating processes. pol” , “M athem atic” —th ey are too
This is how a recording made num erous to be enum erated.
in the com puter language looks: T h ere’s no end to difficulties th a t
0001 - 0000001010. spring from such language discrepan
One th in g is clear to the lay m a n — cy. To tra n sm it problem s form ulated
i t ’s some sort of a code. B ut w hat is for one com puter to another of diffe
it? And t h a t ’s the w ay the com puter re n t design the program m ists have to
reads it: “Add the num ber from the make up a new program m e. W hat
cell num ber ten of the w orking ‘me costs in labour and tim e! Im agine a
m ory’ to the num ber in the sum- calculating centre em ploying lim ited
m ato r.” equipm ent. The day the centre is
Here w e’re up against a definite expanded and the equipm ent is
suprem acy of 0 and 1 over the n atu ral changed the com puter language ex ist
language: ju st fourteen signs, and you ing a t the centre w ill “break down”.
have a sentence of fourteen words. For th is reason program m es com
H igh capacity out of all qualities has piled a t the level of in stru ctio n s—at
c e rtain ly been realized in the com pu th e level of a specific com puter—
ter language. could no longer satisfy the scientists
Owing to th is i t is w ell adapted in conditions of contem porary state
for inform ation exchange between m an of science and in dustry, when gigan
and com puter and betw een com puters. tic am ounts, lite ra lly avalanches, of
Com puter language also helps in case calculations are required. They de
of inform ation exchange between cided to go over from “local com pu
people effected by m eans of com puters. ter dialects” to the language of a u to
I t enables com puters to conduct “d ia m atic program m ing—a veritab le com
logues” w ith other system s and w ith p uter language, which every com
system s w ithin th e com puter itself. puter should be able to under
Its range of action is, as you can see, stand.
q u ite wide. This task is m uch more difficult,
However, besides advantages the m uch more com plicated th an the task
com puter language has some defects. of sim ply enum erating instructions in
And the m ain, q u ite essential one is a program m e.
W hat is the autom atic program m ing language for? The purpose is always iden
tical: to help the program m ist tell the com puter w hat to do.
Specialists compare such an artificial language w ith th a t p a rt of the spoken
language which is used in instructions telling people “w hat to d o ” . For instance,
i t ’s very m uch like th e language of an ordinary cooking book. Analogy w ith some
D E S C R I P T I O N P R O C E D U R E S
IN G R E D IE N T S OF T H E O R D E R OF P R E P A R A T IO N
CAN DY
M ix th e f i r s t f i v e in g r e d ie n t s
in a s im m e r in g p o t s t ir r in g
c o n t in u o u s ly u n t il th e s u g a r
d is s o lv e s
2 c u p f u ls s u g a r
S t ir r in g o c c a s io n a l ly s im m e r
u n t il a d ro p t u r n s in t o a s o f t
b a ll in c o ld w a te r
1 c u p f u l m ilk
" a
V4 a te a s p o o n fu l s a lt
T a ke th e p o t o f f th e f ir e a nd
a dd b u tte r
L e a ve to co o l (no s tir r in g )
u n til c o ld e n o u g h to h o ld
12—616
178 CYBERNETICS A TO Z
recipe from the cooking book is usually resorted to in order to describe the classes
of gram m atical forms characteristic of an artificial language.
L e t’s take the recipe for m aking chocolate candy (see illu stratio n on p. 177).
I t certainly h a sn ’t escaped you th a t the recipe is subdivided in to two parts.
The first consists of a description of stuffs needed to m ake the candy. The second
describes procedures th a t are to be carried out in sequence in order to solve the
problem : m ake the candy.
Sim ilarly, the languages of autom atic program m ing contain gram m atical forms
of two types. They are called descriptions and procedures. D escriptions, for their
p a rt, are subdivided in to tw o types: the description of d a ta —a list of component
p arts, and th e description of program m es—subprogram m es. The procedures th a t
are to be carried out in accordance w ith th e recipe are sim ilar to an ordinary com
puter language program me. N ext follow th e procedural operators—they corres
pond to sentences and describe operations which are to be performed w ith th e com
ponent parts.
In general, operators and d a ta descriptions in the artificial com puter language
consist of expressions which m ay be directly b u ilt up from num bers, w ords, ab
breviated designations of m easures, auxiliary sentences and word groups. Expres
sions corresponding to words consist of com binations of symbols.
This interconnection of the structures of the “w hat to do” type is characteristic
both of the language of a cooking book and of the autom ated language. The diffe
rence lies exclusively in actions of procedural operators, which depend on the
field of application of the language.
The operators “m ix ” , “cool” , “b e a t” , “cut in to p a rts ” , “s tir ” , for exam ple,
are relevant to the process of food preparation. C alculating processes, on the other
hand, en tail different operators: “extract a square ro o t” , “take the 24th degree” ,
“a ttrib u te the v a lu e ” , “repeat the following calculations u n til....”
ordinary natu ral language. Due to journey, since the specialists are a l
this, a “Cobol” te x t looks very much ready discussing the problem of in
like a natu ral language text. creasing the convenience of autom atic
Now you can see w hat a long and program m ing. And the param ount pro
tedious journey the com puter langua blem here is said to be the liberation
ges have made from “dialects” to of the algorithm ic language from its
several general universal languages. d irectiv ity , from its ad ap tatio n to
T h ere’s no doubt th a t th is is rath er specific fields of action, to specific
the beginning th an the end of the class of problems.
LINGUISTICS, MATHEMATICAL 181
1001001—animal
1000100-bird
1001101—eagle, etc.
In compliance with the programme and with this code the computer w ill choose
words close in meaning. The initial text introduced into the computer w ill serve
as a basis for this “creative” work. The “work of literature” is produced in cyc
les. W ith each cycle of programme repetition the computer extends the basic text,
every tim e departing from it still further, however not beyond reasonable lim its,
so that the text would not lose sense altogether.
This is followed by a process of constructing phrases. Using the programme
instructions the computer combines all the words into sentences in accordance
with the rules of grammar.
And now some examples of how to write white verse. The word
computer literature—in recent years reserve of the poet is 130 words. The
a veritable “collection of works” metre of the verse is strictly fixed.
written by different computers has The computer writes 150 quatrains
been built up. per minute. It doesn’t name its verses,
The computer “RCA-301” learned just numerates them.
Verse No. 027
W hile life creates false totally empty images,
W hile slow tim e flows past useful deeds,
And the stars dejectedly orbit the skies,
People cannot sm ile.
Poem No. 929
As sleep blindly streamed
Over shattered hopes,
Cosmos exuded with pain over ruined love.
Your light was slowly exiled
Out of secret men
And the skies slumbered not.
And here’s an example of the writings of MUC—the “electronic brain” of
the University of Manchester.
A Love Letter
My little treasure! My comprehensible devotion
wonderfully attracts your tender delight. You
are my loving adoration, my breast-widening
adoration. My brotherly feeling with secret breath
awaits your dear restlessness. The adoration of
my love tenderly keeps your greedy zeal.
Your lovesick MUC
LINGUISTICS, MATHEMATICAL 185
ventor Raimon Lull, though he na Since in this logic we are dealing
med his creation an “ object of great with mathematics the concrete mea
art”, had no idea of the rules of ma ning of a statement is of no impor
thematical logic which serve as a tance to it. One thing only is impor
basis for the operation of modern tant—whether a given statement is
logical computers. true or false.
The great German mathematician Note also the following. Actually
and philosopher Gottfried Wilhelm everyone of us, so to speak, devotes
Leibnitz is considered to be the foun all his tim e to practising formal lo
der of mathem atical logic. He was gic. Involuntarily we, especially when
the first to try, in the 17th century, trying to stress our point, adapt our
to construct logical calculus of the selves to the laws of linguistic algeb
arithm etical and algebraic types. He ra, for (in the opinion of the specia
narrowed the gap between logic and lists) the mathematical logic can be
calculations, improved and specified said to form the “skeleton of our
the symbolics of logic. thought”. It serves as a basis—natu
On the foundation laid by Leib rally, in a very broad sense—for the
nitz another great m athematician Ge general properties of statements and
orge Boole, father of E. Voynich, the of reasoning.
author of the novel The Gadfly, built Imagine yourselves answering the
the temple of a new discipline—the question “ Why is it light in dayti
mathem atical logic. He devised a me?” with the words “ Because light
special algebra adapted to logic con shines in daytim e”.
structions. As distinct from ordinary You have violated the rules of logic,
algebra, this algebra uses symbols to the logic of reasoning, for you haven’t
designate not numbers, but statements. explained anything. Unsubstantiated
When children learn to count, they method of thinking of this sort was
are embarrassed when having lear branded by Moliere in his immortal
ned, for example, how to add two comedy Le M a la d e im a g in a ire. In
apples and three apples, they have the course of the play a bachelor of
to add three houses and two houses. science is being examined: “ Why
But later at school the children get does opium promote sleep?” The ba
it into their heads that numbers chelor’s answer is: “ Because it con
exist by themselves and do not ne tains soporific power, which is ca
cessarily serve to designate the num pable of lulling senses to sleep.” As
ber of apples, houses, cars, etc. Chil we see from this example, here the
dren learn when studying algebra words of normal language taken to
that the concept a 2 is much wider gether in context are devoid of clear,
than 13 apples m ultiplied by 13. precise meaning.
The same is true of mathematical There is an ancient reasoning that
logic—one has to abstract himself became a classical example of logical
from the contents of statements. They proof: “A ll men are mortal. Socrates
are of no importance to m athematical is a man. Therefore, Socrates is mor
logic, just as algebra is indifferent to ta l.”
what x stands for—the number of Scientists invented unique symbols
fish, cars or stars. to enable precise meaning to be attri-
LOGIC, MATHEMATICAL 189
buted to words. This system of sym and the deduced judgement is called
bols enables the logical structure of the conclusion.
an idea or of a judgement to be disco L et’s take the judgement “ All com
vered. W ith the aid of precise defini puters make m an’s work easier” as
tions of words it is possible to calcu the major premise, and the judge
late the interconnections between sym ment “The arithmometer is a compu
bols of words, the relationships exis ter” as the minor premise. Then the
ting between them. Things which conclusion drawn from these judge
were formerly difficult to express in ments will be “The arithmometer
words of the normal language can makes m an’s work easier”.
now be easily represented by symbol The judgements may be true or
of mathematical logic. false. If they are true, then, provided
H ere’s a simple example: H a0 is we adhere to certain rules of construc
the symbol of water. It states that ting syllogism s, we shall always ob
there are two atoms of hydrogen and tain correct conclusions, true inferen
one of oxygen to a molecule of water. ces from the judgements, as, for
In this transcription the facts relate example, our conclusion about the
to chemistry, the numbers to mathe arithmometer.
matics and the symbols to the logic But if, while adhering to the rules
of symbols. of syllogism construction, we obtain
L et’s undertake a step-by-step in a false conclusion, this means that
cursion into the laws of logic. Three at least one premise was not correct.
different forms in which thought And vice versa, if all the premises
is realized are considered in logic: were true judgements, but the deduc
concept, judgement and deduction. tion resulted in a false conclusion,
“This inscribed angle, resting on this means that some law of syllogism
the diameter” is a concept. In case not construction was violated. We may
all the angles are embraced by it, obtain the absurd conclusion “ All
it is a singular concept. “ All inscri pupils are poor pupils”, if we con
bed angles, resting on the diameter, struct the syllogism incorrectly: “ Pet
are right angles” is already a judge rov is a poor pupil. Petrov is a pupil.”
ment, since it reveals the properties It turns out that the solution of
of the object of judgement. every logical problem, as well as a
The process of deducing a third mathematical problem, has its own
judgement out of the two is called “technology”. I t ’s made up of ele
deduction. Its principal form in logic mentary operations resembling addi
is called the syllogism . One of the tion and m ultiplication. There are
judgements in the syllogism —the ge special rules for these operations, and
neral one—is called the major premi with their aid anyone is able to solve
se, the other judgement—the speci a complex logical problem. The same
fic one—is called the minor premise, is true of an automatic computer.
arbitrary logical judgements? For example, how should the judgement “I shall
certainly go to the football match if I get a ticket, or if a friend invites me, and
if it is not raining” be represented by a formula?
W ell, there is a way. But some logical judgements are needed, by way of ex
planation.
The judgement of the football fan contains several reservations. Here they are:
1. I w ill get a ticket.
2. I w ill be invited by a friend.
3. It w ill be raining.
Simple judgements may be interconnected by the words OR, AND, NOT. Com
posite statements are always expressed by simple ones with the aid of these words.
They, like three jinns, can do everything in logic.
We can agree to designate each of these words, for the sake of brevity, by some
symbol, just as in mathematics corresponding signs are used instead of the words
“plus”, “m inus”, “m ultiply”, “divide”.
The word OR is usually designated by a cross ( + ), and the word AND, by the
sign of m ultiplication—the point (•).
Now it remains, for the sake of brevity, to introduce a letter for each statement.
Its negation, i.e. the word NOT, w ill be designated by the same letter, but with
a dash on top. For example:
I w ill get a ticket—T. I w ill get no ticket—T.
I w ill be invited by a friend—F. I w ill not be invited by a friend—F.
I t ’s going to rain—R . I t ’s not going to rain—R.
Now we shall write with the aid of these symbols the composite statement “I
w ill get a ticket, and i t ’s not going to rain, or I w ill be invited by a friend, and
i t ’s not going to rain”.
LOGIC, MATHEMATICAL 191
Here it is:
T-R + F R
As in algebra, the common m ultiplier can be taken out of the brackets. Now
the formula w ill take the form:
(T + F ).R
M = (T + F ).R
As you know, every logical judgement may be either true or false. I t ’s been
agreed in mathematical logic that true statements be expressed by one, and false
by zero. Again 1 and 0. “I ’ve learned all the lessons”—this was what a pupil
told her class instructor. L et’s designate this statement for the sake of brevity by
the letter L. If this statement of the girl is true, L = l . Thus, if L = l , L = 0 , since
in this case the statement “I have not learned my lessons” would be false.
And vice versa, had the pupil not learned her lessons, L = 0 and L = l .
This is a general rule in logic. If some statement B = l , then B = 0 , and vice ver
sa: if B = 0 , then B = l .
It remains for us to observe a few additional propositions of mathematical logic
to know all its principal laws.
Obviously, B + B = B and B -B = B .
Clearly, the composite statement “I ’m going for a walk or I ’m going for a w alk”
is fully identical to the simple one “I ’m going for a w alk”. Similarly “The recei
ver w ill work and the receiver w ill work” is tantamount to the simple statement
“The receiver w ill work”.
Two following propositions are just as obvious:
B+ B = l and B - B = 0
The need arose of performing tens of The computer does it quicker. But
trillions of arithmetical operations. how much quicker? In a day, an hour,
Even with a speed of ten thousand a minute?
operations per second it would take How many, times quicker—that is
a high-speed computer over four years the question. And then: what should
of continuous operation to do the high-speed computers look like, what
job. For instance, the solution of a elements should they be made of,
problem relating to planning and con what w ill their dimensions be?
trol of the economy requires Now that computer engineering has
10 000 000 000 000 000—a number its history we can glancing back per
with sixteen zeros (1016)—calculations ceive the changes that the computers
to be performed. have undergone from generation to
Three and a half m illion slow-speed generation, and follow the trend of
computers would be needed to comple their development.
te this enormous job. This is unfe The ancestors of present-day compu
asible. Only one alternative remains ters were the electromechanical slug
—to make the computers work faster. gards into which all the ingenuity of
Working eight hours a day you the fourties was packed. Their time
could count to a m illion in three and a of operation was in the m illiseconds
half months. A billion would take range.
you ... 500 years to count.
nishing the volume of various blocks of the computer as a whole but should be
based, in the first place, on minimizing the dimensions of separate components,
on increasing their packing density with a simultaneous decrease in the number
and length of all connections between the blocks. Now the designers went to any
length to make a midget out of the computer.
And the simple transistor which recently was triumphantly marching through
the entire electronics has modestly relegated to the background. Instead, new
components took up the struggle for minimum dimensions and maximum speeds
of computers. Great was their number: optotrons and cryotrons, high-frequency
transistors and tunnel diodes, spacistors and twistors, biaxes and transfluxors,
persistors and cryosars, parametric devices and technetrons. A tenfold decrease
in the volume of various apparatus became at once a reality. And operating speeds
shifted to the hundred thousandths and even m illionths of a second range.
But even this colourful array of subminiature, ultra-fast, super-reliable compo
nents soon made way for thin films—the building stones of third-generation com
puters.
Three generations of computers in the spell of twenty years!
At the beginning of the fifties—tube computers, at the beginning of the sixties
—transistorized computers, and at the beginning of the seventies—computers on
integrated circuits (thin films).
W ith the aid of these new building stones it proved possible to build a world
of tiny giants, to erect electronic cities of unusual architecture. Recently, the de
signer would proudly demonstrate a computing block of the computer the size of
a tin of sardines. Now, engineers dream of packing 350 thousand circuits into
one cubic decimetre.
Where do these magic qualities of the thin film stem from that allow engineers
to achieve so much?
Usually films are produced by evaporation. The appropriate material is heated
in vacuum under ambient pressure of a thousand m illionth of an atmosphere. The
material evaporates and condenses on a glass or a metal plate. This fine work is
made much more intricate by the necessity of arranging the particles not arbitra
rily but in accordance with a strict pattern. “Electronic stitching” is based on the
process of condensation through the slits of a mask repeated many times. The
structure of the layers of the film used in circuits—there may be ten, fifteen or
more of them —must coincide exactly. To show the difficulty of this job it suffices
to state the thickness of the film—i t ’s only 100 Angstroms (one hundred thou
sandth of a millimetre thick). The right to use the word thickness with the film
being thinner than a ten thousandth of the thickness of a safety razor blade re
mains in itself questionable.
This film circuits are complete electronic circuits. Thus we witness not the art
of assembling circuits from separate blocks, from separate building stones, but
a supreme mastery of matter when every particle of it in compliance with the
wishes of the creator occupies a place allotted to it.
This is the quality that now assumes major importance for computer designers.
M illions of devices per cubic m illi ters, or, to be more exact, of highly
metre in conjunction with speeds of productive communities of entire com
m illions of operations per second. puter systems with a total memory
Operating speeds of third-genera of billions of bits and operation
tion computers are expected to rise speeds of billions of operations per
by about two orders of magnitude: second.
108 operations per second, th at’s what An absolute superiority of midgets
their speed is going to be. This is over giants! The midgets won again,
only an order of magnitude less than having succeeded in opening the doors
the lim it set by the velocity of elec of the kingdom, unseen and unheard-
tric pulse propagation in solids. of before, of the kingdom the entrance
And the fourth generation? These to which bore the inscription “na
computers are still more advanced. nosecond”—a billionth of a second.
Speeds up to 10s operations per se Such computer speeds, though it
cond, a working memory store hous sounds like a paradox, can be neither
ing 108 bits of special inter-computer seen nor imagined, but they can be
information units. This unit is equi attained and utilized.
valent to two decimal digits or one This marks a complete triumph of
ABC sign. This amount of 108 bits is technology: something has been made
sim ply hard to imagine. Let us by by hand that defies im agination.
way of an example translate the volu However, as is often the case in
me of computer external memory sto technology, every new achievement
re exceeding 1012 signs into the “book creates a new problem.
language”. The result w ill be m il The lim it for the speed of operation
lions of volumes of 500 pages each! of third-generation computers is set
The fourth-generation computers by the velocity of propagation of
boasting such parameters and such a electric pulses in solids.
structure are, in effect, real communi To overcome the problem one has
ties of the second- and third-genera to make a detour.
tion computers. This trick was ac The way as the specialists see it is as
complished with the aid of a new sub follows. Microscopic devices have been
miniaturization instrument, the BIC. designed for the conversion of elec
Big Integrated Circuits represent tric signals into light signals, and vice
structural complexes of numerous ele versa.
ments. Just compare: the envelope Sim ultaneously fibre optics has co
of one semiconductor device houses me into existence. The latter makes
only the device itself, of an integra use of thin transparent filaments,
ted circuit—up to ten devices, and with the aid of which light can be
of a big integrated circuit—over a transmitted along any straight or cur
hundred. And this is not the lim it. ved path connecting the elements of
In the future BIC w ill, probably, grow the circuit, in the same way as elec
to become GIC—Gigantic Integrated trons are transported along metal
Circuits — containing several thousand wires.
elements. As a result, in addition to the elec
GIC may be regarded as the build tron the particle of lig h t—the photon
ing blocks for fifth-generation compu —was, too, harnessed to the electro-
MICROMINIATURIZATION 201
nic cart. In opto-electronic systems ter with the reliability of the brain
information fluxes flow at the same too bold? Isn’t this dream baseless?
time along electric and optic chan It turns out not. Using fibre glass
nels, whose joints are provided with a laser device can be built that w ill
opto-electronic and electric-optic con operate on the principles of a living
verters. The use of optic connections neuron. Light-conducting filaments
and of optic methods of information w ill play the part of nerves transport
processing has given electronics a new ing impulses. The pattern of operation
degree of freedom, has substantially of such a computer w ill im itate the
increased its possibilities and opened action of the brain’s neurons and the
up new horizons. Thus emerges the nervous system. This hybrid of tech
shape of computers of the following nology and electronics w ill, in effect,
generations. Life w ill be blown into make a synthetic brain.
them not by electric current but by a It would seem that everything that
ray of light. Now the phrase “the could be desired had been realized:
computer radiates thought” w ill sound speeds defying imagination, wonder
quite real. ful reliability and utmost compact
The principles of design of future ness. But people with a foresight are
optic computers w ill be quite diffe already able to discern the dim con
rent from those utilized in the design tours of new machines.
of electronic computers. Light pulses An attentive reader, you have not
of a hundred billionth of a second du failed to notice that the description is
ration switch a lazer system on and sim ply “such m achines”, without the
off practically w ithout delay. customary epithet “thinking”, or the
Fantastic speeds led to fantastic rigorous definition “computing”.
dimensions of the computer—they ha Why?
ve reached the absolute minimum, the New machines are expected to re
part of the calculating element in volutionize technology in the same
them being played by molecules and way as this was done some twenty
even atoms. The most acute problem years ago by electronic computers
here is reliability. Since repair work which replaced electromechanical cal
on such midget computers is not fea culators since their speeds w ill be
sible they have to work without no less than 1020 logical operations
faults. This, the engineers decided, per second. Try and find an epithet
could be achieved. Such devices exist. for such a machine. Their operating
For instance, nature has nursed the principles w ill, too, be quite diffe
human brain. Its reliability is per rent.
fect. It works without repair or stop Imagine yourself reading a book
page about 70 years, despite the not line by line but whole pages at
fact that every hour of human life a glance.
some 1000 neurons die—that makes This is the operating principle of
some 500 m illions during the whole the computer, and such projects al
life. ready exist. They w ill be capable of
Then why not make use of nature’s processing incoming data en masse.
experience? The computer element w ill perceive
Isn ’t the idea of building a compu not a line but a whole picture, nay,
OPT IC A L DE V I C E
M ir ro r
LASER C IR C U IT S
MICROMINIATURIZATION 203
ten thousand pictures at once, each The rate at which such machines
of which will contain 1010 bits of depart from machines that we have
information. In these computers you been accustomed to and become so
w ill be looking in vain for channels m ething defying imagination is quite
transmitting light and electronic sig terrific.
nals. Where does the road of microminia
They are strange, almost bodyless turization take us? Is there a lim it to
creatures. computer advancement?
Their principle of operation, to a The history of the generations of
first approximation, resembles that machines gives a negative answer to
of the epidiascope, which instantly this question. And what about the
displays pictures on a screen and over lim itations set by the laws of nature
lays them one on top of the other. to name, for one, the constant veloci
The computer’s “memory”, if it is ty of light?
based on “pictures”, w ill be able to There is no getting away from the
store a huge library of 500 thousand fact that the rate of information trans
volumes. fer is lim ited by the velocity of light.
The selection of information in Therefore, future optical machines
these computers w ill not be based on must be designed so that light in
the address principle when to get to them would have to travel minimum
the desired cell it is necessary to distances. It may be conjectured that
find the appropriate “street” and they w ill be spherical in shape, since
“house”, but on the principle of asso out of all bodies with an equal vol
ciations. ume the sphere has the minimum sur
Everything we memorize is inter face area.
connected, we memorize groups of Going over from our usual world
information, not separate bits. into the world of the atom we come
That’s how our memory—human face to face with new laws, new con
memory—works. This, too, w ill be the ventions. In our world the capacity
working principle of the new compu of the computer is lim ited by the
ters which have been even before maximum density of information
their birth romantically christened packed into the computer memory. In
“picture logic” computers or “pictu the world of the atom this is of no
re arithm etic” computers. importance since one cubic centime
We have just been introduced to tre of an absolutely condensed nuc
several generations of computers. Ha lear matter weighs 114 m illion tons.
ve you noticed how rapidly their ca What an enormous amount of infor
pabilities increase, and how at the mation could be packed into matter
same time their dimensions decrease having such density! And what about
at no less a rate? living matter? Just think about it,
the material basis of genetic informa sions of computer elements. This obs
tion of the whole population of the tacle, as we have just seen, may dis
world—some three billion people— appear in the future.
could be compressed into the volume Nature’s ingenuity, its ability to
of a rain drop. pack information sparingly, point to
It would be appropriate to remark the shortest road for the computer de
at this juncture that the nerve elements signer to take.
of the human brain—individually— Today it is too early to try to de
act quite slowly. The duration of the fine the place future machines w ill
operating cycle is almost a second. occupy in the life of man. What w ill
Slow speed is made up for by an enor be the job of intellectual automata
mous redundancy, since the elements that memorize and think quicker
are great in number. According to than the human brain? There is at
theoretical calculations an element present no answer to this question,
working at a rate of one m illionth of and not because it lies in the distant
a second does the work of a thousand future—it is not improbable that such
elements each working at a rate of machines w ill be our contemporaries.
one thousandth of a second. Natural The point is, scientists at present
ly the question arises, won’t the de have no clear picture of the future rela
signers turn back and start trading tions of the creator and its m arvel
speed for quantity? Before there was lous creation.
an obstacle in this w ay—the dimen
205
MODELLING
A model for the reproduction of the physical processes—an artificial power system.
rents in a circuit with active volta forces and deformations in some struc
ges.” ture.
The specialist in mechanics is cer The specialist in planning w ill sta
tain that the equations describe the te quite Authoritatively that these
equilibrium of forces in a system of equations serve to calculate the work
levers or springs. ing time of machine tools.
The building engineer w ill inform Five quite different answers. Which
us that these are equations relating is the right one? You shouldn’t won-
208 CYBERNETICS A TO Z
The model can assume a somewhat unfamiliar form as well—the form of mathematical for
mulae.
der—all are right. Yes, a single sys common between the calculation of
tem of linear algebraic equations can motion of celestial bodies and the roll
describe the state of equilibrium in of a ship. Yet, if only the formula
an electric circuit, as well as in a and the equations without words are
system of levers, or in a structure. A ll written, it is impossible to discern
depends on the meaning of the cons which of the two problems is being
tant coefficients a, b, c and of the solved: the equations are the same in
symbols of the unknowns x 1 and x 2. both cases.”
It is appropriate at this juncture to The wonderful mathematical si
recollect the words of the famous m ilarity of diverse phenomena pre
Russian academician A. Krylov: sents tremendous opportunities.
“To think of it, what could be in W hat’s the use of a scrupulously
MODELLING 209
Let us now find out how the electric modelling apparatus, this “mathematical
mirror” reflecting the regularities of the model, works.
L et’s start with a simple example. We intend to study the stresses in a steel
bridge truss loaded by a crane. What do we start with? First of all le t’s find the
geometric image of the truss and its electric model.
14-616
210 CYBERNETICS A TO Z
This is done with the aid of unfamiliar “electric cubes” which specialists call
resistor boxes. These “cubes” can be arranged to outline all sorts of figures: the
body of a dam, the wall of a channel, the blade of a turbine, the wing of a plane,
a rail, etc.
Each of these “electric cubes” consists of several coils of wire and condensors,
connected to a single terminal. The required circuit is drawn along the terminals
inside the “cube”—the so-called junctions. They are connected to current sources,
and this initiates various phenomena in the “cubes”. Electricity acting on each
“cube” through its junction plays the part of water, heat source or mechanical
force—i.e. of the active environment of the real prototype being studied.
But how should these “cubes” be assembled into figures corresponding to dams,
parts of turbines, etc.? These structures or machine parts should, too, be subdi
vided into “cubes”, so that a definite number of the “cubes” would correspond
to a definite dimension of the object being studied. This can be done easily with
the aid of the object’s prints.
L et’s return now to the calculations of the truss. Using the drawing of the truss
we arrange the “electric cubes” to form a geometric likeness of it and make the
measurements. There is no need of complicated switching and connecting of the
ends of one “cube” with those of another. The mass of “electric cubes” is so arran
ged that any figure can be immediately “cut out” by sim ply outlining the object
with a string on the “electric cube” set.
A universal electric model possesses the additional faculty of probing deep into
the element. Is there any way to learn what happens 10 cm inside the truss? L et’s
make a “hole” in the electric model. We can make any “holes” we like, marking
them appropriately on the print. After that we have only to disconnect a certain
number of “cubes” in the corresponding place of the model. After the required
hole has been made, any measurements may be made inside it, and they w ill give
exact answers as to what takes place inside the beam. To probe to any depth into
the “wound” inflicted on the model, one has only to connect a wire to this place.
This is done automatically by pressing an appropriate button on the “electric
cube” set.
The stresses in the beam are studied on an electric model called the electro-in
tegrator. It solves—integrates—with the aid of electricity complicated differen
tial equations sensitive to the smallest changes taking place in the shortest inter
vals of time.
The integrator is one of the existing types of continuous-action electric model
ling machines. Another name for them is analogue computers. Literally day to
day their fam ily is being expanded and modernized.
You have not failed to notice that the principle of operation of the analogue
computers is quite different from that of the digital computer. One specialized
book on modelling contains a very vivid and clear example to this effect,
A tailor who measures a m an’s figure in certain places makes use of numerical
methods.
The shoe-maker, on the other hand, who outlines the shape of a m an’s foot
on paper, makes use of the analogue principle, since his measurements are
continuous. This, too, is the principle of operation of the analogue computer:
MODELLING 21»
The analogy existing between thermal and electric processes enables the distribution of heat in
bodies even of the most complex shape to be modelled.
the waves. The forces acting on the plane change continuously, too. These changes
take place every moment of tim e, and they are not repeated.
On an electric-modelling or an electronic set fast processes can be repeated
any number of times. The mathematical model helps man to achieve mastery over
tim e. W ith the aid of “electric cubes” it enables dynamic variable processes ta
king place in reality to be slowed down or accelerated.
The scientist has only to reduce the m odelling rate to place himself in the po
sition of a man watching a slow m otion picture—he is able to learn things he
would not be able to without the artificial slowing down.
By increasing the modelling rate it is possible to hasten a process which in rea
lity proceeds at a sn ail’s pace and takes years to complete.
Looking into the “mathematical mirror” one sees and quite quickly, too, what
is going to happen to a dam, a lock, an artificial lake several years after they have
been built.
This is the reason why the unquestionable advantages of mathematical model
ling are adopted without reservation by scientists of various specialities.
Physicists occasionally find for the The journey from the test tube to an
models quite fantastic, from the lay industrial installation often takes from
m an’s point of view, fields of applica ten to twelve years. Gan mathemati
tion. Just to cite physico-mathemati- cal m odelling be applied here? Yes,
cal models of the plasma—of the ob and it is already being applied. For
ject the direct study of which, as instance, at the Novosibirsk chemical
physicists themselves agree, is made plant factory tests have been com
very difficult by its peculiar nature pleted of an installation which man
and by the com plexity and great cost aged to “skip” all intermediate deve
of experimental installations. lopment stages and arrived at the
A veritable hamlet in space enve factory direct from the laboratory.
lops the earth—75 space stations. This The “m athematical mirror” disclosed
is the place of destination of rockets its true image. Mathematical models
fired from the Earth. Regular traffic are also used to study the properties
by spaceships is maintained between of new catalysts, as well as in some
the stations. They bring in foodstuffs, other chemical experiments.
equipment and instruments, specia In collaboration with biologists,
lists for servicing and repair work. chemists achieve promising results in
These results of studies carried out by m odelling such exacting and unwieldy
American scientists using the method substances as enzymes, these marvel
of mathematical modelling were ex lous catalysts of life.
pressed in concise formulae. T hat’s You can learn about models in
m odelling as applied to astronautics. biology and medicine from a chapter
And what about chemistry? Here the of this book called “Cybernetics in
usual practice is for a process after B iology”.
leaving the laboratory to be subjected Something remains to be said about
to a protracted, m ultistage and ar one more aspect of modelling, about
duous testing and development work. its role in experiment. The m odelling
MODELLING 213
experiment differs from the usual one object widens the scope of the experi
in that the experimenter experiments ment rendering thereby invaluable
not with the object itself but with a service to the experimenter.
model of it. The model “intrudes” The story about where and when
into the experiment and draws at analogies between different processes
tention to itself. This is a very im and properties with an identical “ma
portant property of modelling, for them atical im age” are used can be
models can be experimented with continued indefinitely. Or it can be
even when the objects themselves for cut short with the words that have
some reason or other are beyond the been born several hundred years ago.
experimenters’ control—as the case “And, above all, I value the Ana
may be, they may be too far away logies, my faithful instructors. They
(the stars), or very short-lived (ele are in possession of all the secrets
mentary particles), or too great (lar of Nature and should therefore be
ge industrial complexes). In these last to be ignored.”
cases the model acting as an interme These are the words of the great
diary in assuming the role of the German mathematician J. Kepler.
214 MUSIC, ELECTRONIC
brush that proved most suitable for to be the source serious and very
computer music, though with one im promising work.
portant improvement: the erasal of An example of such work is the
unwanted blot-notes. “Urals M elodies” mentioned before.
To operate on this principle, a They are interesting as an attempt to
machine’s memory must be provided construct not only chords, as is usual
with a store of sounds from which it ly the case, but also that very impor
can extract combinations according tant component of music, the melody.
to m athematical rules. As you may The author of the new method is
have guessed, the “sounds” in the Soviet scientist R. Zaripov. The essen
machine are represented in terms of ce of his method is contained in a pa
numbers. And, of course, a carefully per entitled “On Algorithmic Descrip
compiled programme is needed. tion of the Process of Music Composi
But what is to be gained by this? tio n ” printed in such a serious and
Who needs this kind of prim itive mu highly esteemed publication as the
sical compilations? The answer is no Proceedings of the Academy of Scien
one—at least no one needs them as ces of the U SSR.
music. However, they have proved
000037
000006
000210
202413
003212
004111
004113
205115
005113
006250
007006
010210
212413
013212
014111
014113
215115
015113
016250
017006
022207
Several processes in the translation of notes and rules of music into machine language: a method
of encoding pitches; a portion of a programme written down in number'code; the same program
me on punched tapes; the result of the machine’s work written down on paper tape.
Mathematicians studying music found that popular songs are drawn from 35 to
60 notes. A statistical analysis of large numbers of songs revealed the following
typical song structure: there is the first part, which we shall call A , covering eight
bars and consisting of 18 to 25 notes. This part is repeated once and then followed
by a part B , also covering eight bars but consisting of 17-35 notes, after which
part A is repeated once more.
Other interesting rules were discovered. If five consecutive notes follow on'an
ascending scale, the sixth w ill always drop, and vice versa. Also, the first note
of part A w ill usually never be the second, fourth or fifth minor note of a scale.
Songs obey such long-established rules of composition as Mozart’s rule which lays
down that an interval between two neighbouring notes must never exceed six to
nes.
218 CYBERNETICS A TO Z
Commence work
, j 9,
89C
R-/89C'J‘
,U709./i
^789/
78<
785
.785
■785
)785
378
57895
o45678901^
.234567890123*
12345678901234.
•234567890123^
734567890123'
234567890123
■•45678901?''
456789011
456789011
. ->4567890125^
' k '2345678901234:
(23456789012345,
.1234567890123451
.1234567890123451
’23456789012345
End operation
You may, on occasion, have listened to an unfamiliar tune and been able to pre
dict the following note before it was played. This happens most frequently with
lyrical songs. Specialists say that, in such cases, each subsequent note carries less
information than, for instance, in music pieces by Prokofiev or Shostakovich which
abound in unexpected inflections. Thus, the amount of information per note is
a parameter, a quantity which can be used to judge the music.
The rules of composition deducted from analyses of musical works are used in
writing music with the help of electronic computers. In addition, the “U ral”
computer was provided with a special generator of random numbers, its purpose
being to present random notes in a digital code. Each note was then examined
under the rules of musical composition, and only when a note was found to meet
the requirements was it entered into the score. If the screening reveals the note
MUSIC, ELECTRONIC 219
N ff
-m-+
• ►* - ■------
m
f
m -
220 CYBERNETICS A TO Z
note required an average of 800 m achine operations, and a whole piece as m any
as 30000.
A m elody synthesis programme for a “B3CM-2” com puter occupies several
thousands of its “m em ory” cells.
W hen the com puter, equipped w ith its program me, finishes its com pilation of
the tune its autom atic printer feeds it out in code on paper tape. The record is de
ciphered and then transcripted in to conventional m usical script.
MUSIC, ELECTRONIC 221
p
p 2 ir. rf. i
j- ^
i —
j. j; _j_. ^ 0 J J liJ
f ' — — s -—_
M~ f ~(i# _
a
i pi'-
222 CYBERNETICS A TO Z
Now th a t you have seen how com the ra th e r laborious task of tran scri
puters compose music and have exa bing a score from one key to another,
m ined several pieces, it is tim e to and i t is sure to m ake no m istakes in
explain why scientists devote so m uch the process.
tim e and effort to produce w hat is A com puter can be usefully em
usually extrem ely sim ple melodies. ployed in deciphering m usical m anus
The thing is th a t com puters deal cripts em ploying the m ethods of quan
w ith symbols. The elem ents of m usic tita tiv e analysis. In a sim ilar way it
are also sym bols. The num ber of sym m ay transcribe volum inous tape re
bols in music is rela tiv e ly sm all, cordings of m usical folklore.
which m akes i t convenient for compu “E lectronic m usicians” can be en
ter experim ents. A com puter makes tru sted w ith such tasks as searching
it possible to trace step by step, note for new tim bers, the im portance of
by note how sim ple elem ents combine which can h ardly be overestim ated for
to produce a tune. Com puter m usic instru m en tatio n , as well as in the
offers an o p p ortunity to study the arrangem ent of sym phonic m usic for
very nature of m usic, investigating in d iv id u al instrum ents.
m usical forms, chords, scales and se I t is hard to say w hether m achines
quences. A com puter is a fine tool for w ill ever be capable of producing
analysing the creative process. anything like genuine works of art,
By teaching com puters to w rite m u and works of creative value. O bvi
sic scholars hope to penetrate the do ously, a m achine producing som ething
m ain of a rt and investigate it through like a work of a rt has no creative
a new, cybernetic approach. urge: i t a ll begins and ends w ith the
Hence the very specific role assig first and la st b ar assigned by the prog
ned to com puters composing music: ram m e. A m achine m ay grind out hun
they are not intended for creative dreds and thousands of tunes, b u t it
work b u t as m an ’s helpers. A com pu w ill never say w ith pride, “This is
te r can, for example, easily cope w ith my best w ork!”
An A d vanced Programme
N Two States
The trigger circuit. When the right tube'is open and the left closed, the trigger registers 0. And
vice versa, when the left is open and the right is closed, it registers 1 .
consists of two electron tubes assem ly closes the open tube and opens the
bled in one bulb. They are electrically closed one. And in strict accordance
connected so that when one conducts with this pulse the trigger immedia
current, the other blocks it. (One is tely changes its state from 1 to 0
open, the other is closed.) One of and vice versa.
these stable states was taken to mean It remains in each state until a
1, the other 0. new pulse reaches it. So “overturning”
Each new electric pulse applied to from one state into the other the trig
the input of the trigger simultaneous ger enables pulses to be registered.
The recording of numbers in a computer by the series and the parallel methods. ^
T H E NUM BERS IN A COMPUTER
ARE R E C O R D E D BY
F o u r s o l d i e r s in
A s e t of p u ls e s
mm a co lu m n
T H E S E R IE S M ETH O D
"o r "
¥ F our
Hi
¥ s o ld ie rs
¥
P u ls e s
£
¥ £
THE PARALLEL METHOD
226 CYBERNETICS A TO Z
It would appear at first sight that Seta to name a prize he would like
to record large numbers one should to receive for his invention of an
have an enormous amount of trigger excellent game—the chess.
cells. At this juncture it is appro Seta laid the chess board before the
priate to recall the Indian legend of Shah and asked for one wheat grain for
the Shah Sheram who offered the sage the first square, two for the second.
NUMBERS IN A COMPUTER 227
four for the third and so on, for all the Sun. And this gigantic number
the 64 squares—that is, for each next the inventor managed to get with
square twice as many as for the pre the aid of only 64 squares.
ceding one. In the modern computer, too, the
The demand of the sage appeared circuit of only 64 triggers is capable
quite modest at first. S till, Seta fai of counting the astronomical “chess”
led to get his prize. number 264.
The mathematicians of the Shah Triggers have been in use as count
calculated that the number of grains ing devices for a long time. Formerly
for the last square is so great that it they were used m ainly to count ato
defies imagination: eighteen quintil- mic particles. Later their use was
lion four hundred forty-six quadril extended to computers.
lion seven hundred and forty-four tril In the Soviet Union the first trig
lion seventy three billion seven hun ger was made in 1918 by the promi
dred and nine m illion five hundred nent scientist, radio engineer
and fifty-one thousand six hundred M. Bonch-Bruevich. He started re
and twelve. search on electron tubes way back
The grain would have filled two in 1916 and was the first to organize
barns stretching from the Earth to the production of them.
228
1 II III
A 6 2 7
B 3 4 8
C 1 5 5
D 2 9 3
Cargo-handling
| capacity of _
each landing stage®1 0 11 11
stages I, II and III. The sand is needed for four construction projects, A, B, C, D.
To solve this problem we w ill have to write several tables. In the first one the
right-hand column gives the daily requirement of sand—30 tons. The bottom row
indicates the cargo-handling capacity of each landing stage. The delivery quotas
230 CYBERNETICS A TO Z
are balanced w ith the requirem ents. The numbers in the squares show the dis
tance, in kilom etres, between the landing stages and the construction sites.
Our problem is to d istribute sand deliveries among the landing stages so as to
keep the autohaulage, in ton-kilom etres, down to the least possible m inim um .
The garage traffic controller has distributed deliveries as shown in the second
table (the numbers in the squares indicate the am ount of sand delivered from the
respective landing stages to the respective construction sites). A ll construction
sites receive the necessary q u an tity of sand. B ut w hat about haulage figures? This
is easily calculated: 1 X 1 + 7 x 2 + 3 x 4 + 8 x 5 + 5 x 7 + 6 x 8 = 1 5 0 ton-kilom etres.
However, a better d istribution is given in the th ird table. The num ber of ton-
kilom etres is reduced by one-half: 8 x 1 + 1 x 2 + 9 x 4 + 1 x 5 + 4 x 7 + 7 x 3 = 100.
B ut the best, optim um v arian t offers a 70 per cent cut of haulage as compared
w ith the first. It is presented in the fourth table.
You see how much trouble is involved in the solution of a “sim ple” economic
problem , a concrete problem w ith few v arian ts and sm all numbers involved.
Today it is realized by all th a t it is hard to work out even a fam ily budget for
a m onth or two in advance to an accuracy of one rouble. The economic estim ates
for an industrial enterprise are a hundredfold more difficult. And w hat about cal
culations for a group of enterprises, a whole industry, or on a nation-w ide scale?
After this try and im agine the trem endous scale of economic calculations for a
national economy of the size we have in the Soviet Union!
See w hat savings have resulted from producing an optim um plan for a single
industry on a country-wide scale. The optim um plan for the geographical d istri
bution of cement m ills throughout our country enabled a reduction of average
haulage distance from 565 to 305 kilom etres. This yielded a saving of 140 m illion
roubles per year. Furtherm ore, it made possible the cancellation of construction
plans of several m ills in unsuitable locations. This yielded a saving of another
192 m illion roubles.
Such economic problems are solved by means of so-called linear program m ing,
which was elaborated by Soviet scientist L. Kantorovich.
We shall try to explain the general The n atu ral question is:|w lia t diet
idea of linear program m ing w ith the can fully m eet the organism ’s physio
help of a sim ple example. logical requirem ents a t the lowest
To function norm ally, the hum an cost? In other words, w hat is needed
body m ust receive a certain daily is the cheapest and at the same tim e
ration of nutrients: fats, proteins, car m edically the m ost wholesome food.
bohydrates, vitam ins, m ineral salts, Thus, the task is to choose from
etc. On the basis of comprehensive 50 groups of food products a d iet
research, m edical science has estab containing the 15 n u trien ts required
lished the necessary ratio n for 15 by the organism . The basic criteria
types of nu trien ts. The substances for an optim um selection is the cost
needed by the body are contained in of the food. I t should be as low as
various q u an tities in different foods: possible. The task, in m athem atical
bread, m ilk, fish, vegetables, fru it, parlance, involves the solution of a
cereals—in all some 40 or 50 groups. set of 15 equations w ith 50 unknow ns.
OPTIMAL CONTROL 231
V) A
0 aa 10 90 C/D A 10 90
0) _
O
o B 40 70 10 W a re h o u se III 2 B 50 70
03 O
LL
C 40 80 C 40 80
W a re h o u se V F a c to ry C
Warehou - V o lu m e o f
se s i 11 III IV V o u t p u t of
fa cto rie s
CO A 12 7 13 8 6 100
CD
2 B 4 5 8 6 10 120
C
LL
C 7 14 15 12 16 120
Warehouse
c a p a c i t y 4U 50 70 90 90 340
OPTIMAL CONTROL 233
variants of the plan in the quest for variants but on a purposeful quest
the optimum one. each step of which brings us nearer
In our problem a clearly defined to the goal. When it is achieved, and
target function ensures the choice of the computer feeds out the optimum
the optimum diet at the lowest cost. variant we can declare with full con
The solution yields such positive va fidence that, given the stated conditi
lues of the variables x x and x 2 which ons of the problem, nothing better
satisfy the conditions of the problem can be devised—this is guaranteed by
(as written down in the system of the methods of linear programming.
equations) and with which the target W ith the current high rates of
function has the minimum of all pos economic development the compo
sible values. The methods of linear nents of the national economy inter
programming are designed to solve act faster and faster, and economic
just these kinds of problems. indices become ever more dynamic.
» A computer is fed the conditions In such circumstances planning must
of the problem in the form of a sys be carried out on a continuous basis,
tem of equations and inequalities and and all sections of the economy must
the target function formula. The ma be managed with the utmost efficien
chine selects some initial variant and cy. The smallest delays in planning
analyses it. The analysis imm ediately and control can result in losses of no
reveals whether this practically ran less magnitude than those due to in
dom variant approaches the optimum accuracies in planning.
or not—something, obviously, high U ntil recently in our country more
ly improbable. If the variant is not than two m illion persons were emplo
the optimum one the analysis indica yed in administration and manage
tes how the next one should be chosen ment. It may come as a surprise to
so that it is better than the preceding the uninitiated that at the Gorky
one and the target function is smaller. Motor Works alone a veritable army—
The new variant is also analysed for more than five thousand people—are
optimum conditions and, if the ans engaged in ensuring the factory’s raw
wer is negative, another, still better, materials supplies and marketing.
variant is selected. The quest pro The people responsible for economic
ceeds step by step until the analysis management handle billions of pa
reveals that the optimum solution, that pers annually. Nowadays simple desk
is, a variant w ith the smallest pos calculators and adding machines are
sible target function has been found. inadequate to cope with the flood of
As you see, the method is based not information, it is virtually impossible
on a random try-out of all possible to calculate any great number of
■4 Study of transport operations by linear programming. The purpose is to obtain the lowest haulage
costs taking into account the output of each factory, the storage capacity of each warehouse
and haulage costs of a unit commodity from each factory to each warehouse. The tables at upper
left and right present two optimum solutions. The numbers in the squares denote the quanti
ties of different goods to be shipped to various warehouses to obtain the lowest overhead. The
table at the bottom gives the cost of transporting cargo from factory to warehouse. The diag
ram in the centre presents the haulage routes with the costs presented on the vans. The thick
arrows give the most advantageous routes.
234 CYBERNETICS A TO Z
A p p ro x im a te p ic tu re
o f a c tu a l
s itu a tio n
A d ju s tm e n t o f th e
m odel o f p ro d u c
tio n to s u it a c tu a l
s itu a tio n
A p p ro x im a te c o rre la
t io n o f p la n and
a c tu a l p ro d u c tio n
plan variants and ensure optimal con automation of planning and economic
trol of the economy. calculations w ill become a reality.
Workers of the Cybernetics Insti Mathematical methods could be cal
tute of the Ukrainian Academy of led the golden key that opens the
Sciences have estimated that the vo doors into the mystery kingdom of
lume of information involved in eco economics.
nomic planning increases in propor Of course, solving economic prob
tion to the square of the volume of lems is an extremely difficult thing.
production. This means that in 20 It has been shown that the higher the
years more than half the population level of a nation’s productive forces
of the Soviet Union would be engaged the more complex is the task of run
in management and administration. ning its economy. The scope and com
The economists are justified in plexity of calculations increase with
w aiting for the day when complete the expanding volume of production,
OPTIMAL CONTROL 235
acceleration of growth rates, appea state of affairs. Then come the peri
rance of new industries, expansion of pheral units for data processing and
economic contacts. This is the reason storage. These are electronic machines
for the extensive scale of work going connected by communications chan
on to use electronic computers to plan nels with the central “computer plants”
the economy of regions, republics and of economic areas, industries or other
the Soviet Union as a whole. amalgamations.
In Byelorussia, for example, machi Further on, at the top, is located a
nes were used to calculate and intro system for receiving, handling and
duce the most profitable plan of tim storing information designed to pro
ber haulage. In Leningrad plans were vide a comprehensive picture of the
drawn up of river and marine ship state of affairs in each specific indus
ping for all of the main river and try and the economy as a whole.
marine basins. The Computer Centre Experts consider that the creation
of the State Planning Commission of of a national network for gathering,
the USSR calculated the best distri storing, processing and transmitting
bution of suppliers and consumers of economic information is a priority
sheet metal and plate. In Turkmenia task the importance of which can
machines have been for several years hardly be overrated.
carrying out a substantial portion of A planned socialist society should
the economic calculations. strive to attain the best possible,
Many such examples could be cited. optimum results in each separate in
It is not for nothing that V. Glush- dustry and in the national economy as
kov, an eminent specialist in cyber a whole. It was this that V. I. Lenin
netics, claims that soon “automatic meant when he said that only that
systems w ill become as essential a construction can deserve to be called
tool of scientific research in the hands socialist which is carried out accor
of economists as atom smashers are ding to a comprehensive general plan
for physicists or electron microscopes aimed at the balanced utilization of
are for biologists”. all economic and financial resources.
Electronic computers not only help V. I. Lenin repeatedly drew atten
to accelerate calculations, they also tion to the importance of scientific
provide for a ten- and hundredfold organization of administrative work,
reduction in the time needed to draw without which it is impossible to run
up plans. Electronic machines used the economy correctly. He wrote that
for economic operations ensure centra people working in management and
lized industrial management. administration must pass the test of
Economic management entails a knowledge of the fundamentals of
kind of computer hierarchy. The “gro theory of our state apparatus, of the
und floor”, so to speak, houses the fundamentals of the science of mana
“lower echelons” charged with the gement.
collection, storing and processing of Nowadays mathematical methods
information. They allow for planning are the basis of scientific manage
the work of factory shops, depart ment without which economic plan
ments and enterprises as a whole, pro ning of industry, science, technology
viding a day-to-day picture of the and distribution of resources, man-
236 CYBERNETICS A TO Z
power reserves, and direct, concrete and for separate industries, how to
factory management and control are make the most rational use of mate
sim ply unthinkable. rial and manpower resources, on the
The best variant of a plan for the questions of economic reform.
development of the Kuznetsk basin The new method makes it possible
coalfields in the years 1962-1970 had to forecast scientific and technological
to be worked out. It was necessary to progress and its effect on the economy,
take into account the number of col assess natural resources, determine the
lieries, the grades of coal, labour ex optimum size of state reserves and
penditure and a thousand and one optim ize the structure of planning
other considerations. There are some bodies themselves.
100 pits and cuts in the Kuzbas for In conclusion, to show the whole
which some 200 development variants importance of application of cyberne
had been provided. By using methods tics in the economy, we should like
of mathematical planning the optimum to refer to a major component of so
variant for solving this complex pro cialist planning—to the work of draw
blem was found. It turned out that ing up the national economic balance.
the development of the Kuznetsk ba This balance is a system of economic
sin coalfields could be carried out indices characterizing the principal
with a capital investment 200 m illion correlations, proportions and rates of
roubles less than envisaged by the production, the sources and reserves
original “hand-made” plan. The com for expanding the country’s national
puter-made plan proved 43 per cent wealth and popular income.
more economical than the man-made Mathematical methods make it pos
one! Today economists increasingly re sible to establish all the proportions
ject “real-life” experiments and cost and connections in the national eco
ly “trial-and-error” quests. They pre nomy with great accuracy and to
fer the better tool of mathematical study its efficiency. In other words,
modelling. Such modelling was em we can now determine much faster
ployed in the stupendous task of de and more accurately whether our na
veloping a scientific model of the tional wealth is being used correctly.
national economic development plan And that wealth is great indeed. For
of the USSR for the years 1971-1975. example, the gross national product
Practical recommendations were ma of the Soviet Union in 1968 was esti
de on how to improve the development mated at 500 000 m illion roubles!
patterns for the economy as a whole
237
mes is a very difficult job requiring between any two of the com puter’s
high qualification. T his is because elem ents. The com m unications chan
the program m ist has first to visualize nel m ust dispose of some sort of a
and then to realize in th e com puter “sw itchm an” who eith er tran sm its
all inform ation transm ission routes inform ation to a channel or blocks
necessary for the execution of a defi the way.
n ite sequence of operations. The elem ents playing the p a rt of
The program m e has been com piled. the electronic “sw itchm an” in the
The routes for the solution of the prob com puter are called valves. The pur
lem have been la id in the com puter. pose of the program m e is to provide
Now the com puter sets to work. instructions for the opening of a cer
Since we speak of problem solution ta in group of valves while keeping the
routes, th is means th a t some commu others closed.
nications, some channel is established The work of the computer is gover-
An example of the realization of a programme in a computer.
PROGRAMMING 239
L e t's try to com pile a com puter code each operation w ith a set [of
program m e for calcu latin g the grea b in ary sym bols. Besides, w e’ll desig
te st common divisor of two num bers nate every operation briefly by letters.
a and b according to the well-known This w ill fa c ilita te the com pilation of
algorithm of E uclid. For th is purpose the program me.
we should have a lis t of operations. H ere’s a short tab le of operations
Suppose we have them w ritte n down. which w ill help us to com pile the
L e t’s arrange them in to a tab le and programme:
240 CYBERNETICS A TO Z
B rief d e s ig B inary
O peration n a tio n cod e
To profit from the tab le one sent to the cell, the num ber which
should keep in m ind several im p o rtan t form erly occupied it is “erased” , and
conditions. the newly sent num ber is in stalled
The first—when a num ber from the in its place. The num ber which has
register cell is sent to the adder it is been sent from the adder rem ains
retained in th e register as w ell. there unchanged. M oreover, w e’ll have
The second—when, on the other to rem em ber th a t, if the num bers a
hand, a num ber from the adder is and b are n ot equal, w e’ll have to
C eil c o n te n ts
C ell num bers d esig n a ted b y le tte r s In b in a ry code
01 Ss—20 00110000010100
02 Sb—21 00100000010101
03 Is—05 01100000000101
04 St 10000000000000
05 Ss—20 00110000010100
06 Sb—21 00100000000101
Cells 07 In—10 01110000001010
containing 08 Sn—21 01000000010100
instructions 09 Rc—01 01010000000001
10 Ss—21 00110000010101
11 Sb—20 00100000010100
12 Sn—21 01000000010101
13 Rc—01 01010000000001
14
15
16
17
18
19
Cells containing 20 number a
numbers 21 number b
PROGRAMMING 241
find out which of them is the greater p utor p rin ts the num ber, and the
and which the sm aller. problem is solved.
Man can do it easily ju st by taking If, on the other hand, the rem ain
a look a t the num bers. The com puter, der is not zero, i.e. the num bers are
on the other hand, alw ays has to com not equal, the control device in com
pare by orders. Therefore, the easiest pliance w ith the conditional in stru c
way for the com puter to find out tio n w ill tu rn to the next instruction
which of two num bers is the greater in cell 05 and la te r to cell 06. In com
is to perform subtraction. pliance w ith the instructions co n tai
The sign of the rem ainder w ill tell ned in these cells the num ber a from
it which is the greater and which is cell 20 w ill again be sent to the adder
the sm aller. w hich has been preset to zero. There
L e t’s s ta rt compiling the programme. the num ber b from cell 21 w ill be
W e’ll place the instructions in the subtracted from it.
working “m em ory” cells No. 1 W hen a rem ainder accum ulates in
through No. 13. The num ber a w ill the adder, the conditional instruction
be placed in cell No. 20, the num ber from cell 07 is m ade to operate.
b in cell No. 21. If the rem ainder is negative, i.e. the
The addresses of the register cells second num ber is less th an the first,
w ill be w ritte n down in the decim al they should, according to the algo
system . rith m , be m ade to change places.
L e t’s see how the com puter works The conditional tra n sitio n in stru c
using the tab le as a guide. The com tio n hands the control over to cell 10.
puter is sw itched on, and an in stru c The la tte r orders the second num ber b
tion is sent to the control device from from cell 21 to be sent to the adder.
the register cell 01. N ext the num ber from cell 20, i.e.
This in stru ctio n w ill be carried out the num ber a, is subtracted from it.
during the next step. The adder pre The rem ainder w hich is positive is
set to zero w ill receive the num ber sent to cell 21 on the in stru ctio n of
from cell 20—the first num ber a. cell 12. E v e n tu ally the instru ctio n
This num ber w ill rem ain in the adder. of cell 13 again hands over control to
N ext an in stru ctio n from cell 02 is cell 01.
sent to the control device. The result The cycle w ill be repeated in com
is the sub tractio n of the num ber b pliance w ith the fifth requirem ent of
from cell 21 from the num ber a. the algorithm .
This, as we shall see below, satisfies The num ber a, in th is case the sub
the second requirem ent of the Euc trahend (because it is the sm aller of
lid ’s algorithm concerning the com the two num bers), is sent to cell 20.
parison of the two num bers. Now it The rem ainder which now occupies
is the tu rn of the conditional instru c cell 21 w ill be subtracted from it, etc.
tion from cell 03. The control device If, on the other hand, the rem ainder
checks the result of the foregoing ope obtained after instructions 05 and 06
ratio n . have been carried out is positive,
If th e rem ainder is zero, the num the instru ctio n 07 w ill be followed
bers are equal, and each of them is by the in stru ctio n from cell 08. The
the greatest common divisor. The com- positive rem ainder w ill be sent to
16—616
242 CYBERNETICS A TO Z
SEQUENCE
QF INSTRUCTIONS
The The
address executive
part part
0001 0 31
1000 0 37
0002 0 35
1000 0 37
0003 0 36
An example of programming. The sequence of instructions is shown on the left below. The trans
fer of the number to the summator is designated by 31, 37 is the multiplication of numbers,
35—addition, etc.
cell 20, and com puter control w ill Thus, the whole program m e is com
be returned to cell 01 from cell 09. piled from 13 instructions. I t enables
T his, as in the case above, w ill be the the g reatest common divisor of any
beginning of a new cycle. pair of num bers to be found. The
D uring th is cycle the place of the num ber of in stru ctio n s is alw ays the
two num bers is, in accordance w ith same. B ut th e to ta l num ber of repe
th e E u c lid ’s algorithm , tak en by the titio n cycles w ill be different depen
subtrahend and th e rem ainder. dent on the num bers a and 6.
In th is w ay the process of searching I t ’s easy to verify th a t if a = 2 1 ,
for th e g reatest common divisor w ill 6 = 1 4 , the th ird cycle w ill yield the
continue cycle after cycle u n til the greatest common divisor equal to 7.
num bers in cells 20 and 21 are equal. In case of m u lti-d ig it num bers, on
PROGRAMMING 243
the other hand, th e num ber of cycles gram m es for the com puters produced
m ay be tens or even hundreds. by the com pany.
C om pilation of program m es for the The program m e lib ra ry com piled for
solution of th e more com plex prob th e “1900’’-type com puter alone com
lems requires great experience and prises over 3.5 m illion instructions.
su b sta n tia l effort. Program m e com pilation has become
I t should be noted th a t w ith o u t the a sort of an in d u stry for the m athe
program mes a ll electronic com puters, m atical provision of com puters. More
even those capable of m illions of ope m oney is spent on th is in d u stry th an
rations per second, are, as a prom inent on the production of the com puters
com puter expert A cadem ician T . M ar them selves. Thus, about one billion
chuk once rem arked, a t best, m erely dollars were spent on the procurem ent
item s of fu rn itu re of in stitu te and fac of program m es for the IBM-360
to ry offices. The entire line of calcu com puter.
lations in the com puter, th e so-called W ell then, w hat is th is m athem ati
processing of inform ation from in p u t cal provision? I t is a com plex—lite
to o u tp u t, is organized by the program ra lly a m u ltitu d e — of program mes
me. I t ’s the program m e th a t ensures assem bled in special libraries. They
the execution of a ll operations assig enable the com puter to operate effi
ned to th e com puter. c ie n tly , to carry out the solution of
The program m ists prepare whole problem s. They, in the words of the
series of stan d ard program m es for the experts, m ean the same to th e com pu
solution of ty p ical problem s. The ters as college education m eans to
greater th e program m e file provided people.
for the com puter the b e tte r i t is adap O ur days w itnessed the b irth of a
ted for its work, the easier is its con new profession—th a t of the m athem a
ta c t w ith th e user, the greater are tic a l provision system s engineer. So
th e fac ilitie s for its use and the grea m uch a tte n tio n is being paid to the
te r is its value. education of program m ists because
Nowadays com puter m anufacturers w ith o u t them th e ex p lo itatio n of
spare no efforts and no resources to electronic com puters a t present em
produce com puters equipped w ith a ployed everywhere is im possible.
com plete set of standard program m es. A qualified program m ist m ust re
For exam ple, a t the A m erican compu ceive a serious m athem atical educa
te r production com pany IBM some tio n . Those who th in k of becoming
one and a half thousand employees are program m ists should study m athe
perm anently employed com piling pro m atics thoroughly.
244 PSYCHOLOGY, ENGINEERING
he receives factual inform ation from the m achine, processes it into comm and in
form ation and transm its it to the m achine.
The m achine, too, has its functions which are no less clearly defined. The m a
chines on m a n ’s comm and perform technological operations.
True, the d istribution of functions between m an and m achine m ay differ g reat
ly from case to case.
W hat does it depend on? On the purpose of the system .
Here is a mechanized production system . Here m an has to perform all functions
of regulation and control.
Now w e’ll go over to a system of a higher ty p e —to the autom ated production
system . Here the m ajor p a rt of these functions is entrusted to the machines. M an’s
job is lim ited to planning operations as a whole. He makes im portant decisions,
controls the work of the system in general, taking over the control in case of devia
tions from the programme.
W h a t’s the way of achieving best, optim um results of the com bination of “m a
chine” and “hum an” properties in the “m an-m achine” system? W h a t’s the way
of a tta in in g in the “m an-m achine” system m axim um efficiency of the machine
and m inim um fatigue of man? The problem m ay be tackled from two opposite sides.
F irstly , the m achine should be b e tte r adapted to m an. Secondly, as it is being
done by engineering psychology, m an should be “ad ap ted ” to the m achine in the
best way possible.
W ith in the scope of engineering psychology come those links of the system whe
re m an finds him self directly connected to the m achine, where the transm ission
of inform ation takes place. Does th a t m ean th a t m an him self, in the first place,
serves as such a link? I t certainly does. He perceives the inform ation by his sen
ses. He processes inform ation and tran sm its it to the control panel. W hen control
ling the machine he translates instructions into a convenient code th a t can be ea
sily understood by the m achine.
Acceleration Acceleration
1 10 1000 to ibo
Temperature Temperature
The comparison between the allowable limits of acceleration, radiation and temperature for man
and automaton on a journey into outer space is not in favour of man.
per m inute) from the norm al as sm all them , are exam ined by psychologists.
as three per cent. S cientists assess in n ate q u alities of
The scientists a t the laboratory of every m an and his a b ility to develop
labour psychology of the Academy of them . This enables to predict the job
Pedagogical Sciences follow ing m eti th a t w ill be to the m an ’s lik in g , and
culous research carried out into the to avoid w asting tim e and resources
properties of technical hearing deve on teaching people jobs th a t they
loped express m ethods of train in g . w ill not regard as fit for them .
W orkers a t the Perm telephone Nowadays engineering psychology
p la n t, before being sent to shops, faces some very im p o rtan t problem s.
before jobs and duties are assigned to There is, on the one hand, the rapid
250 CYBERNETICS A TO Z
specialized m echanical, electrical, and term s to the m achine tool or the pro
optical controllers, fast, com pact, fle duction line, force them to operate
xible, ensuring the necessary regula in a s tric tly orthodox m anner.
tio n procedure. T here’s in add itio n a A t th is juncture it would be bene
v ast arm y of tracking system s to ficial to show how the q u a lity control
w atch over “law and o rd er” in pro is being accom plished and how m u lti
duction shops. They d ictate rigid farious th is process is.
The control process usually consists of two stages. The first is the perception
of inform ation about the state of the object, and the second is the separation from
the incoming inform ation of th a t p a rt of it which pertains to the q u a n tity being
controlled.
This is usually done in laconic language: “yes-no” , “too m uch-standard-too
l it tl e ” , “operational-nonoperational”.
Everyone knows how im portant accurate m easurem ents in q u a lity control are.
Modern technology is able to produce autom atic q u a lity controllers w ith a
memory. Such a controller w ill not only find errors, b u t w ill memorize them , make
an analysis of them , draw conclusions and tell the machine tool w hat to do to
avoid rejects.
Devices working in conjunction w ith autom atic flow-lines m ay serve as an
exam ple of fully autom ated q u a lity control.
There are even whole factories working on the “super-closed shop” principle
w ithout people and, thanks to autom atic control, producing high-quality goods.
This applies to hydro-electric stations and continuous process chemical plants.
Here the centralized control embraces everything: q u a lity of product, sta te of
equipm ent, q u a n tity and composition of sem i-products.
The q u a lity control (technological control) in engineering industry, on the
other hand, is decentralized. It is carried out at principal operation stages, in the
process of transportation and at special m ulti-operational control stations. There
are special machines for sorting and filtering out rejects and for control of finished
products.
Nowadays, w ith the advent of autom ated production control system s the qua
lity control is carried out by special devices coupled to electronic computers.
And w hat about the forms of control? They are numerous. Single param eter
control, as its nam e im plies, controls only one specific param eter w ith one gauge
of the sim plest type.
M ulti-param eter control, as d istinct from the single-param eter control, w at
ches over numerous param eters w ith the aid of m any w ell-coordinated instrum ents,
machines or even whole systems. This control is, as a rule, centralized.
M ethods of control are also subdivided into series, parallel and series-parallel.
In the first, inform ation is received, in tu rn , from several control sources. In the
second, inform ation is received continuously and sim ultaneously from m any sour
ces over m ultiple channels. The th ird , as m ay be easily guessed, combines both
form er m ethods.
T h ere’s also the logic control which entails the u tiliz a tio n of the result of mea-
QUALITY CONTROL 253
These are the operations which have to be performed in case of computer errors of various types.
surem ents and control for self-tuning, for assessment of variations in conditions
of some process.
Sporadic control is carried out as the need arises, from tim e to tim e, or in ans
wer to autom atic signals in case of rare b u t su b stan tial deviations.
A more complex type of control, the m ulti-stage control, abides by the principle
of hierarchy: the higher the control stage, the higher is the q u a lity of m achining
or the accuracy of a process. Such control involves the collection of inform ation
from all the branches of the factory.
I t ’s no easy job to effect q u ality I t m ust be said in addition th a t
control in in dustry. The above list since the tim e electronic com puters
was quite im pressive. And s till, it began to be used, the problem loomed
d id n ’t m ention local control, d ista n t of controlling the q u a lity of th eir
control, telem echanical control. work. Now it is necessary to discou-
254 CYBERNETICS A TO Z
rage the clever m achines from m aking controlling the work of electronic
s illy m istakes, to see th a t they do com puters. They are subdivided into
th e ir calculations correctly. This is the program m e control and circu itry
one of the principal problem s of de control m ethods. E ach of the la tte r is
signing d ig ital electronic com puters. subdivided, in its tu rn , in to proce
The problem is very com plicated for dures of control in the process of prob
even a dust particle on the m agnetic lem solution and of control outside
m em ory tape m ay lead to an error. the operational cycle.
T ry and find it afterw ards! The sim plest w ay of controlling the
U nhappily, existing m ethods of q u a lity of com puter operation, from
com puter control do n ot guarantee logic considerations, would be to re
absolute and continuous q u a lity of peat specific cycles or replace the
calculations. A com plete and detailed fau lty block. I t is also obvious th a t
control procedure is, as y et, lacking, it is easier to control th e inform ation
and the users of the u ltra -fast and transm ission circu its of th e com puter
ultra-accu rate electronic com puters are th a n the a rith m etical operations. The
s till liab le to suffer from th e ir fancies problem of detecting and excluding all
—there are instances when the com possible errors in the com puter con
p u ter hands out an erroneous result. tro l system m ay, however, tu rn out to
There are various procedures for be insolvable.
255
R In d ex of T ru st
R O B O T S
MYTHICAL ROBOTS
CLOCKWORK ROBOTS
ELECTRICAL ROBOTS
ERI C TUM"
RELIABILITY 257
nonrestorable
<§> 1
$o
restorable
nonrestorable
1 restorable
pg
FrrF'
4
ITT vM
The classification of products in the theory of reliability. Some products—the elements of ra
dio-equipment, instruments, resistors, machine parts, bearings—are not regenerated. Others—
electronic computers and control devices, cars, machine tools—are. Some, for instance, devi
ces carried by rockets or satellites, are not regenerated in flight but are regenerated during sto
rage or flight preparation.
the accuracy, in hydraulics the ab le, you have to expect mistakes and
sence of failures in various operation distortions. Some unreliability, even
conditions. The criterion of computer if it is extremely small, is common
reliability is the accuracy of input, to all man-made products. There
processing and output of processed fore, the problem of increasing relia
information. b ility looms large before science and
If a calculating device is unreliab technology.
There are several ways of creating redundancies. The method of “repetition” involves the assem
bly of a new circuit and the multifarious connection of it. The method of “net-like structures”
helps when both the main and the reserve structures are liable to failure. “Voting” involves the
replacement of the entire device or its block by an identical one and the connection of addi
tional special devices—the mixers.
the states leads to another state. Since each state strictly corresponds to its own
control command the defect will lead to a wrong instruction, and the system will
make a mistake.
Can anything be done about it? Yes, it can. The number of relays should be in
creased, and out of the possible states only such should be chosen for control pur
poses which do not in case of errors coincide with some other state.
This is termed creating a redundancy and has an analogy in nature which is
rich in excellent reliable systems made up of less reliable elements. For example,
the human brain as a system is infinitely more reliable than its single element —
the neuron: the organism operates even in case m illions of neurons are put out of
RELIABILITY 259
action. Try and find some technical device that would continue to function pro
perly after the failure of only one of its elements!
Redundancy is a promising trend. Appreciable results have already been ob
tained, still greater results are expected. Yet specialists do not consider this to
be the main road to high reliability.
Couldn’t a system, machine or device be designed that would give automatic
warning of a possible failure of a part, a unit or a block and that would, moreover,
automatically switch a reserve block, assembly or part into operation? Like a
living organism that sets in operation its reserve channels to preclude the possibi
lity of stoppage.
Work in this direction has already started.
Today some machines control their own operations and give warning signals
in case any defects arise.
Other machines are capable of testing their own electronic devices and of detec
ting 99.9% of tubes or other parts liable to failure.
At the Institute of Automatics and Telemechanics of the Academy of Sciences
of the USSR a group of scientists headed by Academician V. Trapeznikov have
built a digital computer which in case of failure hands out its own “diagnosis”
immediately showing the place of failure of some block or part.
As a result of the application of such advances in modern technology the relia
bility of big computers rose to nearly 98%. This means that the computers work
efficiently 85% of the time (with the exception of time reserved for inspection
and maintenance). A case has been reported of a computer, consisting of 13 thou
sand micro-circuits, working faultlessly for 33 thousand hours. This is a great
factor of reliability, though not the uppermost.
nose painted. His looks were very lamps in the room, opened the win
much inferior to those of his mecha dows and shut the doors. “My robot,”
nical predecessors, but there was a said the inventor, “without its shell
lot he could do. is actually an automatic telephone
“T elevox” worked as a permanent exchange, to which several electric
supervisor of the water tanks of one motors are connected instead of tele
of New York’s skyscrapers. He wat phone subscribers.”
ched the water level, switched on These electric motors performed all
water pumps. the actions of “T elevox”. In other
Moreover, “T elevox” had sound re words, “T elevox” was a typical repre
producing apparatus installed in him sentative of “electrical m en”.
and could pronounce several phrases. H is counterparts of the tim e were
He answered telephone calls about in close affinity to him . Among them
water level and pump operation. were the “Englishm en”: “E ric” built
This work is much more complica by engineer Richardson, “Alpha”, the
ted than the work of a mechanical creation of the professor of physics
musician. Judge for yourselves: now Harry May, the robot “W illy ” of
adays the work sim ilar to that of Westinghouse, and many, many
“T elevox” is being done by automa others.
ted traffic controllers in numerous Robots of today are electronic crea
automated plants. tures. One of them, an electronic ci
In addition to serious “professio tizen of Kaliningrad, says to this
nal” work “T elevox” was employed effect: “I have no heart. I ’m doing
in house work. He switched on the well with transistors and intricate
vacuum cleaner and the ventilator, electronic blocks.”
On p. 262 is the block diagram of the robot “Siberian-2”. He can do the job
of a guide to exhibitions, sell lottery tickets and books, advertise products and
polish floors.
The robot consists of 19 blocks. The main blocks are the control blocks of the
head, of the right arm, of leg action and of body rotation. A very important part
in the “Siberian” is played by the programming device and time relay.
H ere’s a description of the robot by its inventors from the city of Omsk: “The
robot draws power from the mains. The voltage of 27 V is applied to the program-
mator through the sound relay. There it is transformed into programme pulses.
For instance, the ‘le g ’ control block receives two programmes. The first sends the
robot on his way forward. To change course or to turn back a second programme
is needed. How does the robot turn a corner? An instruction is sent from the pro-
grammator to the control block, and the motor is switched on. After the corner
has been turned a second signal is sen t—this time to the stopping block, which
switches the motor off. At the same time a tape recorder with recorded speech is
switched on. As a result the robot turns and, for example, tells the visitors to the
exhibition about some exhibit.
“The movements of the right arm are somewhat more complicated. Simulta
neously with the motor a device is switched on which, say, hands out lottery tick-
262 CYBERNETICS A TO Z
Loudspeaker
S ig n a llin g b lo c k
A r m -m o v e m e n t b lo c k 1 H e a d -tu rn in g d e v ic e
Tim e re la y
E le c tr ic m o to r
E le c tro n ic b lo c k
M a g n e tic ta p e
re c o rd e r
P ro g ra m m a to r
T u rn in g b lo c k
A u to m a tic card
d is t r ib u t o r L e g -m o ve m e n t d e v ic e
E le c tr ic m o to r
The “Siberian-2” robot was built by boys from the Omsk municipal vocational school.
ets stored in the robot’s pocket. Having clutched the ticket, the robot firmly
holds it with his ‘fingers’ —contacts. As soon as a visitor takes the ticket the fin
gers are brought into contact, and the order ‘return to zero’ is sent. The arm mo
tors are switched on, and the robot resumes its former posture.”
These “electronic m en” that have sions and the com patibility of these
gone a long way from the superficial professions in a single robot.
likeness to man can do a lot of the Some robots possess faculties pecu
things we can do. liar only to them: for instance, they
The first thing that meets the eye can have sense organs which are de
is the “wide range” of their profes nied to man. For example, the robot
ROBOT 263
“□ T ry ” not only sees light, hears they sink to the depths of the
sound, feels heat, and notices obstac oceans, rise in rockets to carry out
les and avoids them artfully, but cosmic studies.
raacts to radioactivity, as well. Seve W ith the aid of robots functions and
ral metres away from the danger behaviour of living organisms are
zone it sends various warning signals. being simulated with the aim of crea
D on’t you agree that now the time ting better engineering constructions
has come to speak not only of robot and automata; biological processes
toys? Real robot-assistants have made are being modelled to gain insight
their appearance. into their essence.
Nowadays “electronic dwarfs” are Already now scientists are allot
engaged in testing cars and planes, in ting serious scientific tasks to robots.
shops with an unhealthy atmosphere; How much more is in store for them!
s The W ay to a “Clever” Autom aton
always reacts wrongly. So it appears ting system. Its trick is that it chan
at first glance. An organization is des ges from “bad” to “good” by itself.
cribed as good if it works faultlessly L et’s take the well-known example
and operates within some strictly of a ch ild ’s brain. At first its working
defined lim its, i.e. operates well no is such that the child is always attract
matter what it is —a cat, an automa ed by fire. The organization is clear
tic pilot or an automatic plant. ly a bad one. As a result of the expe
Wait a little , says to this a promi rience obtained a new—“good”—or
nent English scientist W. Ashby, ganization comes into being: now the
inquisitiveness is a good thing, but child avoids the fire, the brain as a
just think of the many antelopes who system had adapted itself.
died because they stopped to look at Is such self-adaptation possible in
the hunter’s hat. a technical system? Just think how
There is no quality or faculty of wonderful it would be! The automa
the brain unequivocally accepted as ton would function satisfactorily not
desirable in one situation that doesn’t only under normal conditions, but
become undesirable in another, main under break-down conditions as w ell,
tains the scientist. And he cites some could “level out” in any operating
examples. Here’s one. condition, would work like a man.
Is it good or bad for the brain to Self-adapting systems are classified
have memory? I t ’s good, if the am by their behaviour. The sim plest self
bient is such that the future often re tuning systems choose optimum ope
peats the past. If the events in future rating conditions with regard to am
would be opposite to those in the past, bient conditions. Such systems are
memory would be a disadvantage. increasingly being used in technolo
This situation occurs when a rat gy-
living in sewage tubes encounters a For instance, it is possible to bu
bait. The rat is very suspicious and ild a self-tuning machine tool control
accepts unfamiliar food only in sm all led by a programme. In this case the
portions. But if delicious food ap controlling device must mark the de
pears in the same place for three days viations in the dimensions of parts
running, the rat learns it. On the being produced and autom atically
fourth day it takes the bait boldly effect changes in the programme. In
and poisons itself. this case an in itia lly deficient pro
A rat devoid of memory (a bad or gramme w ill be improved in the cour
ganization by usual standards) w ill se of the to o l’s operation, and rejects
on the fourth day be as suspicious as w ill be m inim ized. Scientists term
on the first and w ill survive. Thus, this tuning for better operating con
in such conditions memory is a dis ditions “self-improvement by the ma
advantage. Does it follow that it chine tool of its algorithm”, of its
isn ’t possible to distinguish “good” working programme.
from “bad”, that any organization If the system can improve the algo
can at the same time be good and rithm of its operation, it can improve
bad? the algorithm of its behaviour, as
At this point we return to our for w ell, make it “flexible”, “searching”,
mer road and arrive at the self-adap adaptable to the ambient. Such a
266 CYBERNETICS A TO Z
system, which changes its mode of It occupies a position one step above
action in unforeseen circumstances^ the self-tuning system,
is usually described as self-organizing.
of the rheostats and electromagnet windings in the circuit. The positions of all
four cores were thus interrelated since the current in each solenoid was dependent
on the position of all four rheostats, and the position of each rheostat was depen
dent on the current flowing through the respective solenoid.
As soon as power was switched on all cores and rheostat contacts began to move.
The resulting situation could be twofold: after some transient process the cores
could occupy a stable intermediate state—in this case the system became sta
tionary; in another case the system failed to find for itself a stable state, and as a
result one of the cores exceeded the range of normal displacement and reached the
limiter. This led to arbitrary connections in the circuit, after which the search
for equilibrium was continued. After several switchings leading to arbitrary con
nections in the circuit of the homeostat the system ultim ately attained a state
compatible with equilibrium and then found this equilibrium.
H ere’s another example of a self of all odd numbers from the computer
organizing system —the m odelling of “memory”.
the process of survival on an electro This “w ise” class of self-adapting
nic computer. automata includes yet another spe
Imagine a computer the memory of cies. It bears the name of self-educat
which contains the numbers from 0 ing. To merit the name the automaton
to 9 quite arbitrarily mixed. In this must first of all be capable of sear
computer all numbers are m ultiplied ching. Having been trained in this
by pairs, and the number on the right faculty the automaton must be given
end of the product takes the place of a “memory” to enable it to accumula
the first m ultiplier. L et’s put the te information in the process of se
computer to work. We know that an arching. N ext a system of prizes for
even number m ultiplied by another successful ventures and of punishment
even number gives a third even num for the unsuccessful ones should be
ber. Odd, m ultiplied by odd, also devised. This method of self-education
gives odd, and odd m ultiplied by is termed the trial-and-error method,
even gives even. The conclusion can and i t ’s being used in some program
be drawn that after m ixed encounters mes for solving educational prob
the number of even numbers in the“me- lems.
m ory” of the computer w ill grow—the H ere’s an example of a programme
even numbers have greater chances of for a self-educating computer.
“survival”. Gradually they w ill take The computer was divided in two.
the places of odd numbers in the com One section played the part of the
puter “m emory”. The computer has “pupil”, the other of several “shops”
“organized its e lf” for “su rvival”. In with varying goods assortment. The
time the practice of this purposeful task of the “pupil” was to learn to
behaviour w ill lead to the eradication find the necessary goods quickly.
SELF-ADAPTING SYSTEM 269
At first the ‘‘pupil” roamed the called “clever”, for they are very
shops in search of the goods. In one much like real live pupils. It is not
shop he was in for success—he “came without foundation that scientists call
across” the goods. He “remembered” such mechanisms “brain-like”.
the “shop”—for this he was given a It should, however, be kept in
“prize”. mind that research in this direction
The goods were changed. Out he is actually just beginning. There is as
went shopping again. New searches, yet no general theory of self-adapting
followed by a new success, for which systems. This field of research may
a prize was due. And, of course, a new be likened to a great field touched here
“knot” in the “memory”. This was and there by the plough of research.
repeated over and over again. The But should this field be properly
training resulted in self-education of cultivated, a generous crop w ill be
the ‘ pupil”. He unm istakingly chose the reward for the time and labour
the right shop for specific goods. spent, a crop of all kinds of self-adapt
Needless to say, automata capable ing systems, true servants of man.
of acting in this manner are rightly
270 SEMIOTICS
Signs are always part of a system. For instance, the identical sign “P ” in the
Latin ABC means the letter “P ”; in the Russian ABC it corresponds to the letter
“R ”, and as a traffic sign it denotes a parking place.
W ith the aid of sign systems people communicate with each other, such systems
enable them to study nature, to work.
Scientists divide the sign systems into two categories: natural and artificial.
Animal “languages”, “odour languages”, “finger languages”, etc., are natural
systems. The most advanced natural sound system is the human language. I t ’s
a very fluid, flexible, well-developed system. Having been born out of communi
cation, the human language serves as a means of communication between peop
le. Our language helps us to express thoughts, wishes, to transmit the nuances of
feelings. V. I. Lenin had a good reason to pronounce the language to be the
main means of human communication.
The higher mankind rises in its social progress the more new sign systems come
into being. For this reason their number is continuously growing. And for the most
part this is due to artificial sign systems. Mathematical and physical formulae,
the symbolics of chemistry and traffic sign code mentioned above can all serve
as examples of such systems.
As a rule the role of artificial systems is auxiliary: they express things that can
be expressed with the aid of natural signs, but more concisely, precisely and eco
nom ically. They at once reflect the ultim ate result and the way that leads to it.
Thus, if anyone would like to “translate” a familiar formula into words made
up of ordinary letters, he would need several textbook pages to express the result
of a simple mathematical calculation.
Try and add the number one m illion seven hundred thirteen thousand five hun
dred and one to the number twelve m illion one thousand three hundred and nine
ty-nine, written down in words. I t ’s no easy job.
And i t ’s so easy with the digits!
, 1 713 501
+ 12 001 399
13 714 900
There are “independent” artificial systems as well. Such systems in increasing
use nowadays include first of all languages serving as intermediaries in computer
translation and logical calculus.
There exists also another subdivision of signs. They are classified as language
and non-language signs. This classification is easy to perform. It, so to speak,
lies on the surface. Evidently, all natural and artificial languages are language
signs. Various schematic diagrams, blue-prints, drawings, signboards, maps,
illustrations, diagrams, dances, pantomine, music, sculpture, etc., are non-lan
guage systems.
Signs and sign systems are studied Obviously, signs attracted the at
by a special branch of science—semio tention of scientists. Since signs ap
tics. It derives its name from the peared as soon as men began to think,
word “sem ios”, the Greek for “sign”. their studies date very far back. Al-
NATURAL LANG UAG E
LANG UAG ES OF A N IM A L S
LANG UAG E
A R T IF IC IA L
Mg, Cu, N
-
SYMPTOMS
w
w
c
o
LANG UAG ES
N O N -L A N G U A G E
m
T m
C o p ie s S ig n a ls S ym b o ls Im ages
SEMIOTICS 273
ready A ristotle and other Greek phi the point of view of th eir value as
losophers gave a thought to the gnos signs serving to express some con
tic value of signs. As m an ’s experien tex t.
ce extended and science thrusted dee T his is the reason for the exceptio
per into the secrets of nature, regula n ally wide range of action of sem io
rities common to various sign sys tics. To support th is claim l e t ’s cite
tem s became clearer. B ut p rim arily the proceedings of such an a u th o rita
the ideas of sem iotics won a place for tive assem bly as the first Soviet sym
them selves in the m athem atical logic posium on sem iotics. I t took place in
w ith its symbols and precise defini December 1962 in Moscow.
tions. Here signs were more in evi The scientists considered from “the
dence th an in other branches of sci sem iotics point of view ” diverse pro
ence. Achievem ents in th is field are blem s, som etim es quite unexpected
connected w ith the names of the Ger and strange for the laym an.
m an scientist G. L eibnitz and the Of course, m uch a tte n tio n was paid
E nglish philosopher J . Locke. to the n atu ral language in its capaci
N ext linguistics fell prey to semio ty as a sign system .
tic ideas. The need for comm unica The description of e tiq u e tte is of
tio n between people leads to the estab in terest from the point of view of
lishm ent of a system of symbolic sem iotics. I t tu rn s out th a t fortune
signs, m aintains the French scientist tellin g w ith the aid of playing cards
F. de Saussure. He cited ceremonies, presents a ttractio n s to sem iotics, for
etiq u ette, w ar signals as examples. the reason th a t “th is rela tiv e ly sim ple
Among all system s he a ttrib u ted the sem iotic system can be of in terest to
m ajor im portance to the language. general sem iotics” . Sim ilarly, street
The m ain principles of the science traffic control “facilitates the estab
of sem iotics were form ulated by the lishm ent of certain regularities com
Am erican scientist Charles Peirce. In mon to sign system s” .
the th irtie s they were extended and C lim bing step by step the ladder
developed by scientists from m any of “establishing certain re g u la ritie s” ,
countries belonging to m any scientific sem iotics reaches the heights of quite
schools. A leading place among them “u n e a rth ly ” subjects. I t provided the
and the most im portant results belong specialists w ith an in tric a te tool en
to the Polish and R ussian schools. abling a language to be constructed
D id n ’t it appear to you th a t semio for com m unication w ith civilizations
tics tries to grasp w h a t’s out of reach? of other planets. Yes, t h a t ’s right.
T h at the field of interests which it is On the E a rth a special language has
try in g to “in v ad e” is too wide? No, been developed for the intercourse
is the answer of the experts. Sem iotics w ith our “brothers in in te lle c t” . Its
tries to tackle m any fields of hum an author, the D utch . scientist H ans
knowledge b u t only from one point F reudenthal, has given it the nam e
of view: different objects of research “lincos” —langua cosmica, the cos
are studied unidirectionally, only from mic language.
A Classification of signs.
274 CYBERNETICS A TO Z
Lincos is based on the u n ity of the the creation of lincos does not dispell
laws in the U niverse, specifically, on th is view. I t ’s true, sem iotics is a
the u n ity of m athem atical laws. I t ’s theoretical discipline, b u t it is still
on the basis of these law s, which of some practical value. T hink, for
reflect the rea lity of the w orld, th a t exam ple, of the artificial sign system s
F reudenthal builds his m ulti-stage —interm ediary languages for cyber
lincos structure. This language has netic m achines. Moreover, sem iotics,
hierarchical organization, it contains together w ith psychology and physio
an in tric a te system of relationships, logy, studies the location of speech
com plications, “advancem ent stages”. centres. The results of these studies
O bviously, appropriate m eans m ust proved to be of undeniable practical
be chosen for cosmic com m unica value: the achievem ents of several
tio n s—a radio signal or a lig h t pulse. Soviet scientists aided in the con
To begin w ith , prim ary m athem ati stru ctio n of a special language for
cal concepts should be coded w ith deaf persons. This is an exam ple of
th eir aid: dig its, equation signs, the help given to m edicine.
essence of the b inary system . Next A nother exam ple is from the field
a more advanced stage is reached: of pedagogics.
the exposition of the rules of arithm e The pedagogues are paying an ever
tic . Then comes the tu rn of algebra, increasing a tte n tio n to signs. T his is
and so on, up to higher m athem atics. not to be wondered a t, since education
F inally, w ith the aid of ab stract m a en tails, to a considerable degree, the
them atics F reudenthal turns to evalu m astery of signs.
ating hum an behaviour, sta rts te ll And those who are not convinced by
ing the story about us, the inhabi the exam ples should not forget th a t
ta n ts of the E a rth , about our life, i t ’s only recently th a t the m arvellous
about the E a rth —the home, where faculties of sem iotics began to be
we were born. displayed. For th is reason i t ’s n atu ral
You m ay accuse sem iotics of being to expect from th is science new disco
a purely theoretical science, for even veries and new successes in the future.
275
A n A u to m a to n D ecides upon a M a rk
Isn ’t th a t a perfect exam iner? The W ell, and w h a t’s going to happen
r e ’s a m achine th a t can check 1025 to the teacher, to the m an who does
exam ination papers, practically for all the teaching now?
disciplines. W ith the introduction of the m ost
The num ber of m achines used for perfect autom ata his role w ill not
education is continuously growing. abate, b u t, on the contrary, w ill grow
Isolated experim ents are m aking way in significance because of new duties.
for wide application of educational These duties w ill include the com pi
m achines. For several years now pro latio n of the program mes and th eir
gram m ed education has been p rac ti constant im provem ent. The teacher
sed a t hundreds of colleges and voca w ill, as before, play the m ost im por
tio n al schools. Over 1000 schools have ta n t p a rt in the entire educational
specially equipped classes. There is process, especially in fostering. And
every reason to suppose th a t in the the m achine w ill be his reliable edu
future such m achines w ill be emplo catio n al assistant.
yed universally.
Not one b u t several hundred types The second group embraces tra in
of teaching machines have already been ing m achines.
b u ilt, some sim ple, others complex, The th ird —testing and teaching m a
among them m idget “exam iner-tea chines. Among th is group the m ajority
chers” the size of a cigarette case and is m ade up of exam ining m achines
large m achines occupying whole rooms. used to test the knowledge of the stu
The teaching m achines are subdivi dents. The m ost w idely used is “The
ded into groups in accordance w ith Sw allow ” .
th eir purpose. As a rule, th eir names speak for
The first group is m ade up of simple them selves: “The L ecturer”, “The Con
m echanized devices. s u lta n t” , “The T rainer”, “The Coach”,
Information output A Knowledge evaluation
4- block block
| STUPE NT |
AV ▼
Z Z ? -----
WORK STARTS
T est q u e stio n ^
Ye s No
Leading questio n ^ ^
No
T e st qu estio n ^ z
es
r ^
WORK ENDS
TEACHING MACHINE 279
“The T ester” , “The E xam iner”. There of the m aterial and proceed further.
are universal “m achine educators” , The weak stu d en t, on the other hand,
as well: they test, consult and exam i w ill be able to work slow ly, not get
ne. tin g nervous and not try in g his u t
The fourth group of educational de m ost to catch up w ith the class. In
vices is the m ost com plicated of them th is w ay the principal doctrine of the
all. I t consists of classes for program old “unprogram m ed” educational th e
med education. ory, th a t had been very difficult to
L e t’s learn about one of them , “The p u t into practice, is realized: the doc
Accord”. I t has nothing to do w ith trin e of in d iv id u al approach to every
m usic. The name is made up of in i stu d en t.
tia l letters of its R ussian designation Program m ed education th a t w ill
m eaning an “autom ated class for have to absorb the best achievem ents
controlled education w ith ram ified of our educational system w ill not
dosage” . L e t’s enter it. only m ake life easier for the stu d en t,
There are th irty tables. On each b u t for the teacher-program m ist, as
tab le there is a sm all control panel w ell. The teacher-program m ist w ill be
w ith lam ps and levers. The big cont able to tra n sm it the knowledge of his
rol panel of the teacher is connected subject to any group of students irres
to the sm all panels. pective of th eir level of knowledge
E very stu d e n t has a chapter from and of in te lle c tu a l developm ent. By
the “educational program m e” in front placing accent on in d iv id u al work
of him w hich he m ust learn. The chap the teacher m akes every stu d en t not
te r ends w ith questions. W hen the only study, b u t learn, as well.
stu d e n t th in k s he has learned his Much has been done in th is country
lesson he operates an appropriate to develop the new educational me
sw itch. thod.
If the answer is correct, the signal Special educational program mes are
“c o n tin u e” w ill lig h t up, if n o t—“re being com piled, the efficiency of va
p e a t” . rious m ethods of program m ing is being
Now, note the following. There are, studied, models of the educational
as usual, th ir ty students to a class, process based on the theory of proba
and there is, as before, one teacher to b ility and m ethods of controlling th is
teach them . The educational process process are being designed, the search
is, no doubt, collective: and, a t the is on for a universal educational
same tim e, in d iv id u al. Here the s tu algorithm (i.e. rules).
dents are not “reduced to the ave However, scientists and educatio
rage” . n a lists endeavour to look farth er ahe
There is a single program m e, but ad. They would like to build educa
everyone can learn it in his own way tio n a l complexes and com pile educa
adapting i t to su it his nature. A ca tio n a l program m es th a t would take
pable stu d en t w ill m ake quick work into account the in d iv id u al propensi-
M A block diagram of the educational machine “The Coach” and the route of education process
from the start to the end of its work.
280 CYBERNETICS A TO Z
ties of students as w ell. The new me even thousands, of consumers. This
thods are being applied in specialized w ill enable perm anent controlled edu
education: in sports, in m usic, in cation to be conducted from class
m edicine. work to home work. This m ay be sup
To teach the blind and deaf a me plem ented by an educational tele
thod of reading h an d-w ritten tex ts vision netw ork.
as they are w ritte n is being developed R ecently Soviet scientist 0 . Belo-
and tested experim entally. Such a tserkovsky read before an in tern a
m achine w ill be able to correct d ic ta tio n al conference an in terestin g paper
tions as they are w ritten . w ith the title “The Effect of Space
A design of a m achine capable of Research on the D evelopm ent of Ge
talk in g and understanding hum an la n neral and Specialized E d u c a tio n ” . The
guage has been developed. Such a paper dealt w ith the em ploym ent of
m achine w ill be able to teach hund com m unications sa te llite s equipped
reds, and in the foreseeable future w ith tra n sm itte rs to tra n sm it educa
thousands, of students. To teach va tional radio-television program m es to
rious subjects and w ith an ind iv id u al wide expanses of the E a rth .
program m e for every stu d en t. The The scien tist envisages the tim e
m achine w ill read w hat the student when powerful television tran sm itters
has w ritte n , hear him speak, in short of the E a r th ’s sa te llite s “hanging s ta
w ill rea c t to everything the student tio n a ry ” a t a height of over 30 thou
does in the process of education, and sand kilom etres w ill s ta rt a new school
this w ith o u t any control panels, le year in the vast school, the desks of
vers or push-buttons! which w ill be strew n over not only
The problem of em ploying big elec towns and villages, b u t over the bo
tronic com puters in mass education is undless expanses of jungle and p la
being considered, too. teau, over islands lost in the ocean
Even the use of electronic teaching and over the oases of the deserts. In
m achines in flats is envisaged. They th is way a wide coverage of various
w ill be connected to the sta te educa sections of the w orld’s population
tio n al m achine netw ork. In fact, two from the little ones to the grown
or three electronic m achines can a l ups could be achieved. T his, too,
m ost sim ultaneously provide educa would, of course, increase the effici
tional inform ation for hundreds, or ency of education.
TRANSLATION, COMPUTER 281
T ranslation from one language E ven if m agnetic tape was used ins
natural or artificial) tead of the punch-cards, th is would
require a huge num ber of dictators.
into another language M oreover, it would take, lite ra lly , an
(natural or artificial) arm y of editors to read and ed it the
tex t.
w ith th e aid
A ren’t those difficulties form idable?
of electronic computers. You bet! Y et, those are not the m ost
serious. The m ain difficulty lies in
the large volum e of the vocabularies
of m odern languages. This leads to
inaccuracies in tran slatio n s from one
•'The Horse [Called] Charley” language into another. An elu cid at
ing experim ent of French linguists,
who tried to assess the degree of pre
On Jan u ary 7, 1954 in New York cision atta in a b le in tra n sla tio n from
the first public dem onstration of an one language into another, brought
electronic com puter in its novel qua the result th a t rem inds one of the
lit y —th a t of a tra n sla to r—took pla fa u lty telephone game. Fourteen expe
ce. The com puter tran slated R ussian rienced tran slato rs took th eir seats
phrases into E nglish. In all, 60 sen a t a round tab le so th a t each knew
tences were tran slated . the language of his rig h t neighbour.
The com puter tran slated . I t did the The first tra n s la to r—a G erm an—wrote
job like a m an ignorant of the langua on a piece of paper: “The a rt of brew
ge: w ith the aid of a dictionary. Man ing is as old as the history of m an
finds unam biguous words and a rra n k in d ” , and handed the piece to his
ges them into a sentence according left neighbour. The la tte r tran slated
to the rules of gram m ar. Needless to the te x t from the German into his
say, such a tra n sla tio n is far from native Spanish, w rote it down and
perfect, y et, it caused great difficul in his tu rn handed it over to his left
ties to the com puter. The first diffi neighbour. The sentence w ent on its
culty can be defined as technical. way round the table, everybody tra n s
Judge for yourselves: the special pro la tin g i t into his n ativ e language. At
gramme containing rules for the tran s la st it returned to the G erm an in
latio n consisted of 2500 instructions, H ungarian. He tran slated it and read
which is m uch more th an for solving w ith surprise: “ From ancient tim es
complex m athem atical problem s. beer has been the favourite drink of
There was, too, another side to the m an k in d .”
technical difficulty. The com puter Idiom s present great obstacles to
could read 1 800 000 letters per m inute, com puters. The E nglish words “ char-
b u t to provide it w ith a corresponding ley horse” the com puter w ill tran slate
num ber of punch-cards 12 thousand as “horse [called] C harley” , when
ty p ists w orking a t a speed of 10 thou actu ally it m eans a “cram p in the
sand signs per hour would be needed. calf of a leg ” . “ Foolproof” —“proof
282 CYBERNETICS A TO Z
ted into the com puter w ith tape carrying the E nglish tex t. The inscription is,
however, not in signs b u t in perforations, as in punch-cards. This is th e code of
the te x t being tran slated . Placed next to this reel is a reel w ith narrow m agnetic
tape carrying the com puter program me for tra n sla tio n operation. The com puter
“m em ory” stores in its cells R ussian words arranged in stric t order next to cor
responding English words.
The com puter tran slato r, like m an, makes use of the dictionary. The only dif
ference is th a t in its d ictionary words are “w ritte n ” not in signs b ut in digits.
The English “a ” became 16, “b ” - 0 6 , “w ”—13, “m ”—11, “n ” - 1 5 , “x ”—09,
etc. R ussian letters, too, becam e numbers: “a ” —16, “6 ” —06, “ b ” —13, “ m ” —11,
“ h ” — 15, etc.
T h e o p e ra to r a t th e le tte r ke yb o a rd T a b le o f E n g lis h
a u to m a tic a lly s u b s titu te s num bers le tte r co d e s
fo r le tte r s (s e e T ab le I) , fo r in s ta n
ce A —16, M—11 :'A- 16: i.M-ll';
B- 06 N -15 V- 29
P u n c h e d ta p e w it h n u m b e rs is -|C-22 0-20 W-12
fe d in t o t h e c o m p u te r 1162226121508 00 120500 210716150527162112151(^2 T h e E n g lis h p a rt
o f th e d ic tio n a ry
Left column Right column
T ra n s la tio n program m e is s w itc h e d on 16226121508-121100 2563254)
1205 2100110 ....*0404...
S u b tra c t th e c o d e s o f th e w o rd s o f th e
le f t c o lu m n o f th e d ic tio n a r y from 8l
2107161508'
100... 41212 56
th e c o d e s o f th e w o rd s , i n tu rn , ^ ____ . 271621088J 1
u n til th e re m a in d e r is 6 . When t h is 191100.256^3254 210011— 0404 The m ea n in g o f
is d o n e , p u t th e co d e o f th e r ig h t c o - l ^ ' th e rig h t-c o lu m n codes
lu m n (fo r it s m e a n in g see T a b le 3) in
Nouns
p la c e o f th e w o rd in th e m em ory d e v ic e
W A R N IN G ! T h e th ir d w o rd h a s n 't
been fo u n d in th e d ic tio n a r y
@ Feminine gender ....
ior Num ber in fhe Rui-
2 32 54: si an part ot the dic-
'/im " tionary
C h e c k th e w o rd u s in g T a b le 4. D o e s 4 The d ia g ra m o f
i t end o n " s " ? N o ! D o e s ite n d o n " e " ? the stages o f check
ing tn e u n id e n ti-
No! D o e s it end on " i n g " ? Y e s ! -------- - fie d w ord
D e le te th e e n d in g and lo o k up the. I Dictionary | no
d ic tio n a r y a g a in . Is th e re an " e ” ?
Y e s . Is th e re an " i s " in fro n t? Y e s .
I t 's c le a n t h is is a v e rb in th e p r e * j
se n t c o n tin u o u s te n s e . l t is tra n s la r 5 The R u ssia n p a rt
te d as p re s e n t.T h e v e rb " i s ^ i s n o t 121100^.256 3254 21100.,.4121
. o f th e d ic tio n a r y
tra n s la te d (th e re is n o th in g in i t s n i
p la c e )
These pieces of tape have been taken out of the computer after translation. Here is the literal
translation: “ 3 to n e H e o 6 x o « H M o .”
The translation starts with the computer searching in the dictionary for the
words recorded on the tape. The computer has found a word. Hasn’t it made a
mistake? How could this be checked? The arithmetical device w ill subtract from
every word-number found in the dictionary the word-number recorded on the
punch-tape. If the remainder is zero, the word has been found correctly. Such
comparison takes about one ten thousandth of a second. The computer can look
through a dictionary of a thousand words in less than a second.
Next the computer turns its attention to the index number of the English word
found in the dictionary. The corresponding Russian word bears the same number
in the dictionary. And this word, too, is written in digits. If we now translate the
se numbers into corresponding Russian letters, we w ill get the Russian word—a
translation of the English word introduced into the computer. The words have
been translated, but the computer cannot, as yet, construct a Russian phrase.
First it has to analyse the grammatical form of the English and the Russian words:
the gender, the number, the case, the declination, etc. In the computer these cha
racteristics, too, have the appearance of numbers and are stored in the “memory”.
Word parts such as suffixes, endings, prepositions and articles of the English words
have been translated into the language of the so-called digital information that
is acceptable to the computer.
Only now does the computer start analysing the English phrase as a whole.
Subsequently it constructs the Russian phrase. This is done on the basis of the
translation programme which contains the paragraphs: “verbs”, “nouns”, “ad
jectives”, “numerals”, “syntax”, “changes in the word order”. The computer
constructs the Russian phrases from the words translated from the English in com
pliance with the rules of the Russian grammar.
TRANSLATION, COMPUTER 285
tronic com puter became the m ost po successfully enters fields which u n til
w erful and the m ost flexible calcula quite recently were considered to be
tin g instru m en t ever to be b u ilt by the exclusive privilege of m an. Ma
m an. A high-speed com puter can per chine tools, shops and factories are
form over a m illion operations per controlled by com puters. Com puter
second. control of production processes ap
C alculations made w ith lig htning preciably increases the pro d u ctiv ity
speed. Yes, t h a t ’s rig h t, electronic of labour and m akes work easier for
com puters do deal in sta n tly w ith a the worker.
flood of num bers. In one second the And the control of factory econo
com puter performs m any tim es the mics? The com puter “econom ist”
num ber of operations performed by an took over from m an m any functions
experienced calculator w ith an a rith of planning and analysing various
m om eter in eight hours. In the tim e economic param eters. Here, too, ad
of several hours the com puter makes vantages are obvious: the control be
as m any calculations as a good m athe comes more operative, the num ber of
m atic ia n is unable to m ake in his people employed in the control sphere
lifetim e. decreases.
U nbelievably high speed of count The fields of application of the elec
is only one of the ex traordinary pro tronic com puter are m uch wider. There
perties of the m arvellous m achine, are com puters th a t do the work of des
which can ex tract roots, in teg rate and igners, tran slato rs, teachers, meteorolo
solve the sim plest algebraic equations gists. How do these electronic craftsm en
as well as m ost complex differential work? The chapters of our encyclopae
equations. dia te ll you about it, as well as about
Now the electronic autom aton can the principle of operation of the elec
cope w ith everything. The m achine tronic d ig ital com puter.
Numbers enter the computer by way of th e coding device. Here numbers and
instructions undergo transform ations and assume the form suitable for com puter
operations. N ext they are fed into the in p u t device and into the working memory
store.
Some of the numbers rem ain here for the tim e being inactive. For the others
the working store plays the p art of a tra n sit statio n . Through it some numbers
reach the perm anent “m em ory” , a sort of com puter’s “notebook”. Numbers are
stored here by m illions, the tim e of storage being unlim ited.
Other num bers are needed for im m ediate processing. They are in sta n tly fed
into the arithm etic u n it, consisting of adders, m u ltiplication, division and sub
tra c tio n circuits. The la tte r perform all arithm etical operations w ith the aid of
addition.
In addition to perm anent, long-tim e “m em ory” the electronic com puter pos
sesses a working “m em ory” as well. This is needed to record d a ta frequently used
in the course of work.
The capacity of the working “m em ory” is not laige, b u t it hands out numbers
quickly a t short notice.
UNIVERSAL ELECTRONIC DIGITAL COMPUTER 289
Man and machine carry out calculations in almost the same manner. The arrows show the paths
of control signals.
The u ltim ate results of the calculations—the finished product of the com puter—
enter the output block and are typed on paper tape or on blanks of a specified form.
The arrows in the draw ing show the general path of numbers and instructions
in the com puter. Red arrows indicate the path of num bers, w hite—the path of
instructions.
If we looked, from a great height, at thousands of trains running in different
directions along steel tracks we would perceive a sim ilar picture. The m otion of the
trains, disorderly at first glance, is governed by a single plan, a single tim etable.
<4 Block-diagram of a high-speed electronic computer. Red arrows denote the routes of digits,
white ones—the routes of instructions.
19— 6 1 6
290 CYBERNETICS A TO Z
The computer works in separate cycles. Look at the control block. Out of all
the com puter parts it is, of course, the m ost im portant, for it is due to this block
th a t autom atic operation of the calculating blocks is possible.
The control block in the draw ing has ingoing and outgoing blue lines of instruc
tions. D uring each cycle of com puter operation an instruction from the working
“m emory” is transm itted to the control block. I t is recorded here and carried out
during the next cycle. A special counter counts the instructions after they have
been carried out.
The process is repetitive: choice of an in stru ctio n —execution; choice of an in
stru ctio n —execution. The execution of the prescribed computer programme con
sists of a great number of such repetitions, each of which, in short, makes up one
autom atic cycle (1—the choice of an instruction from the program me, 2—the exe
cution of the instruction).
All the highways in the com puter pass through the working “m em ory” store. This
is understandable, for it is the receiver of the calculating process programme fed
from punch-tape or punch-cards.
The working of a universal electro here. However, now you have some
nic com puter is, of course, m uch more idea of the basic principles invol
sophisticated th an has been shown ved.
291
V Open, Sesame!
sort of grain it is. And, secondly, it letter-sign per day. N atu rally ,, such
should know w hat to do w ith it. speeds of inform ation o u tp u t are un
W hen m an begins w orking w ith the acceptable.
m achine he provides in itia l data in W h a t’s the solution to th is prob
the form, wholesome for the m achine lem? How could the d isp a rity between
and in stru cts the m achine to carry the cap ab ilities of m an and com puter
out specific actions. The m achine speed be avoided?
reads the d ata, processes them , records Special high-speed in p ut-output
the results obtained and transm its complexes have been designed for
them to m an in a form understan com puters. U sually, they consist of
dable to him . punch-card reading devices, several
Previously m an tra n sm itte d the in m agnetic-tape data accumulators, a tele
form ation to com puters w ith the aid type and a high-speed p rin tin g device.
of keyboard devices, th is is s till be As you know, the punch-card rea
ing done. The underlying principle ding device introduces into the com
has long been used in p rin tin g for puter inform ation contained in per
settin g the types of newspapers, m a forations of the punch cards. The
gazines and books. m agnetic-tape memory devices—accu
Im agine a stan d ard p rin tin g m a m u lato rs—enable the interm ediate re
chine which transform s symbols shown sults of com puter operations to be
oh the keyboard into a definite sequen stored. They can also serve as long
ce of pulses and in te rv als between tim e inform ation store. The teletype
them . You press one key after another, is used for short messages, for in
and pulses corresponding to the in stance to tra n sm it an in stru ctio n to
form ation enter the m achine. the operator to in sta ll a definite m ag
B ut you w on’t be quick in in tro d u netic-tape bobbin. The operator, too,
cing a m ass of d ata into the com puter uses the teletype to send a signal to
m anually, “by the fingertips” . Even begin calculations. The p rin tin g de
a ty p is t of the highest grade cannot vice is used for the ou tp u t of the results.
ty p e more than six signs per second. Modern in p u t-o u tp u t devices ope
The electronic com puter, on the other rate a t the highest possible speeds.
hand, handles several m illion bits They “swallow u p ” some 1000 punch
per second. cards per m inute and hand out data
To m ake th is discrepancy more evi quicker th an m an can read them .
dent im agine the operating speed of And even the teletype, th is appa
the com puter to be a m illion tim es ren tly m ost sluggish block of the com
less. Our slow com puter w ill operate puter, types m uch quicker and much
a t speeds convenient to m an—one more accurately th an a qualified typist.
operation per second. The correspon However, even these high-speed im
ding speed of the keyboard prin tin g plem ents do not enable the possibili
device w ill in th is case am ount to one ties of the reading devices of modern
sign per day! com puters to be fully u tilized.
The operator aw aiting an answer Suppose you would like to feed into
from the com puter could in these c ir the com puter not num bers or letters,
cum stances be com pared to a m an b u t p rin ts, diagram s, graphs, draw
receiving a telegram a t a rate of one ings? W hat would you do?
VIDEO DISPLAY 293
Previously, the user of the com puter X -ray instrum ents, electron m icro
recoded all the graphical inform ation scopy—this “electron device u tilizing
into num erical coordinates himself one or several electron beam s” works
and only then tran sm itted them to everywhere.
the com puter. Now the com puter has Nowadays there is a whole fam ily of
been tau g h t to accept and hand out electron-beam tubes, consisting of ne
graphical inform ation in the form to ar as well as d ista n t relatives.
which m an is accustomed. This L e t’s trace the line of kinship of
su b stan tially speeded up the “exchan interest to us in this fam ily.
ge processes” between m an and m a Take the fam iliar electron-beam tu
chine. Form erly it took fifteen m inu be of the television receiver. This is
tes for a teletype to p rin t the coordi a tube th a t draws pictures. The other
nates of a thousand points of a straig h t name for it is the kinescope. I t bu
line. The screen of a video display ilds up an image by varying the elec
depicts a stra ig h t line in a m illi tron beam.
second, or two! If a highly focused electron beam —a
V isual images are introduced into beam incident in one p o in t—is d i
the com puter w ith the aid of a combi rected onto a screen covered w ith a
nation of photocells and an electron- lum inous compound and made to move
beam tube. Each photocell is so ar across the surface of the screen w ith
ranged as to receive lig h t only from the aid of an electric or m agnetic
a definite point of the screen. The field, th is w ill m ake an oscillograph.
com puter identifies the image dis I t has found wide application in the
played on the screen by the pattern technique of m easurem ents.
of signals from the photocells. And now w e’ll try to build into the
A com bination of a pen and a pho tube several plates, standing in the
tocell enabled the so-called “lum inous way of the electron beam , and con
pen” to be designed. This is a sm all nect them to a system of signalling
hollow cylinder housing a photocell. or control.
This pen is used for the in p u t of draw N ext w e’ll m ake the beam run in
ings, prin ts, diagram s drawn by hand several directions: from the cathode
on a screen resem bling the screen of to different plates, whereby the elec
a television set. tric a l c irc u it between them w ill be
Some 50 years ago the electron- a lte rn a tely connected and disconnec
beam tube was a ra rity . Now i t ’s a ted. This w ill m ake a quite different
perm anent resident of practically eve electronic device—the electron-beam
ry home in town and country. The t u com m utator.
bes live in our television sets, and i t ’s Now l e t ’s place another obstacle in
their screens we watch. the way of the electron beam: a plate
Of course, the sphere of application w ith lette rs inscribed on it, the le t
of the electron-beam tube is not li ters being im pervious to electrons.
m ited to dom estic use, its range of T his tim e we again have a new devi
action is exceptionally wide and m a ce—an electron-beam w riting tube.
nifold-. Suffice i t to nam e, besides tele This is the tube th a t bears the name of
vision sets, m easuring instrum ents, charactron, th e lum inous pen.
radar, com puters, radio, autom atics,
294 CYBERNETICS A TO Z
Since our interest is focused on the charactron, le t’s get down to d etail.
In the charactron a m etal plate (called the m atrix) w ith a set of letters, num
bers or any other signs (called the stencils) is placed in the way of th e electron
beam. To be able to record the signs-stencils on its screen the tube m ust also con
tain some other im portant p a rts—systems for the form ation, collim ation and
deflection of the electron beam . In the charactron those system s are: the electron
projector ( “the electron source”), the system for the selection of signs on th e m a
trix , and the address system .
VIDEO DISPLAY 295
The projector reflects on the cathode the image of all the signs contained in the
m atrix . However, not all the signs are needed a t the same tim e. So the selection
system goes into operation. It selects the appropriate stencil and takes it through
the diaphragm .
However, having passed through the diaphragm the sign w ithout the aid of
the address system would “hang in the a ir ” , would not be able to find its place
on the screen. The address system , so to speak, “takes the sign by the h a n d ” and
leads it to the place reserved for it on the screen.
After th a t the image on the screen disappears, and a new inscription appears.
I t is, however, possible to m ake the image stay on the screen for some tim e; in this
case it is possible to read it or photograph it w ith the aid of a high-speed camera.
Here all depends on the designation of the charactron. The tim e w ithin which the
signs disappear—the scientists call it the tim e of residual illu m in atio n —m ay be
sh o rt—some 10-20 microseconds (ps) or long—5-10 s.
The signs on the charactron are usually quite sm all—2-5 m m. B ut when the
need arises they can be enlarged. To this end some devices are equipped w ith a
special electron lense, which w ith the aid of the electric field can enlarge the signs.
The screen of the electron recording tube is generally large, its diagonal being
from 15 to 75 cm long. The density of signs on the surface of the screen is fairly
high—the screen accommodates up to 16 thousand letters or num bers. The w rit
ing speed of the “luminous pen” reaches 4 thousand signs per second.
The charactrons are w idely used in stored during the tim e needed to pro
electronic com puters, m ainly for data cess all the d a ta received.
o u tput. I t ’s very convenient to ob The use of advanced electronic in
tain the results of the “com puter’s put and ou tp u t in place of electrom e
lab o u rs” quickly and in a visible form. chanical devices for the com m unica
Such a charactron does the work of an tions between m an and m achine brought
o u tp u t p rin tin g device. Its data selec about a thousand-fold increase in
tio n speed of 25 thousand signs per com puter in p u t speeds. Now i t has
second (for inform ation in the form of become possible to erase im m ediately,
letters or num bers) is a convincing as one would wipe off a speck of dust,
proof of its q u ality . an unneeded record or an error, to
“W h a t’s the good of i t? ” you would introduce easily and quickly the ne
ask. cessary data corrections. The “lum i
In less th an a w ink a ll those 25 nous p en ” m ay even be used to con
thousand signs w ill disappear to make tro l the com puter: to th is end the
place for the next 25 thousand. And word or code of an in struction or an
so every second. W h a t’s to be done appropriate image should be w ritten
w ith them ? T h a t’s easy—you just on the screen.
have to resort to a high-speed cine The new devices possess all the ad
cam era. It w ill photograph the image vantages: they are fast, noiseless, re
displayed on a film and thus “preser liable, universal, provide for an in
v e ” it. These “inform ation preser sta n t, direct and com plete access to
ves” —letters and num bers—can be inform ation stored in the com puter.
296 CYBERNETICS A TO Z
The operations m ay be conducted on card w ith the sym bolic num ber, type
a real tim e scale: the com puter need and the nom enclature of the group to
not w ait for m an, and m an need not which it belongs.
w ait for the com puter. The charactron was developed in
The charactron is also used in some 1941. In electronic com puters it was
control system s. In these cases it employed la te r —in 1953—after m any
is u sually connected to a kines im provem ents were m ade. I t won ac
cope. ceptance because of its positive qu ali
In such a charactron-kinescope sys ties, valuable “tra its of c h a ra c te r” ,
tem the kinescope screen produces the among which the specialists u n an i
image of the object, and the charac m ously acclaim the excellent image
tron screen the “su b sta n tia liz e d ” signs q u a lity , the v isib ility of inform ation,
in the form of cards. These serve as the high speed of data selection. It
a kind of notes. For exam ple, while is these q u alities th a t m ake charac
the kinescope screen shows some ob tron one of the principal m eans of
ject, the charactron screen shows its “inform ation rep resen tatio n ” .
w O n ly D o t s in E v e r y L in e __
T his is the lim it, and i t c a n ’t be Five signs give us enough com bi
helped: 22= 4 . nations: 26= 32.
Now, l e t ’s try w ith three signs: H ere’s an exam ple of such a code:
A =000 E = 100 A =00000 1=01000
B =001 F = 101 B =00001 J =01001
C = 010 G =110 C=00010 K = 01010
D =011 H = lll D =00011 L ^ O lO ll
E =00100 M =01100
T his is again the lim it: 2S= 8 . F =00101 N =01101
If we add one more sign, w e’ll get
24= 1 6 . Sixteen com binations is n ’t G =00110 0=01110
enough for our ABC. H = 00111 P= 01111
WORD 299
A Babylonian cuneiform inscription depicts a square with a diagonal. Its side is equal to 30,
and this is written in the left upper part. The approximate value of ]/2 is written alongside the
diagonal.
304 CYBERNETICS A TO Z
The eminent mathematician and astronomer Ulugh Beg calculated the ratio of the circumfe
rence to the radius, i.e. the number 2r., to sixteen signs after decimal point. Gradually increas
ing the number of the sides of the regular polygon to approximate the circumference he obtained
a polygon of 800355168 sides!
in the first half of the 2nd century A.D. and left behind 13 volumes entitled The
Great Collection, or the Great Construction. This was a kind of contem porary en
cyclopaedia of astronom ical knowledge. I t has reached us under the Greek-Arab
title of Almagest.
Almagest contains the results of enormous calculation work performed by P to
lem y. They were presented in the form of sine tables and were intended to help
astronomers in their work.
The tables could be used to find the sines of arcs up to 90° in increm ents of a
quarter of a degree. The Pythagoras theorem enabled any elem ent (a side or an
angle) of a rectangular triangle to be calculated from two known elem ents.
The Ptolem y table is the first trigonom etric table to reach us.
The prolonged period of R om an suprem acy in Europe was not m arked by any
conspicuous achievem ent in m athem atics, but for the Greek scientist Diophan-
tus (3rd century A.D.) who introduced into algebra some original algebraic
equations. The invention of the logarithm proved of param ount im portance
to calculation techniques. I t affected the entire m ethodology of m athem atical
problem solution. People fam iliar w ith the use of logarithm s from their school
days can h ardly im agine the m arvel and excitem ent caused by th eir appear
ance.
The great scientist P . Laplace wrote: “The invention of the logarithm , by
curtailing tfye calculations of several m onths to a m atter of several days, seems
to double the life of an astronom er.”
X, Y, Z—CALCULATION MATHEMATICS 305
It is shown above that the more simple operations with numbers performed in the arithmetical
progression correspond to those performed in the geometrical progression.
306 CYBERNETICS A TO Z
Real working logarithm tables for accurate calculations were first compiled
by John N apier. It took this prom inent Scotch m athem atician tw enty years
to complete the job. N apier explains the reasons for his work as follows:
“I have tried to do everything in m y power to rid myself of the difficulty
and ennui of calculations the tediousness of which usually scares a lot of people
away from studying m athem atics.”
The work of N apier is of outstanding im portance because he was the first
to divulge the essence of the logarithm , a m athem atical q u an tity h itherto u n
known.
The study of dynamic processes often in Mathematics was able to sense the limit
volves the solution of extremely complex when rectangular steps turn into a smooth
differential equations. curve. It found the lim it of an infinite
sum (S ) when the number of the additives
is increased to infinity while each of them
tends to zero. This limit is termed integral.
20* /
308 CYBERNETICS A TO Z
The value of this discovery for the m athem atics of calculations was as great
as th a t of the discovery of the trigonometric functions.
Three hundred and fifty years passed since Napier made his ingenious inven
tion. D uring this tim e over 500 types of logarithm tables were devised. They took
a firm place in the arsenal of calculation im plem ents and to this day occupy a
position of honour as m a n ’s aid in calculations.
Millions of specialists use logarithms in their everyday work, from the highly
accurate twenty-plus sign tables to the “wooden” logarithms, slide rules, indispen
sable instrum ents for technical calculations.
People always lived in a world of incessant motion and continue to do so now.
Everything around us is in motion: the m utual position of the planets, of the
solar system, air pressure and tem perature, forces active in a machine, currents
flowing in an electric circuit, the state of a living cell all change in the course
of time.
The speeds of other processes are enormous, and their duration quite short: some
hundredths, thousandths or m illionths of a second. Swift is the drop in air pres
sure in a region of an approaching cyclone. Colossal is the acceleration of ele
m entary particles in a synchrotron: it takes them seconds to cover cosmic dis
tances.
W hen studying these so-called dynamic processes scientists are primarily inte
rested in changes taking place in time.
To analyse and calculate these processes m athem atical methods are needed which
would be able, like an ultra-high-speed camera, to sense changes taking place
w ithin the m inutest time intervals.
The part of such a “camera” in calculations came to be played by the methods
Mathematicians made a startling discovery: the regularities governing random quantities are
themselves not fortuitous. This makes it possible to assess the frequencies of various deviations
from the point of aim.
X, Y, Z—CALCULATION MATHEMATICS 309
of the differential and integral calculus, the science of varying quantities. It was
created by the m athem atical genii Leibnitz, Newton, Euler and their disciples.
But it is one thing to study phenomena as regular as the change from day to
night, or the alternation of the year’s seasons, and quite another to investigate
processes liable to be affected by various fortuitous circumstances.
A coin throw n into the air is bound to fall to the ground. Even the tim e it takes
it to do so m ay be calculated.
But no calculation w ill be able to predict which side down it w ill fall: heads
or tails. I t m ay fall any way, this being dependent on numerous fortuitous cir
cumstances.
Such em inent m athem aticians as Pascal and Ferm at evinced interest in games
the outcome of which was based on random events. A new branch of hum an know
ledge came in to being, the theory of probability. I t was proved th a t probability
is a q u a n tity which m ay be m easured.
B ut it was only in the 19th century th a t thanks to the endeavours of such m athe
m aticians as Gauss, Chebyshev, M arkov, Lyapunov, the theory of probability
developed into a separate branch of science, a branch of m athem atics the practi
cal im portance of which became stupendous.
Many em inent scientists spent a lot c u latio n of various m athem atical pro
of tim e on tedious calculations try in g blem s, assesses the com plexity, the
to discover new roads in m athem atics cost in labour and the accuracy of the
and to check th e ir calculations. algorithm s and of the m ethods of
M athem aticians-calculators do not com piling com puter program m es for
build m achines or houses, there are them . This is a vast task , for some a l
no scales, test-tubes, galvanom eters, gorithm s contain up to a b illio n a rith
or microscopes in th eir studies. They m etical operations.
do not conduct experim ents. B ut m a The m athem atics of calculations
them aticians arm the m achine desig helps not only its wards: m echanics,
ner and the arch itect, the physicist physics and astronom y, but also such,
and the chem ist, the biologist and the apparently, d ista n t disciplines as geo
economist w ith the sk ill to solve pro logy, m edicine, linguistics and eco
blem s, w ith the m ost up-to-date m a nomics.
them atical m ethods. H ere, too, m athem atics learned how
Together, wre witnessed the tim e to reduce the problem to a “num
calculation m athem atics left its b e r” .
cradle. C alculation m athem atics grew far
H aving been born to satisfy prac and wide. I t needs a great m any spe
tic a l requirem ents it itself is now in cialists.
fluencing theoretical developm ents, The m ost a u th o rita tiv e m athem a
dem anding of the theory new, more tic ia n s are of the opinion th a t now
efficient m ethods differing in p rincip is the tim e to introduce the elem ents
le from the previous ones. of calculation m athem atics and com
Today calculation m athem atics p u ter technology in to secondary-
evolves rules for the num erical cal school program m es.
310 CYBERNETICS A TO Z
“They are waiting- for you, young to build machines that w ill liberate
m en,” says Academician Vinogradov, man’s creative forces from rough phy
addressing himself to young people. sical and mental work. This is a job
“It w ill be up to you to make use of of the next decades, and you should
mathematics to lay courses to distant be ready for i t . ”
planets and, possibly, to the stars,
To the reader