Professional Documents
Culture Documents
Brazil's Computer Market Reserve
Brazil's Computer Market Reserve
Democracy, Authoritarianism,
and Ruptures
Ivan da Costa Marques
Federal University of Rio de Janeiro
In the 1970s and 1980s, Brazil had a special minicomputers and other computer artifacts
industrial and technological policy for com- successfully marketed by Brazilian firms.
puter manufacturing. Officially named the Comparisons between minicomputer systems
National Computer Policy (Polıtica Nacional marketed in the international market and
de Informa tica), also known as the market those marketed by Brazilian firms provide evi-
reserve policy. Today, more than 20 years past dence of how close the Brazilian teams came
its demise in 1990, the market reserve policy to those working in the developed world.3 A
is still considered a capital sin. Popular opin- few years before the policy’s condemnation,
ion argues that absolutely no good came Brazil became one of the few countries with a
from of the policy’s approach of regulating a significant part of its domestic minicomputer
market by means of restricting the origin of market supplied with locally designed prod-
technology. ucts and native trademarks.
In the context in the 1970s, the general The situation rapidly changed in the late
worldwide conditions of computer technol- 1980s, however. The appearance of the micro-
ogy, and the specific political conditions in computer deconstructed and restructured the
Brazil, the market reserve policy emanated computer industry worldwide. Brazilian mini-
from an interaction among academic research- computer manufacturers were not able to
ers, public officials, publically employed engi- hold their position. Most, if not all, aban-
neers, and military personnel. At its inception, doned the policy’s initial goals and the initial
the market reserve police focused on minicom- success turned to failure. Right after its demise
puters and stipulated that only locally designed in 1990, the general opinion in Brazil was that
minicomputers could be legally manufactured the computer industry policy was the main
and sold in Brazil.1 cause of most of the Brazilian industry’s handi-
Early on, the market reserve policy had a caps, such as the technological gap between
considerable degree of success in attaining Brazilian and foreign made products. Even
its goal. In the early 1980s, Brazil was one of today, politicians and journalists cite the mar-
the few countries that had its own minicom- ket reserve policy as an example of a “crime”
puter industry selling locally designed prod- to be abhorred. Unfortunately, easy explana-
ucts carrying Brazilian brand names.2 Teams tions such as “it filled the computer market
of Brazilian engineers and technicians had with technically obsolete and high priced
absorbed foreign technologies initially licensed products” or “there was American pressure
in the late 1970s and independently conceived against the market reserve police” justifying
and designed new hardware and software for today’s continuing intense rejection of the
2 IEEE Annals of the History of Computing Published by the IEEE Computer Society 1058-6180/15/$31.00 c 2015 IEEE
market reserve practiced in Brazil in the 1970s infrastructure of postgraduate education and
and 1980s preclude attempts to better under- research in engineering seemed an adequate
stand the policy as a historical experiment. starting point. The military dictatorship pro-
These arguments are sometimes followed by vided funds for the universities to create and
even more generic references to the final expand postgraduate schools of engineering.
defeat of the very concept of a planned indus- The National Development Bank (BNDE) cre-
trial policy, eventually overcome by a coali- ated a special agency to support this expan-
tion of forces articulated in the neoliberal sion, FINEP (Financiadora de Projetos), by
wave of the 1980s. They occasionally take detaching a specific amount of capital to
note of what would be a striking alliance nur- finance Brazil’s new attempt to become an
tured in the “repugnant archaism” of the active player in the game of constructing sci-
military dictatorship between parts of the left, entific and technological knowledge.
deceitful entrepreneurs, and the nationalist Several universities, including the Catholic
right.4 Generalized conceptual explanations University of Rio de Janeiro (PUC-RJ), Federal
do not lack importance, but given their ubiq- University of Minas Gerais (UFMG), Federal
uity, they lose almost all specificity and hence University of Rio de Janeiro (UFRJ), Federal
obstruct opportunities to learn from a more University of Rio Grande do Sul (UFRGS), and
detailed explanation of what happened. University of Sa~o Paulo (USP), established lab-
This article offers a new narrative that takes oratories and courses on digital techniques
a closer look at the success and failure of and computer programming, paying atten-
the market reserve policy.5 In doing so, we tion to the emerging sector of informatics.
explore three specific sociotechnical develop- FINEP and BNDE sponsored applied research
ments: the special character of the commun- projects at those and other universities, where
ity of Brazilian computer professionals, the professors and students tried to discover the
intervention of the political police of the mili- workings of the black boxes of the imported
tary dictatorship, and the appearance of the computers. Other government agencies, among
microcomputer. Looking at these three proc- them the Ministry of Finance’s Bureau of Data
esses, amid technical and political changes Processing (SERPRO) and Navy’s Research Insti-
within and outside Brazil throughout the tute (IPqM), also started research projects aimed
1970s and 1980s, leads to a new perspective at discovering and trying to reproduce locally
that runs against widespread opinion that imported computer products and systems that
simplistically binds the roots of the market Brazilian engineers hitherto did not know how
reserve to the authoritarian character of the to make.
military regime, allowing for a more complex It is noteworthy that in the 1960s and
explanation. 1970s a technologically differentiated percep-
tion of the computer sector was not just hap-
A Brazilian Community for Informatics pening in Brazil. Computers were extremely
in the 1970s expensive machines, unknown to the general
In the late 1960s and early 1970s the military public. With the exception of the US, the
that ruled Brazil and its technical and intel- number of people directly interested in com-
lectual bureaucracy envisioned a “Brazilian puters was diminutive: hundreds to a few
economic miracle” that positioned the coun- thousand in the majority of the countries.
try on the head of a runway ready to take off Nonetheless, mastering computer techniques
for the sort of long range flights ventured by was the object of numerous studies, discus-
world powers. Some among them claimed sions, and actions in several countries. Mas-
that the Brazilian military’s dream to become tering computer technology was regarded
a world power could not possibly come true worldwide as necessary, unless a nation were
without the articulation of a minimum to remain dependent on foreign knowledge.
autonomous technological capacity. On the Those were the days of the short-lived IBI
one hand, the Brazilian military despised and (International Bureau of Informatics), created
repressed the collective of university profes- within the United Nations. It was a time
sors when they criticized authoritarianism when many developing, and a few developed
and social injustices. On the other hand, the countries, especially France, tried to deal
Brazilian dictatorship would bet its chips on multilaterally with computer technology
technological development. Given the by that was then already perceived as an emerg-
then already clear importance of computers ing phenomenon that would bring about
and the absence of Brazilian-controlled com- new and highly differentiated forms of pro-
puter manufacturers, the creation of a local duction and consumption. Charles de Gaulle
October–December 2015 3
ordered the famous Nora-Minc Report to sub- what kind of goals were legitimate for tech-
stantiate the discussion and implementation nological research in the computer field in
of a program for the development of infor- Brazil. The dominant idea was that Brazilians
matics in France, which led to the Plan Calcul should make a strategic investment to over-
and the Minitel system. The success of 16 come technological dependence. If Brazil
years of an intricate network of protection of became more and more dependent on com-
and incentives to Japanese computer firms puters and Brazilians did not know how to
had become visible. Those were times of effer- make them, then the country would find
vescent political ideas and actions in the itself paying for computers, whatever the
computer field. Among peripheral countries, price exacted by those few who knew how to
South Korea and India also made efforts to make them.
carry out computer industry policies.6 In the mid 1970s, the military dictatorship
Given this international environment, it is minted and circulated the term “relative
perhaps not surprising that, notwithstanding democracy.” The idea was to try to harness
the authoritarianism imbued in the military democratic forms rather than simply elimi-
regime, Brazil witnessed the appearance of a nate them. Relative democracy meant that
critically minded community of computer pro- people could talk, write, and act on an
fessionals as a result and cause of the invest- expanded set of subjects, but they were not
ments on research infrastructure. The sui granted the broader rights of free speech or
generis character of this community did not free assembly.
take for granted the result of the usual The Brazilian community for informatics
“depuration of phenomena,” which estab- was in a good position to take advantage of
lishes an a priori separation between the this idea of a relative democracy. The military
political and technical realms.7 Its members did not mingle much with the researchers and
assumed to be in charge of two realms simulta- managers at universities and other public
neously: (politically) debating what should institutions (state-owned data processing cen-
constitute legitimate objects of research and ters), and the community of computer profes-
(technically) developing a limited, but signifi- sionals acquired and possessed the means to
cant local capacity to conceive, design, and make their voices heard, even if not very loud.
integrate computer products. In the laborato- These means were a common specialized press
ries, professors, graduate students, and resea- such as Datanews and the journal DADOS e
rchers designed diverse computer products Ideias, in addition to joint conferences and
such as modems, video terminals, intelligent seminars such as SECOMU and SECOP.
terminals (precursors of microcomputers), spe- Through their interactions, academic
cial-purpose processors, compilers, and com- engineers, managers, and probably to a lesser
munication protocols. Of course, Brazilian extent, the military realized that, notwith-
professionals did not invent these products. standing profound differences, their percep-
Nevertheless, they reverse engineered some of tions and analyses of technological problems
the newest pieces of computer machinery in Brazil coincided on many points. The three
available for sale in the international market. groups shared the idea that computer tech-
The empirical research at universities and labo- nology was a strategic issue for Brazil, and
ratories consisted of learning how to make each group was able to translate “mastering
these mysterious products, with the goal of computer technology is a national goal” into
acquiring the capacity to design computers by its own more specific interests. Professors at
teams of Brazilian professionals. According to the then recently established postgraduate
this conception of research, which one might schools of engineering were concerned about
today say “situates” knowledge in a territory, creating a local labor market for their
Brazilian researchers “discovered” how to con- students. Their claims for more money to
ceive of and design products.8 enhance labs and infrastructure would remain
This empirical and conceptual seeking of a legitimate only to the extent that a market
limited, but significant technical capacity to demand for professionals able to conceive of
conceive of and design computer products and design computers would materialize in
cannot be separated from a sort of friendly Brazil. In a similar vein, engineers and manag-
political activism facing the military dictator- ers at state-owned data processing centers
ship. Both technical and political action showed that viable solutions to their specific
shaped a special character for that Brazilian problems would span much more broadly if
community of informatics—that is, a charac- they could count on local capacity to design
ter based on explicit cognitive debates on even small pieces of computer hardware and
October–December 2015 5
Figure 2. Front and back of the commercial specification of PPF. This industrialized version of the PPF
was made by the Brazilian firm Microlab.
“Brazilian community for informatics,” the skirmishes involving civil servants, univer-
current state of affairs led to a social-technical- sities, IBM, and the Brazilian government.13)
political-economic proposal, to introducing The market reserve policy for the develop-
an expedient in the market that made invest- ment of the minicomputer technology in
ment in local minicomputer system develop- Brazil was conceived in this sui generis com-
ment in Brazil more attractive. An ade- munity,9,10,13 which gathered in the 1970s a
quate expedient would be a market reserve broad spectrum of specialized professionals
for firms that would perform product R&D in from universities, state-owned data processing
Brazil. Civil servants at the Commission for centers, and the military. Specialized journal-
Improving Data Processing Activities (CAPRE), ists, politicians, debutante entrepreneurs, and
the government agency in charge of the task labor union leaders joined them by taking
of “rationalizing the use of computers in the advantage of the relative democracy. This more
federal government,” had been accompany- ample social community never acquired a uni-
ing and regularly nurturing discussions in the fied formal structure, but it effectively existed
community of informatics since the early and performed coordinated political action as
1970s. With the support of the community of its members attended seminars and conferen-
informatics, CAPRE implemented the mini- ces, read the same specialized journals, estab-
computer market reserve by issuing an inter- lished common goals, and mobilized to
national competition in 1977 to select firms to support their goals (see Figure 3). During the
manufacture and sell minicomputer systems 1970s, the character and the dynamics of the
in Brazil. One important condition was secur- market reserve minicomputer industry policy
ing a commitment from the winners to per- in Brazil depended on a kind of activism of this
form local product R&D. In the end, the more or less heterogeneous broad community
selected minicomputer manufacturers were of the 1970s that I refer to as the Brazilian com-
the Brazilian firms Edisa, Labo, Sid, and Sisco. munity for informatics.
Using legal import controls, CAPRE had the
means to effectively keep other firms out of First Rupture: The Political Police
the minicomputer market. (This is a somewhat Intervene
crude linearization of a much richer narrative. A turning point in the computer industry
Vera Dantas provides a detailed story with policy occurred in 1979. In the late 1970s,
many back and forth movements, including the political democratization of the country
October–December 2015 7
Some activists were able to see in that conclu- later on with undisguised distaste. CAPRE’s
sion a recommendation for increasing the journals were open to the community and
local production of computer technology—a aware of the opportunities of the relative
victory for the ideas for which the Brazilian democracy, whereas behind a surface-level
community for informatics had fought until technical focus, SEI’s journals sought to hide
the start of the commission. They nurtured an authoritarian mistrust of a community
mixed feelings about those who might possi- that had shown a capacity for collective tech-
bly represent new powerful alliances within nical and political action. One can say that in
the dictatorship’s repressive apparatus. The the early 1980s the market reserve policy dis-
Cotrim Commission proposed the extinction mantled its ties with its diverse and demo-
of the small CAPRE and substituted it with a cratic origins in civil society and part of the
new agency on the highest possible level of military, abandoning a Brazilian universe of
government bureaucracy. The proposed new possibilities and lodged itself in the closed
agency, Special Secretary for Informatics world of anachronistic authoritarianism.
(SEI, Secretaria Especial de Informa tica) was
promptly created as a part of the office of the Second Rupture: The Appearance of
president of the republic with ministerial sta- the Microcomputer
tus. SEI’s posts were immediately occupied by The second rupture came with the micro-
the commission’s colonels themselves and computer and in particular the dissemination
their close collaborators.16 By means of a ser- of the standardized platform for the personal
ies of successive substitutions, the colonels of microcomputer: the IBM PC. The microcom-
the Cotrim Commission remained in direct puter’s effects on the Brazilian computer
or indirect command of the computer indus- industry were not pronounced until a few
try policy from SEI’s creation in 1980 until years after its launch in the US in 1981. With
SEI’s much applauded abolishment in 1990. it the computer left behind the esoteric limits
Historians Emanuel Adler, Franz Stefano of its creation as an “electronic brain” that
Erber, Peter Evans, Hubert Schmitz, and Jose only large organizations might consider
Cassiolato all take notice of the SNI interven- using. Large mainframes were a specialized
tion, but in their analyses, they construct segment of the capital goods market, involv-
frameworks or narratives not of rupture but ing relatively few people—in Brazil only a few
of continuity. For instance, Evans argues, thousand people, possibly only a few hun-
“More surprising than the political demise of dred. For the great majority of people, large
the old tecnicos was the fact that their mainframe computers were esoteric
agenda proved politically robust. CAPRE’s machines kept in refrigerated aquariums.
ideological legacy lived on two quite distinct At this point, it is crucial to note that,
but politically complementary forms.”17 although the minicomputer systems that
They failed to see that the substitution of SEI appeared in the 1970s were much smaller in
for CAPRE represented an inflection point, size and price than the large computer main-
literally a rupture in the derivative function, frames, their market had the same characteris-
in the ethos of the minicomputer industry tics as their larger siblings. Of course, mini-
policy. The Cotrim Commission received computers greatly enhanced the diversity of
detailed handbook-level instructions on how products, suppliers, buyers, and applica-
to forge ahead the implementation of a tions. Nevertheless, considering how deci-
minicomputer (not microcomputer) market sions about buying and selling were made
reserve. They confiscated market surveys, def- and how procedures about use and mainte-
initions of priority for local design, drafts of nance conditions were established, we can
calls for international competitions for prod- see a marketing continuity. Table 1 shows
ucts for which there was a demand that could the results of a survey published by CAPRE
not be supplied with local design, drafts of in 1979 on all computers installed in Brazil
contracts and legal commitments of the listed by their size (there were no microcom-
winners to invest, and local designs for the puters).18 Although minicomputer systems’
following generation of products. Ironically, sizes and population differed substantially
as I shall argue, SEI did use them. from those of the mainframe computers,
However, while CAPRE nurtured the Brazil- their price, delivery times, and maintenance
ian community for informatics by investing conditions were still negotiated between
in its relations with computer professionals, specialized buyers and sellers for use in
SEI treated the community of computer pro- firms. Economically, this characterizes mini-
fessionals initially with symbolic violence and computers systems as typical capital goods.
October–December 2015 9
renewed through live discussions. They of the microcomputer in the early 1980s
claimed that the Brazilian government should required a renegotiation of the policy pro-
resort to the minicomputer market reserve to posed by the Brazilian community for infor-
decrease the technological dependence of matics and adopted/adapted by CAPRE.
Brazil. By 1980 the five selected Brazilian compa-
Second, among professionals, the Brazil- nies (Cobra, Edisa, Labo, Sid, and Sisco) had
ian community for informatics consubstan- successfully established themselves in the
tiated a kind of decentralized collective intel- reserved minicomputer market. Since 1977
ligence that followed up and discussed the they had honored their commitments to
implementation of the minicomputer market invest in local technological capacity to con-
reserve, constantly estimating government’s ceive of and design products. They had
and firm’s behaviors vis-a-vis their fit to the successfully launched enhancements and
objectives that, according to discussions in upgrades of the originally licensed models
the community, made the minicomputer and new models entirely developed by Brazil-
market reserve legitimate. In other words, the ian teams. After five years, they had become
community was also a source of a more speci- independent of the original foreign sources
alized, technical, or professional critique of of their technology. The policy worked just
the minicomputer market reserve policy. The fine for the minicomputer market, as envi-
community disseminated discussions that sioned in the 1970s.
created two poles, where support for what However, the market changed, and the
was considered right decisions and opposi- emerging microcomputer market had a very
tion to what was considered wrong decisions different structure. Microcomputers required
and behaviors led to a more or less consen- industrial and commercial organizations dif-
sual goal of diminishing technological ferent from those of minicomputers, more
dependence on the conception, design, and appropriate to a much more standardized and
fabrication of minicomputer systems. Thus, highly scaled technology, where issues of hard-
in the context of Brazil’s relative democracy ware manufacturing costs, software standard-
and high-tech industry constructs, the com- ization, and prices differentials became much
munity for informatics acquired, although more important and would have to be
insecurely, the character of a technical-politi- approached in new ways. Microcomputer
cal collective actor. technology required a revision of a series of
Table 1 shows that in 1976 the market previous decisions that had proved effective
reserve practiced by CAPRE affected only 11.3 for minicomputers, such as whether to make
percent of the computer market. At its incep- different subassemblies, parts, and compo-
tion, the computer market reserve issued spe- nents in Brazil or buy them from abroad.
cial rules for exploring a small, although First off, should there be a market reserve
increasing part of the market. The lion’s share for microcomputers, and what should be the
was that of larger machines, lying above the general architecture of a market reserve for
reserved segment, a point well stressed in microcomputers? What was the cost and gain
articles and discussions during the 1970s. This of a microcomputer market reserve? What
point did not escape those who have studied kind of technological commitment was it
the market reserve policy: “minicomputers’ appropriate for the government to require
technology was more accessible, investments from firms in exchange for a license to
were not as large as for the big computers, and explore the Brazilian microcomputer market?
there were no minicomputer manufacturers Because it was created in 1980, SEI was born
in the Brazilian domestic market.”19 The facing the strategic problem of deciding what
microcomputer market, however, came up kind of reformulation should be made to the
from below the reserved segment. Microcom- minicomputer market reserve policy, given
puters were complete computer systems that the radical changes already visible on the
were smaller and more integrated than mini- horizon with the emergence of the micro-
computers, and their market started to grow computer market.
at high rates. The first strong sign of market So what did the colonels of SNI do? They
discontinuities came with the 8-bit machines did not reformulate the policy, one they con-
using the CP/M operating system. After the sidered so “timid” as to be nonexistent. CAPRE
IBM PC, it was clear that the market reserve and Digibras were a source of handbook-level
would soon be sandwiched between the mar- instructions on how to implement and man-
ket for larger machines and the fast growing age a minicomputer market reserve that
new market. In other words, the emergence seemed to be working very well. It appears that
October–December 2015 11
Table 2. Revenue growth of the Brazilian computer industry.*
explanatory capacity because they ignore or taking advantage of the first opportunities in
deny the importance of local sociotechnical the process of political opening that the mili-
(political) specifics. Of course, changes in those tary dictatorship was in the end unable to
specifics are changes in the entities that those shun. It was only after the SNI’s intervention
numerical indicators try to represent or meas- that the Brazilian computer market reserve
ure. The strict economics behind those indica- policy diverged from the political opening of
tors, however, frequently fail to take into the country, isolating itself from the rest of
account these kinds of changes. In the case of society. Thus, while its democratic period is
the Brazilian computer sector, analytical frame- clearly associated to its success, its authoritar-
works based on aggregate sectorial data such as ian closure inexorably led it to its now
sales, number of employees, and international emblematic failure.
price comparisons have reinforced the impres-
sion of continuity obtained when SEI adopted
for microcomputers the same procedures that References and Notes
CAPRE had created and adopted for minicom- 1. Resolution 1 of the Plenary Council of CAPRE
puters. As a result, competent Brazilian and for- (the agency at the time in charge of
eign analysts have ignored evident ruptures “rationalizing the use of computers in the fed-
and attributed to the market reserve policy in eral government of Brazil”), published in the
the 1970s and 1980s a unity and a continuity Official Diary of the Union (D.O.U.) on the 15
that does not exist. The intervention of the July 1976. See DADOS e Id e ias, vol. 2, no. 1,
political police is especially ignored, which is Aug./Sept. 1976, p. 39.
all the more surprising because Vera Dantas 2. P.B. Tigre, Computadores brasileiros : indu stria,
has a whole chapter of anecdotes covering the tecnologia e depend^e ncia [Brazilian Computers:
works of the Cotrim Commission.13 The omis- Industry, Technology, and Dependence], Editora
sion seriously jeopardizes economists’ under- Campus, 1984.
standing the Brazilian experience in computer 3. Technical comparisons were based on processing
industry policy.21 speed (in millions of instructions per second),
The crucial point is that the analysis memory capacity (in RAM), ROM and disks (in kil-
offered here highlights that the Brazilian obytes or megabytes), and maximum number of
experience of a computer market reserve, terminals. Among the many accounts acknowl-
contrary to popular opinion in Brazil, was a edging this successful phase, see E. Adler, The
positive interaction between the democratic Power of Ideology: The Quest for Technological
principles and a project of technological Autonomy in Argentina and Brazil, Univ. of Califor-
development until 1979. The Brazilian com- nia Press, 1987; V. Dantas, Guerrilha tecnolo gica :
munity for informatics had become a hetero- ria da polıtica nacional de
a verdadeira histo
geneous technical and political collective informatica [Technological Guerrillas: The True
October–December 2015 13