1984, São Paulo: what Brazil's market reserve actually built

The sentence I am the evidence for

The catalogue article on Brazil's market reserve ends on a careful claim. The hardware bet failed: no globally competitive Brazilian computer manufacturer emerged, and the firms that survived did so by leaving for banking automation. But the human-capital bet, it argues, worked sideways — the policy trained a generation who had to build, adapt and reverse-engineer locally because phoning a vendor abroad was not an option. Its closing line is that people who went on to build Brazil's internet started on hardware the reserve created.

That sentence is written in the third person because the catalogue is written in the third person. Here it is in the first: I am one of those people, and the hardware in question arrived in my house in 1984.

The machine

It was a Microdigital TK-82C, a Brazilian clone of Sinclair's ZX81. My father loaded the first program from a cassette tape onto the family television. Soon after, I was writing BASIC on it myself. I was eight.

By world standards the machine was already old and it cost more than it should have. Both of those facts are in the article, and the article is right: Brazilian machines of that decade were largely clones of the ZX81, the Spectrum, the Apple II, the MSX standard and the IBM PC, sold late and expensive, obsolete on arrival and priced as though they were not.

None of that was visible to a child. What was visible was that there was a computer, in the house, and it did what you told it to. The policy's critics and its defenders have argued for forty years about whether the reserve was worth its cost. From inside a living room in São Paulo in 1984, the relevant fact was simpler: the machine existed and it was reachable. Without the reserve there would have been a different machine, or — far more likely for a family of that kind, in that country, at that moment — no machine at all.

What the policy could not supply

Here is the part I want to argue, because it is the part the industrial-policy debate always misses.

The reserve put the hardware in reach. It could not put the reason to care next to it. That came from somewhere else, and in my case it came from the man who loaded the cassette.

My father was a civil engineer and an amateur radio operator. He worked the local and long-distance bands, and the QSL postcards arrived from strangers a continent away to confirm a contact made on the airwaves. Think about what that is, in the vocabulary I would spend my working life using: a protocol for establishing a link with an unknown peer over an unreliable medium, and an acknowledgement, physically mailed, to prove the exchange happened. He had been doing packet-switching-by-correspondence for years before I understood what any of it meant.

So when a computer arrived in that house, it did not arrive as an appliance. It arrived in a household where it was already normal that machines talk to each other across distance, that you build the link yourself, and that the interesting part is not the box but what it can reach. My mother's mother, who had crossed an ocean from Germany and started her life over, had already taught me to look at the garden — the insects, the weeds, the flowers — as things worth taking seriously enough to ask questions about. Between the two of them, the machine landed on prepared ground.

An industrial policy can create a market for computers. It cannot create a father with a call sign. I got both, and only one of them was policy.

What the eight years looked like from inside

The reserve ran until 1992, which means it covered exactly the years in which I became technical. The pre-1996 record has the sequence: the cloned Sinclair in 1984, the ham bands through the late eighties, and then, from 1991, a bulletin board running on a single phone line, and a modem into a university's UNIX shell — a text prompt on a machine with no centre and no owner, which is the first time the world got bigger than the country.

By 1991 I was also importing components and assembling custom PCs to sell. That was the last year of the reserve, and it belongs to a pattern the article describes precisely: a parallel economy formed around the policy, and the people running the most current hardware in Brazil were frequently the ones going around it. I was a teenager building machines out of parts, which is a small and unremarkable version of the same phenomenon — the demand the policy created did not stop existing because the supply was restricted. It found other routes.

And there was the phone network, which is documented on the era page more candidly than most people would document it, including the detail that mattered: the corporate systems reached by war-dialling were left deliberately untouched. That distinction — between wanting to know how something works and wanting to take something from it — turned out to be the whole of the professional ethic that followed, and it was formed before there was any career to protect.

The World Wide Web arrived in 1993, through a program compiled on that same shell account that turned the dial-up session into a tunnel. My first formal job came in 1995. Cabletron in 1996. Everything after that is on the rest of the site.

Both halves, held at once

The way this subject usually goes wrong is that people pick a half. Either the reserve was a nationalist folly that set Brazil back a decade, or it was a visionary programme that built domestic capability. The article refuses to pick, and I think that refusal is the correct position rather than a diplomatic one.

The hardware bet failed on its own terms. Itautec, the last Brazilian firm still making computers, left that business in 2013. Nobody in São Paulo is running a domestic alternative to a foreign laptop today, which was the entire stated purpose.

The other bet paid out in a currency the policy was not measuring. It produced engineers — a lot of them — who learned on machines that had to be understood because there was no support line to call, and who then built the things that came next. Brazil's banking automation is genuinely world-class. Its internet governance was multistakeholder before the word existed. Its payment infrastructure has been ahead of most of the world twice. None of that was on the policy's scorecard, and all of it was built by people who had a computer in the eighties because of the policy.

I am a small instance of the same accounting. Nobody in Brasília in 1984 was trying to produce a network instructor who would spend three decades explaining these systems to people on four continents. That is not what the law was for. It is nonetheless one of the things the law did.

Why this matters beyond nostalgia

Because the same argument is live right now, in different words, everywhere. Every country currently debating semiconductor subsidies, sovereign cloud, domestic AI capacity or data localisation is having Brazil's argument again, and mostly without knowing that Brazil already ran the experiment for eight years and published the results.

The results are not "protectionism works" or "protectionism fails." They are stranger and more useful than either: the stated goal is frequently not what the policy produces, and the thing it does produce may be worth more than the thing it promised — while still not excusing the cost. Both halves. An expensive failure at building a computer industry, and the reason a generation of Brazilian engineers exists.

Policies are graded on their promises. They should occasionally be graded on their consequences, which take twenty years to appear and do not resemble the promise at all. I am one of the consequences. The machine that produced me was obsolete before it was bought, which turned out not to matter in the slightest.