10/04/2019

What does chess tell us about the future of humanity?

O russo Garry Kasprov é derrotado pelo Deep Blue, computador da IBM, durante jogo de xadrez de exibição pública / Divulgação

Reproduction of Marcelo Viana's column in Folha de S.Paulo.

In college, I spent hours in the library. Among other gems, I found the little book "Computers, Chess and Long-Term Planning" by the Soviet engineer Mikhail Botvinnik (1911-1995), published in 1970.

Botvinnik, world chess champion between 1948 and 1963, was a celebrity in my circle. The fact that he had written about how he played and how his decision-making processes could be replicated in a machine was an incredible discovery for a young person learning the rudiments of programming. I decided I was going to implement his method!

I knew that others must have had the same idea, starting with the pioneers of computing in the Soviet Union. And that, if they hadn't succeeded, perhaps my chances weren't good. I went ahead anyway: I failed, of course, and I was happy trying.

Botvinnik proposed "selective searches": the programmer chooses a formula to evaluate each position on the board, and the machine tests the movements of the pieces (at the time, they could test four moves ahead), abandoning the search when the evaluation becomes poor.

A few years later, the machines were already more powerful. Selective searches were replaced by brute computing power. The primitive chess machines of the 1980s became more competitive. But we "knew" that humans would maintain primacy; after all, no machine can play better than its programmer, right?

Until it happened. In 1997, IBM's Deep Blue computer, capable of evaluating 200 million positions per second, beat world champion Garry Kasparov. It was a shock, but it didn't prove that computers had become more intelligent than humans.

To read the full text, visit the newspaper's website or check the print version.

Folha allows each reader access to ten articles per month even without a subscription.