Soviet chess and artificial neural networks: is there a link?

Some days ago, I was faced with the task of explain to a heterogeneous audience the mechanism of an artificial neural network called SOM (acronym for Self Organizing Maps). Translate algorithms, numbers and equations into words is not always easy, so I was trying to find analogies.

Well, one of the most important things about SOM is that it is a kind of competitive neural networks (whose neurons compete to be activated by a given numerical input pattern) in which competition is allied with something unusual in competitive contexts: cooperation! Once a neuron wins the competition, it shares its reward with its neighbors, so that they can better compete in the next cycle. The global result is that SOM performs better than all ordinary competitive networks.

Coming back to my talk, how could I explain that without being so technical, or boring?

So, I scrutinized my brain looking for some example, some case that could show an interaction of competition and cooperation in the real world and nothing was coming over… until… yes, how could I forget this? The soviet chess!

Soviet chess is a perfect analogy for SOM, and I wonder if Kohonen was aware of that in early 1980’s when he developed SOM.

Chess in Soviet Union (USSR) was taken seriously (maybe too seriously). It was something the politicians at the time liked to show as a proof of soviet’s mind superiority. Grandmasters were maintained by the government, and they were some kind of special citizens in the socialist regime, to whom some privileges were permitted.

The relative good life of a chess grandmaster in USSR was subject to some conditions, among others:

be a very good player;
be obedient to the regime (or at least neutral).

Naturally, only these two things aren’t enough to explain the success and hegemony of Soviet Chess in the long period between the 1950’s to early 1990’s (something only broken by Fischer in 1972). The great secret was COOPERATION.

To the masters it was allowed to have great animosity and rivalry against each other inside the country, but whenever they have to play abroad or to play in USSR against non soviets, they were obliged to share knowledge in conjoint training sections, to prepare together before the games, to analyze together when the games were adjourned and so on.

Between some of the best chess players in the world at the time, world champions included, no great secrets could exist, an improvement found one day, must be shared with others the next day. Is was a State Policy!

That’s why we saw so many world title’s contests played exclusively by soviets, champion and challenger. Only exception: Spassky (USSR) – Fischer (USA) in 1972.

Even after the fall of the USSR, the players from the old soviet school still show great predominance in chess. Until the moment, only three world title matches were played without any ‘former soviet bloc’ representative: Anand (India) – Topalov (Bulgaria) in 2010, Anand – Gelfand (Israel) in 2012 and Anand – Carlsen (Norway) to be played later this year.

It worked! People understood the message, and the remaining topics were much more easier to explain. Maybe competition and cooperation should walk together more often, as the recent (SOM) and past (Soviet Chess) success seem to show. If so, certainly it would be more easy to find a good analogy for my next talks.