Uri wrote:The next version of Deep Hiarcs probably could and maybe even the next version of Deep Junior. Also the next version of Vitrovius could maybe too.
There are many powerful commerical chess engines on the market to choose from today and their playing strength becomes even better in each new generation.
I don't think we are that far off from perfect play actually. It probably would take another 600 Elo points from Houdini 3 before engines will start to play perfect chess.
Over the last 30 years, we have seen an increase in strength of the engines/hardware combination of close to 1000 Elo. Many of the weaknesses that we saw in the eighties are still there; they are very difficult to rectify. If positional knowledge could be added that eliminated a substantial number of those weaknesses, the consequences of those additions would be that an overall increase of more than 600 Elo would be realised.
You can't "start to play perfect chess" anymore than you can start to be dead. Correspondence chess GM Robin Smith, who is now, very regrettably, totally dead, once said that if an opponent relied entirely on Rybka, he could always beat him. Hardly surprising, because he was adding GM knowledge, technique and judgment to the attributes of Rybka. While I wouldn't claim that his input was adding 600 Elo to the Rybka output, it was adding a considerable amount and the combined output of the two was still a long way from perfect.
Incidentally, a few days ago, I saw the groundless and fanciful claim that an engine would have to reach 16000 Elo before it could play perfect chess.