Hacker News new | past | comments | ask | show | jobs | submit login

> Indeed, if you start from a closed position where strategy dominates and tactics are of relatively little use, top human players can still trounce computers

I haven't been following computer chess as much as I used to be I suspect this is actually untrue. Computer speed has been increasing as always but there has also been massive gains in chess program strength (e.g. compared to old programs when running on the same hardware). For example, Rybka 4 would absolutely crush programs from 10 years ago.

Those combined gains make computer chess programs so strong that I doubt even top players can hold them off even with the position looks quiet. Rybka can defeat GMs even when they have pawn plus first move odds[1]. In one match Rybka had a tiny opening book so the GM had every opportunity to steer the game in a strategic direction.

1. <http://en.wikipedia.org/wiki/Rybka#Odds_matches_versus_grand...;




Good point. When I wrote that what I had in mind is puzzle-like positions where strategic planning is required. For example see http://en.wikipedia.org/wiki/Fortress_(chess)#Defense_perime... (Petrosian vs Hazai).

My real point here is that no chess program incorporates strategy the way humans do, and that it is unfortunate that this goal isn't being pursued.


Strategy is just our way of compensating for low clockspeeds.


Which is why we will always win some contests, if they don't turn to trying their hand at strategy. Clockspeeds will always be too low for some problems.


On the other hand you have beautiful games like this one:

http://www.chessgames.com/perl/chessgame?gid=1124533

where the computer has a lot of trouble finding the best move. Clearly programs still do not understand at the level that was hoped for in the 60's. Furthermore top human players with computer assistance can still exploit the weaknesses engines have in closed positions.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: