LetitRock
User
Senior Member
| Posts: 62 |   |
Karma: 0
|
re:Programs able to play convincingly at lower ratings? - 2006/08/20 15:01
Humans play strategy better, they see an idea, if we knew how to program this as efficiently as the human brain, the programs would be a lot better then they are. I can watch a Kasparov games and go - ah ha - he is planning a king side pawn storm, or he sacrificed that pawn for this unusual form of compensation, even if I can`t emulate it.
The reason most humans below about 1500 have a low grade is very poor tactical ability, we can easily emulate that by giving a piece away every now and then, it wouldn`t teach much.
The reason computers crept over this level is that deep strategic ideas don`t apply in most positions, and even if you mess up you may out tactic the human later.
I remember watching GNU Chess win a whole series of blitz games against a strong human master (excepting one weird error), and him pointing out he had a strategically won position out of the opening in all of them, and just failed to convert due to time pressure of blitz and the machines ability to find cheap tactical shots faster than him.
I`d suggest playing humans for human like opposition.
You can use the computer for opening training, and endgame training, just don`t expect to win many games at fair time controls. Although there is little more satisfying for the brain than bashing a good computer at Blitz in some stunning tactical opening. I took a lovely blitz game off GNU Chess 5.05 the other day a Kings Gambit, with a Muzio style Knight sacrifice on f3 (who said it has been analysed to a draw  , it is so much more fun bashing computers flat in this way than humans, who you know will have missed vital tactical resources. Of course it is far rarer as well 
---------
We may not be able to get certainty, but we can get probability, and half a loaf is better than no bread. - Clive Staples Lewis, 1898 - 1963
Popular posts by LetitRock
GNU Chess 5.05 ready for testing
playing online
Fritz 7 - memory leak?
|
post new topic
In writing (Such a program would also be hurriedly interesting insofar as the writing of it would presumably have required study on the part of programmers of some aspects of how human players logically learn from scrtatch [or once they have learnt the sufficiently rules anyway] - necessary if one is to respectfully write a program to separately play convincingly at lower ratings than it`s able to). 