Computer Analysis a Waste of Time?
5 posters
International Chess School Forum :: International Chess School Discussion :: Main Course: General Discussion
Page 1 of 1
Computer Analysis a Waste of Time?
When I looked for the threats in Test Problem 2 (I won't spoil the problem with details here) I chose a threat that was different than the one given. Zero points for me on that section! When I ran the problem for Black to move with Rybka lo and behold Rybka game my line 30 centipawns advantage over the book answer and at a 20 ply depth.
Still, I think that the book answer is better despite what Rybka is showing. For strategy analysis it appears that computer analysis has limited value beyond verifying that there are no tactial oversights. The authors state that if you have a material imbalance of about 1 pawn you don't bother to do a positional evaluation. I suppose that means that we shouldn't put much stock in small delta numeric chess engine valuations either.
Still, I think that the book answer is better despite what Rybka is showing. For strategy analysis it appears that computer analysis has limited value beyond verifying that there are no tactial oversights. The authors state that if you have a material imbalance of about 1 pawn you don't bother to do a positional evaluation. I suppose that means that we shouldn't put much stock in small delta numeric chess engine valuations either.
PawnCustodian- International Master
- Posts : 453
Join date : 2010-08-05
Re: Computer Analysis a Waste of Time?
I think your attitude is good: if it's within a pawn especially. Especially common is that a position a computer could easily win would be very hard for a human, because of the perfect and extended ability of the computer to see so many moves ahead. For instance, computers don't like to simplify an ending, but go for mate in 20. I prefer to trade queens and get an easy pawn endgame with few complications.
We had an extended argument/debate/fight about this topic at this thread:
https://icsdiscussion.forumotion.net/general-chess-discussion-f4/new-chess-improvement-plans-t162.htm
We had an extended argument/debate/fight about this topic at this thread:
https://icsdiscussion.forumotion.net/general-chess-discussion-f4/new-chess-improvement-plans-t162.htm
Re: Computer Analysis a Waste of Time?
Thanks BDK, a very informative thread.
I should explain that I know that RYBKA has outstanding positional understanding, after all Larry Kaufman who did the chess "understanding" in RYBKA wrote the book on material valuation. However, I do not accept the proposition that there is a "best" move in all situations - tablebases excepted. Certainly the three moves that RYBKA suggested ahead of the book solution are all acceptable moves, the big difference is that the book solution has been explained to us by a GM.
I should explain that I know that RYBKA has outstanding positional understanding, after all Larry Kaufman who did the chess "understanding" in RYBKA wrote the book on material valuation. However, I do not accept the proposition that there is a "best" move in all situations - tablebases excepted. Certainly the three moves that RYBKA suggested ahead of the book solution are all acceptable moves, the big difference is that the book solution has been explained to us by a GM.
PawnCustodian- International Master
- Posts : 453
Join date : 2010-08-05
Re: Computer Analysis a Waste of Time?
Oh, and a link to another discussion here that we had:
https://icsdiscussion.forumotion.net/ics-month-one-f3/question-about-pme-position-t136.htm?sid=e3328050a0f1797a727151fca25d22c6
https://icsdiscussion.forumotion.net/ics-month-one-f3/question-about-pme-position-t136.htm?sid=e3328050a0f1797a727151fca25d22c6
Re: Computer Analysis a Waste of Time?
Still i don't know your level PawnCustodian, but why worrying about +/-0.20 moves when you can positionnally blunder anytime and make -1.50 moves or even worst.
You should use primarily computer as an oracle (telling that a move is good or a move is bad, or better & worst), and evaluate your play to find such positionnal mistakes and see why you did them along with your strategy plan.
Only when you are truly confident with your play you can start discussing 0.20 points variations... In the meantime just apply ICS strategy ideas, it does not garantee always the "best" move but an average always good move, which helps print good reflexes in your mind.
It's more easy to remember a good move which you statistically see often, than a tricky crushing move that you see once in a game.
You should use primarily computer as an oracle (telling that a move is good or a move is bad, or better & worst), and evaluate your play to find such positionnal mistakes and see why you did them along with your strategy plan.
Only when you are truly confident with your play you can start discussing 0.20 points variations... In the meantime just apply ICS strategy ideas, it does not garantee always the "best" move but an average always good move, which helps print good reflexes in your mind.
It's more easy to remember a good move which you statistically see often, than a tricky crushing move that you see once in a game.
HangingKing- International Master
- Posts : 371
Join date : 2009-04-21
Re: Computer Analysis a Waste of Time?
Thanks HangingKing, I agree with you on this. My point is I don't think that there is any merit to discussion on numbers since there is really no way to know what they are based on.
Anyway, as to my level, I guess I'm guilty of not introducing myself at the start. I am returning to chess after years of inactivity. I left with a 1600 USCF OTB rating and played correspondence for a couple of years reaching about 1800 postal. To get back in shape I used the Conveka modules and have been carrying a 2230 ELO with CT-Art modules and about 2050 with the middlegame and endgame modules for the past several months.
The ratings are not really meaningful predictors of OTB play in my opinion - there's no substitute for games, recent games! In my initial outing I did abysmal job by failing to manage my clock leaving several won positions on the board. The lost positions I left on the board were both the result of poor positional judgement, one game I simply mishandled a dynamic pawn center, and the other game I made an unsound positional sacrifice and ran out of time trying to recover.
Another of the problems I'm trying to correct is that all of that puzzle solving has me searching for kills that aren't there. I'm hopeful that the ICS training on thinking and an active play schedule will correct those deficiencies!
Anyway, as to my level, I guess I'm guilty of not introducing myself at the start. I am returning to chess after years of inactivity. I left with a 1600 USCF OTB rating and played correspondence for a couple of years reaching about 1800 postal. To get back in shape I used the Conveka modules and have been carrying a 2230 ELO with CT-Art modules and about 2050 with the middlegame and endgame modules for the past several months.
The ratings are not really meaningful predictors of OTB play in my opinion - there's no substitute for games, recent games! In my initial outing I did abysmal job by failing to manage my clock leaving several won positions on the board. The lost positions I left on the board were both the result of poor positional judgement, one game I simply mishandled a dynamic pawn center, and the other game I made an unsound positional sacrifice and ran out of time trying to recover.
Another of the problems I'm trying to correct is that all of that puzzle solving has me searching for kills that aren't there. I'm hopeful that the ICS training on thinking and an active play schedule will correct those deficiencies!
PawnCustodian- International Master
- Posts : 453
Join date : 2010-08-05
Re: Computer Analysis a Waste of Time?
Convekta tactical software estimates are a really bad predictor of OTB performance, IMO. The best way to stay in playing shape is to play slow games (imo).
Re: Computer Analysis a Waste of Time?
BOY does that sound familiar!PawnCustodian wrote:I am returning to chess after years of inactivity. I left with a 1600 USCF OTB rating and played correspondence for a couple of years reaching about 1800 postal. To get back in shape I used the Conveka modules and have been carrying a 2230 ELO with CT-Art modules and about 2050 with the middlegame and endgame modules for the past several months.....Another of the problems I'm trying to correct is that all of that puzzle solving has me searching for kills that aren't there. I'm hopeful that the ICS training on thinking and an active play schedule will correct those deficiencies!
BorgQueen- Grandmaster
- Posts : 690
Join date : 2010-07-06
Location : Adelaide
Re: Computer Analysis a Waste of Time?
Once again, I am with you on this BDK, the software estimates are at best an ego boost only.
Blue Devil Knight wrote:Convekta tactical software estimates are a really bad predictor of OTB performance, IMO. The best way to stay in playing shape is to play slow games (imo).
Chess?- National Master
- Posts : 198
Join date : 2009-04-14
Location : canada, west coast
Re: Computer Analysis a Waste of Time?
Chess? wrote:Once again, I am with you on this BDK, the software estimates are at best an ego boost only.
Well, I suppose it sells software too...
PawnCustodian- International Master
- Posts : 453
Join date : 2010-08-05
Re: Computer Analysis a Waste of Time?
Yes, but the only problem with that is that it really hurts your clock. I've played lots of online correspondence lately and although I play reasonably well, given TIME, my clock suffers badly and once in time trouble, my game falls apart. Focusing too much on 'slow games' has this downside.Blue Devil Knight wrote:Convekta tactical software estimates are a really bad predictor of OTB performance, IMO. The best way to stay in playing shape is to play slow games (imo).
BorgQueen- Grandmaster
- Posts : 690
Join date : 2010-07-06
Location : Adelaide
Re: Computer Analysis a Waste of Time?
BorgQueen wrote:Yes, but the only problem with that is that it really hurts your clock. I've played lots of online correspondence lately and although I play reasonably well, given TIME, my clock suffers badly and once in time trouble, my game falls apart. Focusing too much on 'slow games' has this downside.Blue Devil Knight wrote:Convekta tactical software estimates are a really bad predictor of OTB performance, IMO. The best way to stay in playing shape is to play slow games (imo).
not if you only play slow games!
Chess?- National Master
- Posts : 198
Join date : 2009-04-14
Location : canada, west coast
Re: Computer Analysis a Waste of Time?
Define "slow games". I went from playing 3 days per move games to 60+30 otb games and my clock was a big problem!
BorgQueen- Grandmaster
- Posts : 690
Join date : 2010-07-06
Location : Adelaide
Re: Computer Analysis a Waste of Time?
i would call 60+30 games long.
Chess?- National Master
- Posts : 198
Join date : 2009-04-14
Location : canada, west coast
Re: Computer Analysis a Waste of Time?
Not long enough after you get used to 3 days per move!!!
BorgQueen- Grandmaster
- Posts : 690
Join date : 2010-07-06
Location : Adelaide
Re: Computer Analysis a Waste of Time?
I guess it depends on your goal. If your goal is to do well in OTB tournaments at long time controls, then practicing in such conditions is really important. This will then push you to manage time better, an important part of the equation!
While it doesn't help my chess, I got a lot better at time management by playing blitz games online with the same increment as in the OTB tournaments I play in (that is, 5 second increment).
I also do the same with friends in person, as that showed me that it is actually possible to make a move within five seconds, and have time left over.
Just don't get addicted to such blitz play, it is a thought killer.
While it doesn't help my chess, I got a lot better at time management by playing blitz games online with the same increment as in the OTB tournaments I play in (that is, 5 second increment).
I also do the same with friends in person, as that showed me that it is actually possible to make a move within five seconds, and have time left over.
Just don't get addicted to such blitz play, it is a thought killer.
Re: Computer Analysis a Waste of Time?
Agreed. I made the mistake of playing the 3 days per move stuff for a long time and then returning to OTB chess. As I already said, my clock suffered badly as I was soooo used to sitting for an hour or three on one single move.
I am now doing much less 3dpm games and trying to get in more blitz and "rapid" (20 mins or so) to speed up my thought process. It's working... slowly
I am now doing much less 3dpm games and trying to get in more blitz and "rapid" (20 mins or so) to speed up my thought process. It's working... slowly
BorgQueen- Grandmaster
- Posts : 690
Join date : 2010-07-06
Location : Adelaide
Re: Computer Analysis a Waste of Time?
BTW, yes, my goal is to get as good as I can with OTB 60+30 games.
BorgQueen- Grandmaster
- Posts : 690
Join date : 2010-07-06
Location : Adelaide
Similar topics
» Time to be controversial?
» How Best To Budget Time With ICS
» I am starting month four today.
» Why pdf's instead of browser based? It's too time-taking
» How Best To Budget Time With ICS
» I am starting month four today.
» Why pdf's instead of browser based? It's too time-taking
International Chess School Forum :: International Chess School Discussion :: Main Course: General Discussion
Page 1 of 1
Permissions in this forum:
You cannot reply to topics in this forum