1 edit
Originally posted by greenpawn34im with you on this one pawn dude, computers are mince, and just to prove it, i shall start a thread entitled 'why Kramnic was so crap', simply because he sat in front of his computer when he would have been better off playing blitz with the lads.
Typical of you lot.
Dodge the question because you don't like the answer.
So far you have compared Fritz12 to CM5000, Motor Cycles, girl tennis
players and now opening analysis of the 80's.
Beetle.
I can see by looking at games played today and games from the 80's
that there has been an improvement.
Here we have a program that none of us e
serious over the chessboard encounters.
FRITZ helps you to find friends......😕
5 edits
Originally posted by wormwoodActually, I think there's no point in getting fritz 12, because the improvements will probably be negligible.
compare against what? f10? how could you possibly know which one gave the best analysis? what do you compare it against?
engine vs. engine games can't solve the issue:
lets assume f11 was stronger than f12, in some abstract and absolute chess theoretical way. then lets assume that in certain restricted type of games, a subset of All Chess, f12 at his old harware can run it as well as the older fritz.
at least f11 goes to eleven.
however, this cannot be generalized in upgrading all chess engines. man I don't even want to respond GP anymore, but let's just say I find his point ("why upgrade if it beats you already" ) nonsense.
anyway, back to your question.
some examples:
1)there are many users waiting for Rybka to finally fix its endgame issues. If I didn't have fritz 11 along with rybka (which makes my problem go away), I would be much more pissed off about this, because the engine has real problems dealing with king&pawn endings.
2)Rybka also has a bad bishop issue. at the moment, it can jump over the problem because of its other parameters like mobility, piece activity of whatever they are, but it lacks the specific concept, and this makes it somewhat weaker in CERTAIN positions than CERTAIN other engines.
I noticed the 1st issue initially by myself, it wasn't rocket science, and I'm a patzer. the second issue was brought up in the rybka forums by stronger players, and I suppose Vas and Kaufman confirmed it.
humans don't have to be super GM strength to distinguish one engine from the other. I think correspondence players would have a lot to say about this topic. There are people who know which engine to use for which type of positions. Fritz 10 was known to be a much more tactical engine for example, and Fritz 11 was criticised to lose it's originality and have a style closer to Rybka.
Hiarcs is known to be brutal in kingside attacks. Deep Shredder is (was, before Rybka 3) known as the "endgame specialist". Many players still prefer shredder to analyze certain type of endings.
3)this you don't have to believe of course, and I don't know how to explain it really, but analyzing a game with Rybka 3 Human, jumping from one variation to another and exploring your ideas is a completely different world than analyzing with any other engine in my experience. maybe I can describe it as it's much closer to having a GM explain the positions to you, compared to any other engine. (including rybka 2.3.2a).
so my point is, humans CAN tell the differences between engines they use for analysis.
here's a concrete example:
take this position. and by the way, this is actually in my repertoire (as black), so it's really important to me.
I fire up Deep Fritz 11, which can beat me anytime, and it evaluates the position as won by white.
I fire up Rybka 3 Human, which also can beat me anytime, and it evaluates the position as roughly equal.
it makes all the difference in the world to me, while analyzing games that arise from these type of positions.
1 edit
Originally posted by wormwoodFor a huge test set of positions, varying in nature (tactical, positional, endgames, etc.) if both engines eventually agree upon the best move but Fritz X generally takes longer than Fritz Y, then Fritz Y is better for the analysis of these positions.
compare against what? f10? how could you possibly know which one gave the best analysis? what do you compare it against?
Hi
"Actually, I think there's no point in getting fritz 12, because the
improvements will probably be negligible. "
Thank You PP.
Come on Cimon it's a bit unfair calling someone names just
because they are having a bit of fun.
You know who I am.
I do not hide behind an alias and call people names.
I have no idea who you are. 😉
I wish my friend Korch would come back - he would stand up for me.
5 edits
Originally posted by Varenkawhat if that move is wrong? is it of any consolation that f12 made the wrong decision faster? how many of the moves they 'find' will ultimately turn out to be wrong? we don't really keep a score on how often engines change their evaluation for a certain position, depending on the depth they see and the built-in evaluation function. they change their mind close to every ply, which is another way of saying the previous verdict was incorrect. how do we know the last one is correct?
For a huge test set of positions, varying in nature (tactical, positional, endgames, etc.) if both engines eventually agree upon the best move but Fritz X generally takes longer than Fritz Y, then Fritz Y is better for the analysis of these positions.
the only thoroughly analyzed position in chess is the opening position, and even there there's much more ground to cover. we know for a fact that all engines do poorly in that position, compared to the hundreds of years of human analysis. sure the engines have brought up novelties in opening as well, but it's a tiny subset.
I think it's safe to say the human knowledge of the opening position beats that of engines hands down. and I don't see any reason why the same wouldn't apply to any equal but rich position. to me it seems like strong evidence for how much the pruning algorithms miss, and how far from correct the over all engine analysis is.
the fact that even strongest GMs lose to engines is irrelevant, because we can't rule out humane weaknesses like losing concentration, fatigue etc. -what I'm trying to say is that relative to the quality of judgment, humans still play higher quality chess. the engines win because of tactical oversights, not because of better chess.
in a way, the machines are winning with cheapos. a win is a win, but does it mean cheapo chess was better?
even in the famous kramnik loss to a 1-mover, he was winning wasn't he? everything he thought about the position was right, on an extremely high chess level, everything except the 'cheapo' he missed which got him mated. and although that's a massive blunder compared to a slight advantage gained 10 moves deep, I think it's essentially the same exact event. loss of concentration, missing one of the vast number of candidates at some point. missing the cheapo.
most of this is not directed at you, varenka, but also other posts above. on that same vein I'd like to also explicitly point out that I have absolutely no argument with engines winning statistically more games against GMs, even crushingly these days. my beef is with the correctness of their evaluation, and the missguided intuition that engine evaluation was some kind of god given absolute truth, or that it could be one even in a theoretical sense. because until chess is solved, it can't be.
Originally posted by philidor positionI don't really understand why it makes a difference to you what one engine or another thinks about this position - what's improtant is what you think.
1. e4 e6 2. d4 d5 3. Nc3 Bb4 4. e5 Qd7 5. Qg4 f5 6. Qg3
take this position. and by the way, this is actually in my repertoire (as black), so it's really important to me.
I fire up Deep Fritz 11, which can beat me anytime, and it evaluates the position as won by white.
I fire up Rybka 3 Human, which also can beat me anytime, and it eval ...[text shortened]... e difference in the world to me, while analyzing games that arise from these type of positions.
Sure, if the engine showed that the line you were thinking of playing was tactically flawed then that's great, the engine has probably saved you from losing a game, but if it's just the engine's opinion about a materially equal position then it doesn't matter a fig whether it thinks it's +0.2 or +1.0 for White.
You're going to be playing humans, not engines, so you've got to decide for yourself whether you like the Black side of that position.
You could, of course, play a few games against an engine starting from that position, and it may show up some interesting plans for one side or the other. However I've had very limited success using engines in that way.
Originally posted by wormwoodwhat if that move is wrong?
There are people who make up high quality test sets. These positions get verified by running lots of various engines for long periods of time. Also the lines get played out using interactive analysis. If other people find mistakes in the analysis, they report it and the quality of the test set improves. After some time, the percentage of wrong "best moves" is very low indeed.
the only thoroughly analyzed position in chess is the opening position
Wrong. Many endgames have been analysed very thoroughly. Go through Dvoretsky's endgame manual and try finding some mistakes. I'm not saying there aren't any, but the positions have generally been analysed very thoroughly. Also, many famous games have been analysed thoroughly (e.g. Kasparov's Great Predecessors) - again, go through and see how difficult it is to find mistakes in the latest analysis.
the engines win because of tactical oversights, not because of better chess
Tactics are part of chess, are they not? And chess ratings are based on results only. You don't get compensation for losing due to a tactical oversight.
1 edit
Originally posted by Fat LadyBut in order to improve we sometimes need feedback on what we think. For example, some people are too materialistic and never consider sac'ing the exchange. So they think for a given position "I won't exchange my rook for the knight". But if an engine frequently highlights that sac'ing is in fact a strong move (with lines to back it up) then gradually they will develop a better feel for when to play such moves.
I don't really understand why it makes a difference to you what one engine or another thinks about this position - what's improtant is what you think.
Most players agree that getting a stronger player to go over your games is a good idea. Why does it matter what a stronger player thinks, but yet not a strong engine?