Uri Blass wrote:I suggest that you simply ignore Vincent.
It is typical for him to give claims that are clearly false.
Well Uri I have to agree with Vincent. There was something fishy about DB - Kasparov rematch. I too think that Kasparov could have done much better and that DB in fact played nothing spectacular.
DB is perhaps the most overhyped and overestimated chess computer in history. But there are people who still buy the fairytale even today
No wonder they dismantled it after the match. And now IBM no longer hosts the games/logs.
DB reached 11 plies on averge in the midgame (software). Add 4 hardware plies, that's 15 fat plies, not bad but hardly good enough to match anything strong today IMHO (and not good enough to beat world champion for sure).
noctiferus wrote:Are you planning to release this tool with the next Dabbaba version?
Sure.
(And even better I hope this function will be incorporated in some GUI's, so the analyzes and the new presentation can easily be done in one step.)
As I wrote in an earlier post:
But anyone can do the analyzes.
I assume Arena 3.0 will produce the correct analyzed pgn-games with any engine.
And Dabbaba is free for anyone to convert the pgn-file to the list with the performance rating.
(this version is not published yet, but I can send it directly to those who are interested - just post me your email-adr)
I will release Dabbaba when its playing strength is improved - or the interest for this new function increases.
So far I will mail it to those who are interested.
I would be pleased if someone would analyze the matches Fischer-Spasskij 1972 and 1992.
I would also like to hear what people think of the analyzes of their own games.
Uri Blass wrote:I suggest that you simply ignore Vincent.
It is typical for him to give claims that are clearly false.
Well Uri I have to agree with Vincent. There was something fishy about DB - Kasparov rematch. I too think that Kasparov could have done much better and that DB in fact played nothing spectacular.
DB is perhaps the most overhyped and overestimated chess computer in history. But there are people who still buy the fairytale even today
No wonder they dismantled it after the match. And now IBM no longer hosts the games/logs.
DB reached 11 plies on averge in the midgame (software). Add 4 hardware plies, that's 15 fat plies, not bad but hardly good enough to match anything strong today IMHO (and not good enough to beat world champion for sure).
Saying that kasparov could do better is something clearly different than saying that he played at level of 2100 and 11 plies on average in the middlegame are clearly enough to beat every 2100 player by more than 3.5:2.5(even if it is not fat plies but only plies with no pruning).
I also do not know how do you define something strong.
I strongly believe that many chess programs of today can beat Deep Blue of 1997(including relatively weak programs that are more than 400 elo weaker than houdini1.5 64 bit 1 cpu in the rating list)
so relative to chess programs of today it is not strong but I also believe that not many humans
can beat DB of 1997 so relative to humans it was strong.
Uri Blass wrote:I also do not know how do you define something strong.
I strongly believe that many chess programs of today can beat Deep Blue of 1997(including relatively weak programs that are more than 400 elo weaker than houdini1.5 64 bit 1 cpu in the rating list)
so relative to chess programs of today it is not strong but I also believe that not many humans
can beat DB of 1997 so relative to humans it was strong.
Yes by strong I mean something like that, unfortunately it cannot be verified as DB won't play any more games.
Sure it was good enough to beat most humans (unless it would play blunders like in game 1).
But we are talking about a world champion at that time.
Hard to say, perhaps Kasparov simply was not in good shape, perhaps he underestimated the preparation.
Perhaps he was too much self-confident. But these are pure speculations.
DB did beat Kasparov, that's a fact. But doubts remain IMHO.
Hi Jens.
Thanks for your effort!
50 elo difference in performance in Deep Blue's favour when it actually lost the match in 1996 is obviously wrong.
200 elo difference in performance in Deep Blue's favour in the second match when it actually won by a very narrow margin with just 6 games is obviously wrong too.
I think Kasparov was very willing to win both matches, as Kasparov is like that - he would go heart and mind into an affair when his champion's status was being questioned. I am convinced he was well prepared and he actually played better than in many human events. Looking into the games will leave you with the impression of a superb quality indeed (glitches are always there).
I think what is wrong with measuring the performance rating is the use of a single engine that is able to assess better the other engine's moves than the grandmaster's moves.
Rybka sees many good tactical moves that Kasparov does not see and does not see a number of really good positional moves that Kasparov does. You can rely on Rybka, but not anywhere close to even 80% of cases.
I expect that redoing the exercise with Karpov will return even lower performance rating, as Karpov's style is even more positional.
If you use Houdini for doing the measurements, both Kasparov and Karpov's performance ratings could jump, simply because Houdini is a better positional player than Rybka, and it would also be critical of a number of Deep Blue's moves, probably lowering its rating. In this way I think the picture of both matches will look pretty much sensible.
Lyudmil Tsvetkov wrote:Hi Jens.
Thanks for your effort!
50 elo difference in performance in Deep Blue's favour when it actually lost the match in 1996 is obviously wrong.
200 elo difference in performance in Deep Blue's favour in the second match when it actually won by a very narrow margin with just 6 games is obviously wrong too.
I think Kasparov was very willing to win both matches, as Kasparov is like that - he would go heart and mind into an affair when his champion's status was being questioned. I am convinced he was well prepared and he actually played better than in many human events. Looking into the games will leave you with the impression of a superb quality indeed (glitches are always there).
I think what is wrong with measuring the performance rating is the use of a single engine that is able to assess better the other engine's moves than the grandmaster's moves.
Rybka sees many good tactical moves that Kasparov does not see and does not see a number of really good positional moves that Kasparov does. You can rely on Rybka, but not anywhere close to even 80% of cases.
I expect that redoing the exercise with Karpov will return even lower performance rating, as Karpov's style is even more positional.
If you use Houdini for doing the measurements, both Kasparov and Karpov's performance ratings could jump, simply because Houdini is a better positional player than Rybka, and it would also be critical of a number of Deep Blue's moves, probably lowering its rating. In this way I think the picture of both matches will look pretty much sensible.
Thanks again for the effort.
Best,
Lyudmil
Dopre kazano Lyudmil.....
Napalno se saglasiavam s tvoite izgledi svarzani s tozi istoricheski match....
Dr.D
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….
mar wrote:
DB is perhaps the most overhyped and overestimated chess computer in history. But there are people who still buy the fairytale even today
It regularly beat the crap out of other computers at the time. Even its predecessors (ChipTest and Deep Thought) did so.
I know that Kasparov and his defenders like to give the idea that Deep Blue was an obscure computer that only played against Kasparov, but it's really not true.
I recommend this book which features the other side of the story, it's interesting and informative:
rbarreira wrote:
It regularly beat the crap out of other computers at the time. Even its predecessors (ChipTest and Deep Thought) did so.
I know that Kasparov and his defenders regularly give the idea that Deep Blue only played against Kasparov, but that's really not true.
I recommend reading this book which features the other side of the story, it's interesting and informative:
Well it ought to beat other computers at the time (single core Pentia).
Except that a DB prototype lost to Fritz in 1995 - but that's what a prototype is for.
As for the book: thanks but I'm not interested, I believe Hsu wrote better stuff than that.
rbarreira wrote:
It regularly beat the crap out of other computers at the time. Even its predecessors (ChipTest and Deep Thought) did so.
I know that Kasparov and his defenders regularly give the idea that Deep Blue only played against Kasparov, but that's really not true.
I recommend reading this book which features the other side of the story, it's interesting and informative:
Well it ought to beat other computers at the time (single core Pentia).
Except that a DB prototype lost to Fritz in 1995 - but that's what a prototype is for.
As for the book: thanks but I'm not interested, I believe Hsu wrote better stuff than that.
It didn't just beat them, it beat them very convincingly and it improved gradually to a point where it became possible to take on Kasparov and possibly win (which it did). So I still don't see what you meant by "overhyped and overestimated", its prototypes / predecessors had enough games played to reach a conclusion about their strength.
What do you recommend by Hsu? All I have read is that book and some article about Deep Blue's internals.
Uri Blass wrote:I also do not know how do you define something strong.
I strongly believe that many chess programs of today can beat Deep Blue of 1997(including relatively weak programs that are more than 400 elo weaker than houdini1.5 64 bit 1 cpu in the rating list)
so relative to chess programs of today it is not strong but I also believe that not many humans
can beat DB of 1997 so relative to humans it was strong.
Yes by strong I mean something like that, unfortunately it cannot be verified as DB won't play any more games.
Sure it was good enough to beat most humans (unless it would play blunders like in game 1).
But we are talking about a world champion at that time.
Hard to say, perhaps Kasparov simply was not in good shape, perhaps he underestimated the preparation.
Perhaps he was too much self-confident. But these are pure speculations.
DB did beat Kasparov, that's a fact. But doubts remain IMHO.
Kasparov beat himself. His opening preparation for this match was silly. He would have been better off playing Deep Blue like if it was a human. He also gave Deep Blue too much respect.
