Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: is hydra now stronger than deep blue?

Author: Dann Corbit

Date: 16:55:32 05/25/05

Go up one level in this thread


On May 25, 2005 at 19:17:08, Roger D Davis wrote:

>On May 25, 2005 at 16:05:20, Uri Blass wrote:
>
>>On May 25, 2005 at 15:49:57, Roger D Davis wrote:
>>
>>>On May 25, 2005 at 15:06:40, Uri Blass wrote:
>>>
>>>>On May 25, 2005 at 13:35:59, Roger D Davis wrote:
>>>>
>>>>>On May 25, 2005 at 13:10:34, Uri Blass wrote:
>>>>>
>>>>>>On May 25, 2005 at 12:58:46, Roger D Davis wrote:
>>>>>>
>>>>>>>On May 25, 2005 at 05:35:14, emerson tan wrote:
>>>>>>>
>>>>>>>>s hydra now stronger than deep blue?
>>>>>>>
>>>>>>>We know Kasparov, even then, was a much stronger player than Adams is today. If
>>>>>>>Hydra, supposedly stronger than Deep Blue, loses to a much weaker player, then
>>>>>>>that provides a strong argument that Hydra is weaker than Deep Blue.
>>>>>>>
>>>>>>>On the other hand, if Adams loses, then it says nothing about Hydra's strength
>>>>>>>relative to Deep Blue.
>>>>>>>
>>>>>>>I guess you could always argue that Deep Blue can beat Kasparov and Kasparov can
>>>>>>>beat Adams and Adams can beat Hydra and Hydra can beat Deep Blue, but it doesn't
>>>>>>>seem likely. Particularly if Adams can get a convincing score.
>>>>>>>
>>>>>>>Roger
>>>>>>
>>>>>>I think that you have no way to compare Adams of 2005 with Kasparov of 1997.
>>>>>>Humans today have more experience against computers relative to 1997 and it is
>>>>>>not clear to me that Kasparov of 1997 was stronger against computers relative to
>>>>>>Adams of 2005.
>>>>>>
>>>>>>I also think that the fact that Kasparov lost says nothing because the 2 games
>>>>>>that kasparov lost were because of stupid mistakes of him because of
>>>>>>psychological reasons(resigning in a drawn position and playing a line that he
>>>>>>was not ready to play).
>>>>>>
>>>>>>Hydra is also more known than Deeper blue was known at the time of Kasparov
>>>>>>
>>>>>>Kasparov could get no games of something similiar to deeper blue(deep thought
>>>>>>was clearly weaker) when Adams has no problem to get games of something similiar
>>>>>>to hydra.
>>>>>>
>>>>>>Uri
>>>>>
>>>>>What I said was that it provides a strong argument. I don't think it's a matter
>>>>>of certainty. I think it's a matter of making probabilistic statements, and
>>>>>knowing their limitations. In addition to not knowing whether the Kasparov of
>>>>>1997 was stronger against computers relative to the Adams of 2005---as you
>>>>>pointed out---we don't know whether Deep Blue's style might have been
>>>>>particularly deadly to Kasparov for some reason, or whether Hydra's style might
>>>>>be particularly vulnerable to Adams, or whether Adams has been reading this
>>>>>bulletin board and picking up pointers on the weaknesses of computers. We don't
>>>>>even know how successfully Hsu's team managed to tune Deep Blue against
>>>>>Kasparov. Maybe it will eventually emerge that it's always possible to tune a
>>>>>strong enough hardware beast against any particular human and defeat him. Who
>>>>>knows. Maybe Kasparov wouldn't freak himself out today and lose with stupid
>>>>>mistakes and then again, maybe he would.
>>>>>
>>>>>So...lots of unknowns.
>>>>>
>>>>>Comparisons are interesting and inevitable. Humans will find a way of making
>>>>>comparisons whether we want them to, or not. I think you can continue to 2nd
>>>>>guess yourself ad infinitum about most anything. I prefer not to do that and
>>>>>just stick with my statement that an Adams victory provides a strong argument
>>>>>that Hydra is weaker than Deep Blue. Does it establish it with certainty.
>>>>>Obviously not. But it agrees with commonsense, and that's the ruler that most
>>>>>people will bring to the interpretation if Adams wins. I think if you're looking
>>>>>for certainty, it's best to stick with mathematical proof. Everything else is
>>>>>fraught with contention.
>>>>>
>>>>>Roger
>>>>
>>>>I think that one assumption that you make is wrong in all the discussion.
>>>>
>>>>Adams is not much weaker player than Kaspparov and the rating difference between
>>>>them is only 75 elo.
>>>>
>>>>Here is the fide rating list:
>>>>
>>>>1  Kasparov, Garry  g  RUS  2812  12  1963-04-13
>>>>2  Anand, Viswanathan  g  IND  2785  25  1969-12-11
>>>>3  Topalov, Veselin  g  BUL  2778  25  1975-03-15
>>>>4  Leko, Peter  g  HUN  2763  25  1979-09-08
>>>>5  Kramnik, Vladimir  g  RUS  2753  13  1975-06-25
>>>>6  Ivanchuk, Vassily  g  UKR  2739  17  1969-03-18
>>>>7  Adams, Michael  g  ENG  2737  25  1971-11-17
>>>>
>>>>Uri
>>>
>>>
>>>
>>>I think you're assuming that all ELO intervals are created equal.
>>>
>>>Kasparov is generally considered to be the most powerful player in the history
>>>of the game. There is no other player in the world so widely known. No other
>>>player commands the kind of fame that Kasparov does.
>>>
>>>Yet, in 1997, the year of the Deep Blue match, Kasparov was rated about 2795,
>>>less than what he is today.
>>>
>>>Historically, I'd bet that seldom has Kasparov been 75 points stronger than the
>>>#2 ranked player. Maybe never.
>>>
>>>So it's obvious that it takes 75 points or less to make a Kasparov.
>>>
>>>Roger
>>
>>Kasparov did not beat Karpov easily.
>>13-11,12.5-11.5,12-12 are results that I remember.
>>
>>Later Kasparov also had problems in the beginning of the match against anand and
>>the result was 4-4 and I remember that anand even won the first game in the
>>match.
>>
>>I think that Kasparov had never big advantage relative to player number 2 or
>>even player number 7.
>>He was better but only slighlty better.
>>
>>He could win convincingly(see the match against short) but also lose(see the
>>match against Kramnik).
>>
>>He could win a tournament but also could lose in a tournament and I remember
>>that Karpov won some tournament that kasparov said that the tournament is going
>>to show who is the real champion(Karpov was the fide champion at that time but
>>only because Kasparov and Short did not agree to accept the conditions of
>>fide).
>>
>>Uri
>
>One thing we can probably agree on: Since Deep Blue beat Kasparov and Kasparov
>beats Adams, then if Adams beats Hydra it becomes the commonsense position that
>Deep Blue is stronger than Hydra.

No way.
The Los Angeles Clippers beat the Dallas Maverics:
Mar Opponent Box Score W-L High Pts High Reb
  Wed 2 Dallas  W 101 - 92  26-32 C. Maggette 34 E. Brand 10
The Dallas Maverics beat the Detroit Pistons:
Mon 28 Dallas  L 88 - 95  43-27 C. Billups 26  B. Wallace 17

Can we therefore assume that the Clippers are stronger than the Pistons?
Clearly not.
It takes a large number of contests to determine even approximate strength.

You are right in a sense that most people will believe what you say.  But the
problem with that is that it is not correct.

>A lot of argumentation has to be provided to support any counterargument. Notice
>I didn't say there was no counterargument, merely that supporting it requires
>some cognitive somersaults, like falling back on confidence intervals,
>possibility that Kasparov wouldn't comment errors in a rematch, possibility that
>Hydra is just vulernable to Adams and wouldn't be vulnerable to other GMS,
>possibility that Deep blue was tuned well against Kasparov, but wouldn't have
>played as well versus other GMs, etc.

A tiny contest like this cannot demonstrate a thing like that.

>In fact, if the difference between Deep Blue and Kasparov was 75 points, and if
>the difference between Kasparov and Adams is 75 points, and if the difference
>between Adams and Hydra is 75 points, then Adams will barely win, and the
>difference between Deep Blue and Hydra could still be 75 x 3 = 225 points. In
>this scenario, each barely gets a win other the others, and the difference could
>be 225 points between Deep Blue and Hydra. We'll never know since Deep Blue and
>Hydra will never play

Elo does not work like that.  If the player with a larger Elo always won, there
would be no need to play the games.  In fact Elo cannot be used to determine the
outcome of any experiment a-priori.  It is only an indicator of what might
happen.

>"Could" is the key word here.
>
>I think Adams will lose, but I think it's more interesting if he wins.

It is equally interesting to me either way.
It's probably more surprising if he wins.



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.