Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Is Deep Blue still considered better than Deep Junior ?

Author: Robert Hyatt

Date: 10:30:09 08/20/02

Go up one level in this thread


On August 20, 2002 at 07:26:53, Vincent Diepeveen wrote:

>On August 18, 2002 at 22:34:04, Robert Hyatt wrote:
>
>>On August 18, 2002 at 20:47:58, Vincent Diepeveen wrote:
>>
>>>On August 18, 2002 at 17:09:08, Jonas Cohonas wrote:
>>>
>>>>On August 18, 2002 at 15:27:44, José Carlos wrote:
>>>>
>>>>>On August 18, 2002 at 12:41:58, Vincent Diepeveen wrote:
>>>>>
>>>>>>On August 18, 2002 at 11:31:54, Chris Taylor wrote:
>>>>>>
>>>>>>>On August 18, 2002 at 09:06:02, Jorge Pichard wrote:
>>>>>>>
>>>>>>>>   Kasparov proved that he can defeat programs at fast time controls when he
>>>>>>>>defeated Deep Thought in a game/90 two games match in 1989. This program was
>>>>>>>>weaker than Deep Junior is today, as it searched well over 2,000,000 NPS, but
>>>>>>>>didn't have as much chess knowledge as Deep Junior.  He also defeated Deep Blue
>>>>>>>>in 1996. This program is obviously much faster than Deep Junior is today, but in
>>>>>>>>my opinion Deep Junior still has more chess knowledge than Deep Blue had back in
>>>>>>>>1996.
>>>>>>>>
>>>>>>>>PS: It is hard to compare Deep Blue of 1997 vs Deep Junior of today, but in my
>>>>>>>>opinion Deep Junior Chess Knowledge could make up for the difference of Deep
>>>>>>>>Blue super calculating power of 1997.
>>>>>>>>
>>>>>>>>Pichard.
>>>>>>>
>>>>>>>One way would be to play some games with Deep Blue and Deep Junior.  Guess that
>>>>>>>would settle once and for all who is the strongest.  Or would it just pour fuel
>>>>>>>on the **whos** best fire.  Put together the blue box and match it up.  After
>>>>>>>all it did beat the best player in the world at that time!  The advert could be
>>>>>>>quite powerful.  The machine that beat Kaspy goes for Junior. Methinks there
>>>>>>>could be some money to be made here? So this may not happen, shame?
>>>>>>
>>>>>>that will of course never happen. Just like fischer still is world
>>>>>>champion, deep blue will be world champion in some scientist eyes forever
>>>>>>too.
>>>>>>
>>>>>>To be clear. I feel that any 2650+ player of todaywill wipe out fischer
>>>>>>if he plays like he played in 1970.
>>>>>>
>>>>>>New theory, better tactics, more insight in strategies, better training.
>>>>>>
>>>>>>A 2650 player of today is going to crush any world champ from before Karpov
>>>>>>of course. No doubts.
>>>>>>
>>>>>>Robert J Fischer when the rating list started had 2780 or something. that was
>>>>>>superb compared to anyone in those days. He was the best back then. No one
>>>>>>was as good.
>>>>>>
>>>>>>But the level has improved a lot. Many will say now: "this is not a fair
>>>>>>compare a modern 2650 player against someone who had only an old
>>>>>>book from capablanca and tarrasch, if he could read german anyway".
>>>>>>
>>>>>>In fact a grandmaster did this comparision. He compared a top tournament
>>>>>>in 1991 with a top tournament from 1920. The grandmaster was called Nunn
>>>>>>if i remember well.
>>>>>>
>>>>>>The last few players in that tournament around the start of the 20th
>>>>>>century, they simply blundered away piece
>>>>>>after piece. Would be rated at most 1500 nowadays.
>>>>>>
>>>>>>The 'better players' in the tournament, considered *clear world top*
>>>>>>back then, they blundered on average 5 times a game.
>>>>>>
>>>>>>*no modern topgrandmaster is doing that*.
>>>>>>
>>>>>>The level of the world top increases. This is logical. Suppose you
>>>>>>get to the tennis court with a wooden racket. Even if you're called
>>>>>>John McEnroe you will be of course get completely annihilated. A wooden
>>>>>>racket and services of 160KM/hour (the speed at which McEnroe served) it
>>>>>>is no compare to the 180-220 KM/hour services of modern tennis of today.
>>>>>>
>>>>>>He won't manage a single break of course.
>>>>>>
>>>>>>This is logical. Sport progresses. computerchess even faster. saynig that
>>>>>>deep blue/deep thought was good in its days is justified. It beated some
>>>>>>GMs. That the GMs played big shit games because they cared shit as they
>>>>>>had nothing to proof and would get money anyway, that's no issue here.
>>>>>>
>>>>>>The issue is that it is so *obvious* that software in 2002 is much better
>>>>>>than in 1997 that i am amazed that only Hyatt here doubts it.
>>>>>>
>>>>>>>Chris
>>>>>
>>>>>  Couldn't disagree more. Give Capablanca or Lasker a couple of months to train
>>>>>against today's GM's and they'll do quite well. Talented players learn fast,
>>>>>Vincent. It's not like programs. A program is "closed", it can't modify itself.
>>>>>A player such as Lasker would be able to catch up in very little time.
>>>>>  As for tennis, a fair comparison would be give McEnroe a new racket and see
>>>>>how long does he need to get used to it. Otherwise it's not a fair comparison.
>>>>>Lasker brain would not be obsolete today. His knowledge would be, but knowledge
>>>>>can be learnt.
>>>>>
>>>>>  José C.
>>>>
>>>>I agree with you completely, there is a reason that todays top GM's show a lot
>>>>of respect for players of the old days when talking about them...
>>>>
>>>>You could also flip the tennis analogy and have todays top players go back in
>>>>time and see how they would fare with wooden rackets and slower balls, against
>>>>the top players from back then.
>>>>
>>>>Regards
>>>>Jonas
>>>
>>>I hope you realize very clearly that Deep Blue can't get upgraded.
>>>Yes it was a big achievement for 1 man to make all that. The chip programmed
>>>by 1 person. It's incredible.
>>
>>Would you mind explaining that?
>>
>>1.  The software part of the search most definitely could be "upgraded"
>>as it was done many times...
>>
>>2.  The hardware could _also_ be upgraded.  Deep thought was upgraded to
>>deep blue.  Deep Blue was upgraded to deep blue 2.  Why would you write
>>something that is so obviously wrong???
>
>the chip was improved no doubt. it was not even in verilog Bob.
>How to *ever* upgrade that to todays 0.13 or 0.18 micron?

The same way Hsu took deep thought to the first deep blue chip design.  The
same way Hsu took deep blue 1 to the deep blue 2 design.

Verilog is _not_ a requirement.



>
>>3.  Even if the chess chips were "static" could not IBM give them a _bigger_
>>(or faster) SP2?  They could easily fab more chips.  And run faster...  Of
>>course that overlooks points 1 and 2 above completely, which is not reasonable.
>
>10000 chips gives just 13 ply *at most*.


Says who/what?  I can do 13 plies with _one_ chip...

And my "chip" is not an ASIC implementation of anything, just a generic
microprocessor.



>
>>
>>
>>>
>>>But please don't compare it with 2002 software. It's going to get raped,
>>>dicked and completely annihilated *everywhere*. Opening, first moves out of
>>>book with structure, mobility, even good/bad bishop, king safety (see the
>>>major errors it made there which none of todays program makes), exchanging
>>>to endgame (2 horrible exchanges of a queen in the 1997 games, one Qxg6??
>>>being a full blunder making from a better position a lost one) and finally
>>>endgame.
>>>
>>>So the comparision is not fair, but i hope you realize how bigtime it
>>>would get completely *destroyed* with induction against todays programs.
>>>Not a single area where it is impressing. Let's not discuss tactics.
>>>If you get 11-12 ply then tactics do not decide the game. I feel it
>>>wasn't bad tactically at all if i look to the design specifications of
>>>the extensions.
>>>
>>>Best regards,
>>>Vincent



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.