Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Junior 9 Test (40'/40) after 300 games

Author: Graham Banks

Date: 13:44:59 12/31/04

Go up one level in this thread


On December 31, 2004 at 16:38:53, Sandro Necchi wrote:

>On December 31, 2004 at 16:32:38, Graham Banks wrote:
>
>>On December 31, 2004 at 16:17:02, Sandro Necchi wrote:
>>
>>>On December 31, 2004 at 15:44:22, Graham Banks wrote:
>>>
>>>>On December 31, 2004 at 15:21:23, Sandro Necchi wrote:
>>>>
>>>>>On December 31, 2004 at 15:03:40, James T. Walker wrote:
>>>>>
>>>>>>On December 31, 2004 at 12:18:36, Kurt Utzinger wrote:
>>>>>>
>>>>>>>Matches at 40’/40 + 40’/40 +40’ time control
>>>>>>>Junior9-GUI, ponder=off, 3-/4-men EGTB
>>>>>>>own books, no book learning, no learning
>>>>>>>on 4 Athlons 1.3/64 MB hash for all engines
>>>>>>>Details and games for download as usual at
>>>>>>>http://www.utzingerk.com/jun9_test.htm
>>>>>>>Mfg
>>>>>>>Kurt
>>>>>>>
>>>>>>>(3) Junior 9                  : 300 (+109,= 97,- 94), 52.5 %
>>>>>>>
>>>>>>>The King 3.23 T05             :  50 (+ 18,= 20,- 12), 56.0 %
>>>>>>>Chess Tiger 15.0              :  50 (+ 20,= 17,- 13), 57.0 %
>>>>>>>Fritz 8                       :  50 (+ 22,= 15,- 13), 59.0 %
>>>>>>>Hiarcs 9                      :  50 (+ 13,= 17,- 20), 43.0 %
>>>>>>>Shredder 8                    :  50 (+ 11,= 18,- 21), 40.0 %
>>>>>>>Gandalf 6.0                   :  50 (+ 25,= 10,- 15), 60.0 %
>>>>>>
>>>>>>Hello Kurt,
>>>>>>While I find your results interesting and others with similiar results with
>>>>>>"Ponder off/no learning", I have to wonder if these test are worthwhile.
>>>>>
>>>>>I agree. If a program has a better learning why not to use it just to put all on
>>>>>the same level?
>>>>>With this idea we can start removing other parts as well.
>>>>>I believe programs should be tested as they are in their best tournament mode.
>>>>>If someone wants to test new books it should test the same program with its own
>>>>>book and with another book and not only the second choice.
>>>>>
>>>>>>The problem is that pondering is part of the program.
>>>>>
>>>>>I agree here too.
>>>>>
>>>>>>If you are trying to test which is best at playing chess then cripling all >programs is not necessarily cripling them equally.  What if some programs are >better at predicting others moves and therefore gain an advantage by pondering more accurately.
>>>>>
>>>>>I agree 100%...since these are CHESS PLAYERS how would be to force a chess
>>>>>player not to think while the opponent is thinking?
>>>>>This really makes no sense to me.
>>>>>
>>>>>>The same for learning/book learning.
>>>>>
>>>>>Again I agree. Some years ago there was no learning/book learning at all and
>>>>>that was a big missing...which made a big difference with the human players.
>>>>>Now that we have them we need to improve the learning features and not to turn
>>>>>them off!
>>>>>
>>>>>>I'm getting suspicious that most of the improvements in new programs is just >some "book-up" tricks against certain programs to gain quick Elo points.  >Disabling learning will allow these "tricks" to work continiously while book >learning/learning will eventually nullify them.
>>>>>
>>>>>This is quite correct...it would damage the older programs only as the new ones
>>>>>may include variations which may put in trouble the old programs and turning off
>>>>>the learning will make harder to the old program to find a way to avoid
>>>>>loosing...so this is an unfair things to old programs and can only give higher
>>>>>expectations on new versions; much more than they really are!
>>>>>
>>>>>>I don't know if you've seen my blitz database ratings but it seems the longer >I play them the closer they get in ratings.  My ratings also closely immitate >the SSDF list by showing only a few points increase between the Chess Tigers >and Shredders.
>>>>>
>>>>>This is true, but in case of Shredder do not forget the different GUI and the
>>>>>book learning + different book mode between the 2 GUIs.
>>>>>Shredder 8 CB is the update of Shredder 7 CB or Shredder 7.04 CB
>>>>>Shredder 8 UCI is the update of Shredder 7 UCI...unfortunately SSDF did not test
>>>>>it.
>>>>>
>>>>>>Junior programs are showing up in the same fashion lately.  I
>>>>>>currently have Junior 9 trailing Junior 8 by 2 Elo points.  I'm getting
>>>>>>suspicious that top programs are hitting a "wall" and showing no real
>>>>>>improvement in strength, only a change in the way they play.
>>>>>
>>>>>I think that even the top programs can be improved...let's see the new Shredder
>>>>>9 if it will be keep the promises...
>>>>>
>>>>>>Just food for thought.
>>>>>
>>>>>Thanks for your smart post!
>>>>>
>>>>>>Regards,
>>>>>>Jim
>>>>>
>>>>>Sandro
>>>>
>>>>
>>>>At what time control should learning take place though and this is a huge issue
>>>>as far as I'm concerned.
>>>
>>>At all levels because a good learning allows learning from long time control
>>>levels to fast ones and not the other way around!
>>>
>>>>Does a GM automatically alter his 40/2hrs play based on what has happened in a
>>>>blitz game.
>>>
>>>It could happen because they are able to judge a position about immediately, so
>>>it is not unlike.
>>>
>>>>I'd suggest he'd take time to study his experience in greater depth
>>>>before aending his play.
>>>
>>>It depends on the GM as some needs time to calculate and others do not.
>>>
>>>>This is where learning has its faults in computer play. Is learning that >occurs as a result of blitz play going to be constructive learning when >you're looking at lower ply depths? Such learning could actually have a negative impact.
>>>
>>>See above.
>>>
>>>>
>>>>Graham.
>>>
>>>Sandro
>>
>>But we're looking at programs having different learning experiences with
>>different users, so where's the consistency?
>
>OK, but this is a player feature, so it must be used in any cases if we want to
>make comparisons between them.
>
>>It's like cloning a GM and having the clones play in different tournaments. >They will each adapt their play according to their learning experiences, so >in fact they become different!
>
>This is true, but if we test without learning they will loose again and again
>the same game...does it make sense to you?
>Quite interesting is how they try to solve these problems bringing new moves
>every game until they find one!
>
>>
>>Graham.
>
>Sandro


I think we've both made some good points.
The main thing I wanted to get across though is that when testers test with
learning off, at least you're getting consistent data. Of course if I get any
duplicate games they are replayed.

Regards, Graham.



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.