Author: Uri Blass
Date: 21:08:56 02/02/05
Go up one level in this thread
On February 02, 2005 at 17:05:30, Ingo Bauer wrote: >On February 02, 2005 at 16:47:28, Uri Blass wrote: > >>On February 02, 2005 at 15:25:29, Ingo Bauer wrote: >> >>>>>After one day you may have a couple of hundred, maybe thousand games at 3 0 at >>>>>playchess. Unfortunately this is completly worthless for Shredder! I tried there >>>>>with two different accounts (both S9). One is playing only games with an hour or >>>>>longer, the other run 2 nights only 3 to 5 min no increments. Both on the same >>>>>Hardware. The difference at the end was about 100 points. Of course the long >>>>>game account is much better! >>>>> >>>>>Bye >>>>>Ingo >>>>> >>>>>PS: Both account did not play versus other Shredders! >>>> >>>>I do not understand how do you compare between different time controls. >>>> >>>>You can say that you cannot learn from blitz about long time control but the >>>>knowledge if the new version is better at blitz is also an interesting >>>>information. >>> >>>All I wanted to say is >>> >>>1. that Shredder needs some time to let the search algorith settle. >>>2. On playchess the vast majority of games is 3 0 and therefore a bit unlucky >>>for Shredder to show it's real strength. >>> >>>Bye >>>Ingo >> >>If I understand correctly you mean to say that the difference between Shredder9 >>and older version was bigger at longer time control. > >Why are you making it so complicated? > >All I wanted to say is that expecting a "real" value out of some blitz games at >playchess within one day is impossible. I disagree. We cannot expect real value at long time control based on some blitz games but we can expect real value at blitz. I am sure that most here did understand >that quite well after my first post. Why you didnt is beyond my imagination. > >Sorry for the inconvienience I gave you >Ingo I believe that most did not understand you because you did not explain yourself well. You said: "On playchess the vast majority of games is 3 0 and therefore a bit unlucky for Shredder to show it's real strength." You did not say "On playchess the vast majority of games is 3 0 and therefore it does not give a good estimate for the strength at long time control." If a program is unlucky because of the time control then it means that it can perform relatively better at different time control. If Shredder is unlucky because the games are 3 0 then it means that it can perform relatively better at long time control and this is the reason that I understood you in the way that I understood. I also do not agree with you that the real strength is expressed only by long time control games. You may be interested only in long time control but a lot of people are interested more in blitz and for them long time control are unimportant(for me both are important). I think that 3+0 games is part of the picture of the real strength of shredder. Uri
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.