Author: Vasik Rajlich
Date: 12:28:44 02/13/06
Go up one level in this thread
On February 13, 2006 at 09:22:43, Albert Silver wrote: >On February 13, 2006 at 03:44:32, Arnold Gove wrote: > >>On February 13, 2006 at 03:02:50, Vasik Rajlich wrote: >> >>>On February 12, 2006 at 22:58:57, Timothy J. Frohlick wrote: >>> >>>>It appears that Vasik reworked or deleted some code in Rybka 1.01 13D. It is >>>>strange how a slight change can have a 50 point effect in chess strength. I >>>>suffer from gout and can assure you that a sore big toe is only 0.1% of the body >>>>mass but can make you crawl rather than run. >>>> >>>>TJ >>>> >>>>PS Now to play Rybka 13B vs Rybka 13D. I have no doubt that on my slow 1500 MHz >>>>AMD with 512 MB RAM in 32bit Win2000 Pro that 13D will be demolished. >>> >>>Tim, >>> >>>it is almost certainly statistical noise. Various Rybka Beta 13c and 13d >>>candidates played >5000 3-minute games agains Beta 13b. The final one played >>>>1000 games. >>> >>>Vas >> >>That, Vas, is the problem with a creeping rollout - as with frequent updates >>through a subscription service. There will be insufficient (certainly reduced) >>periods for testing between versions, placing supposed incremental improvements >>in reasonable doubt. >> >>AG > > >Bear in mind, Vas tests his versions quite a lot before releasing them, though >other may confirm or debunk his results of course. > > Albert Note that the same issues apply to changes which are not "released". There is no perfect way to measure progress, all you can hope for is to go forward more than backward. Vas
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.