Author: George Sobala
Date: 04:10:02 10/07/04
Go up one level in this thread
On October 07, 2004 at 05:57:13, Odd Gunnar Malin wrote: >On October 07, 2004 at 03:27:12, George Sobala wrote: > >>On October 07, 2004 at 03:06:12, Kurt Utzinger wrote: >> >>>On October 07, 2004 at 02:27:56, George Sobala wrote: >>> >>>>First result: I let Ruffian 2.0.0 (Shredder's nemesis) run the gauntlet >>>>overnight against Shredder 8 UCI and Shredder Classic 1.1 at 3'+0" blitz. >>>>Ruffian used its own book, 64M hash, Centrino 1.6GHz, both Shredders used the >>>>downloadable big book from shredderchess.com. I used the Classic GUI. >>>> >>>>Results: >>>> >>>>Ruffian v Shredder 8 : 16.5 - 13.5 >>>>Ruffian v Classic 1.1 : 14.5 - 15.5 >>>> >>>>Interesting. >>> >>> Not interesting in so far as results >>> are within the usual statistical error >>> margins. Much too less games played >>> to conclude something. >>> Kurt >> >>That is statistical nihilism and is simply not true. >> >>I can statistically conclude the following: >> >>1. Shredder 8 is not HUGELY better than Classic 1.1 against Ruffian2 at 3+0 >>blitz. >> >>2. There is more than a 50% chance that Classic 1.1 is STRONGER than Ruffian2 at >>3+0 blitz. >> >>... which I find interesting. :) > >Did you watch the games, was it a lot of random results in drawn endgames? >I usually prefer to have increment when I test with one timecontrol, eg. 5'+10 >blitz to get something usefull for comparing analysis strength. > >Odd Gunnar No, I didn't ( I ran it overnight), and I fully agree that this is a problem with blitz time controls with no increment. They two Shredders are currently playing a 30 game match against Deep Sjeng 1.6 at 5'+5" and the current score is: DS1.6 v S8: 1.0 - 7.0 DS1.6 v Classic1.1: 5.5 - 2.5 Even more interesting! :)
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.