Author: Peter Skinner
Date: 16:37:19 03/02/04
Go up one level in this thread
On March 02, 2004 at 16:13:03, George Tsavdaris wrote: >On March 01, 2004 at 20:37:00, Peter Skinner wrote: > >>On March 01, 2004 at 15:10:56, Dann Corbit wrote: >> >>> >>> # Name 1 2 3 4 5 6 7 8 9 10 11 12 >>>13 14 Score Buch Sommb >>>------------------------------------------------------------------------------------------------------------- >>> 1 Ruffian_202 **** =1= 01= =01 =10 0==1 1=0 11=1 111 111 11= 1110 >>>0=1 1=11 30.0/43 882.5 595.50 >>> 2 Ruffian_105 =0= **** 1=0 === ==1 =11 1111 111 1010 001= =01 =10 >>>1111 111 29.0/43 892.5 574.00 >>> 3 Ruffian_210 10= 0=1 **** 1=0 =001 ==0 011= 110 1=11 0001 110 1=1 >>>111 111 26.5/43 910.0 526.25 >>> 4 Ruffian_101 =10 === 0=1 **** === 00= 1=1 101= 101 0101 1110 111 >>>01=1 101 26.0/43 898.5 524.75 >> >>This is almost bang on with the results that I have attained. Version 1.0.1 in >>my testing finished ahead of 2.1.0 by only a half point. So those two just >>flopped in our testing. >> >>My games we G/15 and G/30. It seems that Ruffian 1.0.5 and 2.0.2 are just about >>equal in strength and 1.0.1 and 2.1.0 are equal in strength. >> >>I read a post on another forum the other day where someone did some more in >>depth testing is came to the conclusion that it is possible that 1.0.1 has been >>optimized and renamed 2.1.0 and the same goes for 2.0.2 and 1.0.5. > >Mmm... interesting. I would search a little about this. > >> >>If you take certain test positions and analyze them with the two similiar >>versions almost 99% of the time the same variation happens. If there was a >>"huge" strength improvement like some would have us believe that would not be >>the case. >> >>Unfortunately the post was removed as the administrator felt the post was >>attacking the author or accusing his of fraud. While the poster probably was, >>there is a hint that this is exactly what could have happened. >> >>Statistics do not lie... > >Yes but statistics on the above tournament don't say that Ruffian 1.01 ~= >Ruffian 2.1.0 and Ruffian 1.05 ~= Ruffian 2.0.2. > >> >>Peter I have tested, and I have read all the testing others have done, and the same data always seems to come forward: 1. Ruffian 1.0.1 finshing within a single point of 2.1.0. Usually is happens to be a .5 point differnce. 2. Ruffian 2.0.2 and 1.0.5 seem to finish within 1 to 1.5 point of each other. 3. Very few test results have shown 2.1.0 or 1.0.5 to be stronger than 2.0.2, and 1.0.5 respectively. I know in the Ridderk tournament 1.0.5 did finish lower than 1.0.1, but that was only by 4 points.. luck could have been a contributing factor. 4. When analyzing positions with those 4 versions, 2.0.2 and 1.0.5 come out to the same result, just 2.0.2 does it quicker. Same goes when analyzing with 1.0.1/2.1.0. 5. Personally I don't believe Per-Ola would do something like this, but the data does speak volumes. It is hard to just toss it aside. I do want to go on record and state that I don't believe this to be the case, or rather I am seriously hoping this is not the case. It would constitute a major fraud.. Peter.
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.