Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: who will be the 1st program to hit the 2800-2900 rating barrier?anyone?

Author: Uri Blass

Date: 03:12:52 01/17/03

Go up one level in this thread


On January 17, 2003 at 05:41:50, Sune Fischer wrote:

>On January 17, 2003 at 01:07:10, Dann Corbit wrote:
>
>>On January 16, 2003 at 21:56:45, stuart taylor wrote:
>>
>>>On January 16, 2003 at 16:47:53, Dann Corbit wrote:
>>>
>>>>On January 16, 2003 at 16:16:02, david wight wrote:
>>>>
>>>>>please help with your thoughts.
>>>>
>>>>I, Dann Corbit, as a human player, am at 2900.  Of course, that puts Kasparov at
>>>>2900+1900, = 4800 (roughly speaking).
>>>>
>>>>The numbers are arbitrary and have no meaning in absolute value.  The only
>>>>purpose of ELO is to demonstrate strength levels via DIFFERENCES.
>>>>
>>>>HTH
>>>
>>>So you DO admit that there are playing strength levels in chess, just that it
>>>can't be gaged exactly. One year 2800 might be a certain level of game, and the
>>>next year it might take about 3500 to play the same game!
>>
>>The Elo formula is useful to predict outcomes over a broad range.  If a
>>measurement is calibrated based upon data in a single pool, then the difference
>>can be used to predict future outcomes (on a broad basis -- not individual
>>measurements, of course).
>>
>>Not sure how that connects with your question/statement.
>>
>>My point was that Kasparov can be 2800, 3800, 50,000,000 or whatever we choose.
>>As long as someone 300 ELO beneath his ability is given (Kasparov's ELO-300) the
>>calculations will be the same.
>>
>>It is obvious that Elo figures drift over time.  An Elo of 2500 from 1875 has
>>little connection with an Elo fo 2500 today.
>
>While true in principle, there is 1 fixpoint commenly used: beginners rating at
>1000 Elo. If you also put random play at 0 Elo you have a good (fixed) scale.
>
>-S.

People who play their first tournament are often not beginners so you cannot use
this information to compare rating of today players with the past players.

Putting computers that play random moves as players in tournament and having
more computers in the rating list(for example computer that does 3 ply search of
all the moves and choose a random move that is not losing more than 0.4 pawns
relative to the best move) can help but unfortunately people object to
participation of computers in tournament so it is not going to happen.

The sad truth is that
people in general are not interested in correct rating and this is the reason
for the stupid rating system when you can lose rating from winning a game.

The rating system was also changed such that people can calculate their rating
because people in general prefer wrong rating that they can calculate and not
right rating.

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.