Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hydra???

Author: Robert Hyatt

Date: 12:31:39 02/14/04

Go up one level in this thread


On February 14, 2004 at 12:15:56, Ricardo Gibert wrote:

>On February 14, 2004 at 10:28:00, Robert Hyatt wrote:
>
>>On February 14, 2004 at 04:45:12, Ricardo Gibert wrote:
>>
>>>On February 13, 2004 at 15:56:00, Robert Hyatt wrote:
>>>
>>>>On February 13, 2004 at 15:54:18, Mike S. wrote:
>>>>
>>>>>On February 13, 2004 at 15:26:23, Robert Hyatt wrote:
>>>>>
>>>>>>You can safely assume that those ratings are totally meaningless, if you are
>>>>>>thinking within the context of human FIDE ratings...
>>>>>
>>>>>Within the context of human FIDE ratings, a previous version of Hydra, aka
>>>>>Brutus, has achieved a tournament performance of 2765 :-)
>>>>
>>>>Tournament performance != Fide rating, unfortunately..  Otherwise Kasparov might
>>>>have hit 3200 more than once.
>>>
>>>You're exaggerating. That would mean that the average rating of his opponents
>>>was about 2800 and he would have to win every single game. This has never been
>>>possible.
>>
>>Did you see the key words "might have hit"???
>>
>>There have been multiple 3000+ tprs.  Given the rating inflation of today, it
>>won't be long before there is a 3200+ tpr.
>>
>>>
>>>For example, a 4 man tounament today with the 4 top players would consist of:
>>>
>>>  Kasparov 2831
>>>  Kramnik  2777
>>>  Anand    2766
>>>  Svidler  2747
>>>
>>>The average rating of Kasparov's opponents would be 2763, so he could get s
>>>performance rating of 3163 tops. I don't believe such a tounament ever happened
>>>and certainly not twice e.g.
>>
>>I'd claim 3163 is "pretty close" to 3200...
>
>
>You can claim it, but a tournament in the past where Kasparov performed over
>3100 you will not find. That is the *point*.
>

The only *point* here is that "might" != "definite fact"...

I don't see why that wasn't clear...

Picking a single TPR does not prove a program is 2800+ as the Paderborn
crosstable seems to suggest...

The fact is that more than one GM has produced a TPR _way_ over his actual best
rating...  using it to conclude anything is not particularly sound.

>
>>
>>
>>>
>>>Top Tournament performances
>>>   The highest winning percentages for tournaments
>>>   where at least four of the top ten were present.
>>>
>>>   1872 June-July, London, won by Steinitz 7.5/8, 93.8%
>>>   1930 Jan.-Feb. San Remo, won by Alekhine 14/15, 93.3%
>>>   1862 London, won by Anderssen 11/12, 91.7%
>>>   1876 Mar.-Apr. London, won by Blackburne 10/11, 90.9%
>>>   1873 July, Vienna, won by Steinitz 22.5/25, 90%
>>>   1867 June-July, Paris, won by Kolisch 21/24, 86.4%
>>>   1883 Apr.-June, London, Zukertort 22/26, 84.6%
>>>   1994 Linares, Karpov 11/13, 84.6%
>>>   1899 London, Lasker 22.5/27, 83.3%
>>>   1867 Dundee, Neumann 7.5/9, 83.3%
>>>
>>>As you can see, Kasparov does not even make the list. I think a 3100 performance
>>>would be a record.
>>>
>>>Check out http://www3.sympatico.ca/g.giffen/records.htm
>>>
>>>The most astonishing performances were in match play by Fischer in his now
>>>legendary drive to the world championship match with Spassky. 6-0 vs Taimanov
>>>and 6-0 vs Larsen. This will be very difficult to improve upon.
>>>
>>>>
>>>>>
>>>>>http://www.chessbase.com/newsdetail.asp?newsid=1131
>>>>>
>>>>>(this was 40/2h)
>>>>>
>>>>>Regards,
>>>>>Mike Scheidl



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.