Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: My proposal to SSDF: decrease level by 100 points

Author: Chris Carson

Date: 11:49:53 07/28/00

Go up one level in this thread


On July 28, 2000 at 10:25:14, Bertil Eklund wrote:

>On July 28, 2000 at 03:49:58, blass uri wrote:
>
>>On July 28, 2000 at 01:16:36, Lanny DiBartolomeo wrote:
>>
>>>On July 28, 2000 at 00:58:14, Jouni Uski wrote:
>>>
>>>>I made comparison between Carson's latest human list and SSDF to clear
>>>>possible level difference. But it's very difficult, when there are not many
>>>>exactly same software and hardware combinations. I removed multiprosessor
>>>>results and made assumption, that difference between 200MMX and 450 Mhz is 79p
>>>>(for comp-comp, sure not for human-comp). Here's some figures:
>>>>
>>>> Program   average speed   Against humans   SSDF rating         difference
>>>>
>>>> Rebel         431Mhz      2549/33 games    2594 (200MMX+79p)     -45
>>>> Fritz6        500Mhz      2457/15          2721 (450)           -264
>>>> Hiarcs6       272Mhz      2573/20          2518 (200MMX)         +55
>>>> Shredder      500Mhz      2495/8           2676 (450)           -181
>>>> CM5/6K        375Mhz      2521/14          2653 (200MMX+79p)    -132
>>>>
>>>>This means, that average difference (Against human - SSDF rating) is -85p.
>>>>(Weighted difference for all 90 games).
>>>>
>>>>I think this confirms general feeling, that SSDF must drop level by 100 points.
>>>>Counting these multiprosessor Junior/Fritz may still lower level.
>>>>
>>>>I this this kind of list is much more reasonable than current:
>>>>
>>>> 1  Fritz 6.0 128MB K6-2 450 MHz          2621
>>>> 2  Junior 6.0 128MB K6-2 450 MHz         2589
>>>> 3  Chess Tiger 12.0 DOS 128MB K6-2 450   2571
>>>> 4  Fritz 5.32 128MB K6-2 450 MHz         2554
>>>> 5  Nimzo 7.32 128MB K6-2 450 MHz         2553
>>>> 6  Junior 5.0 128MB K6-2 450 MHz         2526
>>>> 6  Hiarcs 7.32 128MB K6-2 450 MHz        2526
>>>> 8  Crafty 17.07/CB 128MB K6-2 450 MHz    2524
>>>> 9  Nimzo 99 128MB K6-2 450 MHz           2523
>>>> 10 Hiarcs 7.32 64MB P200 MMX             2477
>>>> 11 Fritz 5.32 64MB P200 MMX              2475
>>>> 12 Chessmaster 6000 64MB P200 MMX        2474
>>>> 13 Hiarcs 7.0 64MB P200 MMX              2460
>>>> 13 Fritz 5.0 PB29% 67MB P200 MMX         2460
>>>> 15 Nimzo 99 64MB P200 MMX                2449
>>>> 16 Junior 5.0 64MB P200 MMX              2437
>>>> 17 Nimzo 98 58MB P200 MMX                2425
>>>> 18 Rebel 9.0 47MB P200 MMX               2420
>>>> 19 Hiarcs 6.0 49MB P200 MMX              2418
>>>> 20 Rebel 8.0 51MB P200 MMX               2410
>>>>
>>>>Jouni
>>>
>>>Yes, but if you leave Fritz6 on top it is too inflated compared to human games,
>>>  and dropping hiarcs rating would be a problem as well since it earned its
>>>rating 2573, it seems like the damage is done and you need a separate ratings
>>>list for comps vs human.  I do not think you can just lower the ratings straight
>>>across the board.
>>>  Lanny
>>
>>I think that the leader of the ssdf(if it is not for a short period) will always
>>be inferior against computers(unless there is a big difference of more than 100
>>elo ssdf rating).
>>
>>The best players will always buy the number 1 in the ssdf and prepare against it
>>so programs number 2 and number 3 can get better result against humans because
>>humans are less prepared against them.
>>
>>Fritz is leading the ssdf most of the time in the last years.
>>
>>Some other programs like Junior and tiger and chessmaster could be the leaders
>>only for a short time.
>>
>>This is the reason that people buy more fritz and this is the reason that fritz
>>is inferior against humans(I talked with one of the players in the israeli
>>league and he told me that he prepared against Fritz but did not prepared
>>against other programs).
>>
>>Uri
>Hi!
>
>I fully agree with you because i have spoked with a lot of strong players and
>almost everyone plays and analyzes with Fritz all day long.
>
>In the next list august 4 or 5 we are going to subtract maybee 100 points (still
>under discussion)
>
>Bertil SSDF

That just shows that a lot of players respect the work you do
to produce this list as well.  I have heard the same as you and Uri
about Fritz being used a lot.  Great work on compilation of the
list and all the hard work and research to keep it calibrated for
both computers and human results.

This is still the most valid/accurate/reliable source of information on
computer chess strength.

Best Regards,
Chris Carson



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.