Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: SSDF Rating list

Author: Dann Corbit

Date: 13:40:00 06/12/01

Go up one level in this thread


On June 12, 2001 at 15:53:20, Uri Blass wrote:
[snip]
>>I disagree here.  I would have guessed 50-70 ELO increase, and within the bounds
>>of uncertainty, the actual result was *well* within expectations.
>
>No
>1)I think that people expect more than 50-70 elo increase from better hardware.

Within the bounds of certainty, the increase could easily be more than 100 ELO.
Did you examine the error bars for one standard deviation?

>Let see the increase from better hardware from the same engine:
>
>Junior5 2432,2534(102 elo increase)
>Hiarcs7.32 2473,2525(52 elo increase)
>Fritz5,32 2478,2551(73 elo increase)
>Mchesspro8 2390,2477(87 elo increase)
>Nimzo99 2446,2489(43 elo increase)
>
>You can see that the average increase is 71 that is slightly more than 50-70
>
>2)People expect better engine so they expect more than 80 elo increase
>
>3)if the average expectation is above what you get it is a disappointment.
>The disappointment can be explained by a statistical error but it does not
>change the fact that it is a disappointment.
>
>Programmers who fail to win a tournament are not going to say:
>"I am not disappointed because it is only a statistical error"

I will.



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.