Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Nimzo99 MMX - Hiarcs 6 P90 SSDF game 12/20 1-0 Now: 10 - 2

Author: Tony Hedlund

Date: 10:47:36 06/01/99

Go up one level in this thread




>I wasn't saying new products defies reasoning to test older models, but rather
>once a newer model has proven itself superior to the older model - that would
>indicate further testing of the older model to be of little, if any,
>significance.

That's just not true. Read Professor Elo's book.

>I make the above point a major issue because I prefer to see program A tested
>against program B on the same hardware. It is written in the manuals of both
>Hiarcs7 and Fritz 5.32 that the speed of the processor affects the stength of
>the programs. I have no doubts about that.

Neither do I.

>Therefore, I feel that if only the
>best representatives from each manufacturer were tested - this may allow the use
>of equal hardware because there wouldn't be a need to test unnecessary programs.

That means that every tester must buy two new computers every time we change
hardware. And to get a rating for a new entry we must play it against entries
with established ratings. That's no way around that.

>I know that SSDF claims that testing on equal hardware isn't necessary to get
>true ratings, but I do have a problem with that.

You don't believe in Arpad Elo's method of calculating ratings? Odd, FIDE use it
to. You should talk to them also.

>It is most logical to me that
>testing be done on equal hardware to eliminate as many variables as possible and
>get the most accurate rating possible.

About logic, take a look at my answer to one of your earlier posts, that one you
didn't respond to.

>Mel

Tony



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.