Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: I plan to settle this.

Author: Robert Hyatt

Date: 10:01:48 06/14/01

Go up one level in this thread


On June 14, 2001 at 10:42:19, Mark Young wrote:

>
>I don't know. I'm I missing something. Where did Bob apologize?


Just look for this:


::::::::::::::::::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::::::::::::::::::::::

Below...



>
>1.	It is not acceptable to put quotes around words that one has never used.
>2.	This is a typical Hyatt deflecting tactic he has used on many occasions over
>the years. He purposely misstates ones position to take attention off the topic.
>(It crosses the line when he starts quoting me one something I never said.)
>3.	Bob has no Idea how the ratings at www.chessmetrics.com were generated, Bob
>state he knows how they were generated.



Bob didn't state any such thing.  I stated, quite clearly and distinctly,
that it is _impossible_ to predict ratings for players of earlier years.
Ratings are based on a rating pool of players.  I will state it again.  If
you don't have a real rating for a player that was active in (say) 1890,
then there is _no_ way to compute a rating for him today.  Nor is there any
way to compare his old rating (if he had one) with the ratings of today.

Different rating pools make Elo numbers non-comparable.

That is _all_ I said.  I don't know _how_ they computed old ratings.  It doesn't
matter.  However it was done is scientifically invalid, which is what counts.










?That implies that you can take ratings
>from today, compute ratings for players active 20 years ago, then take those
>ratings, and back up another 20 years, and repeat he above, until you go back
>far enough to compute one of these whacko ratings? This is just dead wrong. See
>web site.
>4.	This has nothing to do with what was being discussed in the thread. (see 2)
>5.	a. The only points in the thread that I have made is Ratings is the only
>measure we use in chess, b. Ratings can be generated for any pool of players
>past or present because we know the outcome wins, losses and draws of the games
>they have played.. Neither of which Bob can refute (see 2)

I can refute that point all day.  I take pool one to be 10 players of GM-level
today and beyond.  I take pool two to be 10 players of class C level only.  I
let both pools play to establish their ratings.  Both pools will be rated
roughly identically.  But what happens when a 2000 player from pool one meets
a 2000 player from pool two?  No question.

So "ratings" can be generated from any pool, but the pool must be a "common"
pool... not two disjoint pools.  Which means ratings prior to the acceptance of
the Elo system are all "hypothetical ratings" with very little meaning.







>
>
>>
>>                                          Albert
>>
>>
>>>Nothing
>>>new for you. I will keep posting, you do what you wish.
>>>
>>>>
>>>>
>>>>
>>>>>
>>>>>
>>>>>
>>>>>>
>>>>>>Publishing a rating list that spans 200 years implies the ratings are
>>>>>>absolute.  Otherwise the numbers are meaningless.  Which they actually are.
>>>>>>





::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::;
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

Just read the _next_ sentence, which was quoted by you when you responded..

I think that will clear it up _nicely_.


You accused me of posting without "reading first".  Sound like a pot and
kettle issue?  Since you _obviously_ didn't read the sentence below???









>>>>>>If I misunderstand what you provided below, then accept my apology for
>>>>>>interpreting it incorrectly...
>>>>>>



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.