Author: Uri Blass
Date: 13:26:45 09/12/05
Go up one level in this thread
On September 12, 2005 at 15:58:39, Dann Corbit wrote:
>On September 12, 2005 at 13:34:08, Fernando Villegas wrote:
>
>>Dear Amir:
>>You say:
>>"....Of course playing strength is of utmost importance,
>>more than any other measure of a chess program. This is obvious from the
>>professional and academic point of view...."
>>
>>Yeah, right: from the academic and professional point of view.
>>What about those that does not care a shit about the academic and profesional
>>point of view?
>>
>>I can understand that you, a known and reputed programer, on of the very best,
>>can be delighted for such things as "this engine plays now 12 points elo
>>stronger than before". But believe me that, to me and many others, that we
>>cannot care less. 2612 instead of 2600? So what? In the times of engines playing
>>in the range between 1600 and 1800 I DID care and it was sound that I cared at
>>all. Those 12 points made a difference. Better if 50. 50 points, in those years,
>>were the difference between a really dumb engine uncapable to give fun and a
>>decently strong engine capable of delivering the merchandise.
>>
>>Even in the times of 2000-2200 elo points engines or machines, the 50 points
>>issue was important for players like me, no pro, not academics in the field of
>>AI.
>>
>>Not now anymore.
>>
>>For us it is matter of better trainning facilities, better guis, more fun, more
>>reasons to buy the thing. So no, I will not give you that break :-)
>>
>>There is much that can be done in this field and I do not care if chessbase will
>>go into that because they evade a lose of Elo competitivity OR if they really
>>are engaged on that for the reasons they say.
>>
>>I only ask you to consider this, totally different point of view, beyond the Elo
>>points.
>>
>>And please, do not say to me that I just must go to Chessmaster and leave you
>>all in peace with your professional machines.
>
>The friendly sort of features that you speak of are not really even connected to
>the engine very much.
>
>There are lots of weak engines if you want to play them (and I genuinely enjoy
>playing them from time to time).
>
>If you want the best analysis, then a stronger engine is better than a weaker
>one.
>
>If you want a message like "You pushed your queen in front of that pawn, and it
>is not a good idea to do that with no compensation" {and I do like messages like
>that also, as long as they don't add things like "... you blubbering patzer..."
>or "... you really ought to try another game. Have you considered checkers
>against a three year old?") then those sorts of things do not really depend on
>the engine much, but are a function of the GUI as it translates the analysis the
>engine has performed.
>
>"We think engine strength is not important anymore." is another way of saying
>"We don't think we can compete on strength anymore."
>
>"We want human-like play." is a pure cop-out because we can't really even define
>that.
>
>"We want it to play at reduced levels." is also a cop-out because there are lots
>of engines that will make dumb mistakes if that is what you would prefer.
>
>"We want it to play like a patzer in the same way as a human plays like a
>patzer." is more hogwash. No two patzers are going to play the same way. One
>guy doesn't understand king safety and another guy leaves hanging pieces. Both
>of them are patzers but they don't play alike at all.
>
>If you want weak engines, there is already a big basket of free ones.
>
>If you want human like analysis, then ChessMaster has a nice model for that.
>Other interface designers can copy their idea.
>
>If you want good analysis, then you need the strongest engine. For those who do
>not care about the best analysis, then any engine will do, I suppose.
>
>And there will be some who want the strongest engine possible just for online
>play to get a big rating at FICS or some other online game place. For these
>people strength does matter. They will even buy $6000 computers just to have
>the hottest hardware for that purpose.
>
>At any rate, "We are no longer interested in strength" is a pure smokescreen.
>You can bet if they won the WCCC hands down they would be blowing a trumpet and
>beating a drum. The fact that they (to some degree) no longer compete shows
>that they are afraid of losing.
>
>I salute the commecial engines that had the guts to show up and compete at the
>world championships (Shredder, Junior, Deep Sjeng) and give a golden raspberry
>to those who hid in the weeds (Fritz, Hiarcs, ChessMaster, Chess Tiger {sorry
>Christophe}) and a double golden raspberry to those that claimed improving
>strength is a waste of time afterwards.
>
>In the case of Hiarcs, ChessMaster {the King engine} and Chess Tiger, perhaps
>the problem was no 'Deep' version and so they knew their goose was cooked before
>the game started (who wants to give away 150 Elo before the starting gun
>fires?). But Fritz has a deep version.
I do not think that it is 150 elo.
4 processors of junior or Shredder against one processor of Fruit give probably
less than 100 elo.
4 processors may do the program 3 times faster and being 3 times faster at long
time control is probably less than 100 elo improvement.
Uri
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.