Author: Ingo Bauer
Date: 03:07:18 08/03/04
Go up one level in this thread
Hi >On August 03, 2004 at 05:10:45, Ingo Bauer wrote: > >>Hi >> >>>http://f50.parsimony.net/forum200336/messages/2551.htm >>> >>>tried these styles... >> >>What you propose is to use different styles vs. different engines. > >Since the original style is not working, a different style is the only >alternative. If the style is not working ONE different style makes sence but 1. The style seems to work against Fritz (till now) so it doesnt seem to be that bad and 2. Why did the programmer choose a style that is not working (I would call that a bad beta testing!) >>I think this >>is pretty worthless. > > >? > >why ? > >when i make a special experience that is bad, i do not try this again 300-500 >times. > >this senseless testing of failing setting is IMO not helping. We have two differnt things here. I agree with you if I see that programm A is loosing against Program B with one to 19 I will not do another 480 games to be sure. But to test setting A against Engine B and setting C against engine D is not the solution because you never get a point where you can say engine X ist the best engine. You only get that setting A(of engineX) is better against the default setting of engine B ... It never ends this way. >of course. always when you change something in a setting it changes anything. >this is true for all programs. >therefore the right styling is part of the product or of the preparation >in a tournament. For a single tourney ok, but you never get a conclusion about the "better" engine (as the SSDF is doing it (ok there are some weak points in their concept but it is the best we have!) But it is unfair if you do it in a tourney only for your prefered engine vs. a default opponent. If a "real" tourney happens everyone is doing it for his engine, then it is fair again. Optimizing a single engine against a "default" opponents is not fair in my eyes. In a tourney where the programers do not prepare against a specific opponent because this opponent is not the same as the default one. Everyone is trying to get the best "overall performance" out of its engine. Again: As long as everyone is optimizing it is ok but not chose a single engine and use this one. Example: What about a car race between a Citroen C5 and a Golf V, and VW is allowed to tune the Golf until it is really the better car. What do you think how big the interest in such an event would be? No way, either all cars can be tuned or none! >and you believe comparing engines with a random style done by the programmer >makes more sense. Pro deo or Rebel or XYZ comes out with a style. > >who says this style works ? >God ? In this case yes, the programmer is god. He choosed the setting because usually he thinks these are the best overall settings! >The styles that come with the progams are only an effort. they are not the last >word. ed god XY % with his style, he said. > >HE said. thats the important information in the sentence. > >Ed said he had xy % + in relation to the older rebel programs. > >But what does this mean ? did ed test his latest Pro Deo style versus Shredder8 >? > >And did it get the xy % against this program ? Again, the programer choosed these setting as the best ones. If you change them pls change the other settings in your tourney to fit best against each single opponent as well or at least use one setting against ALL opponents. But it is hard to do that properly so I prefer the "out of the box settings" over any particular setting. Or use at least one setting against ALL opponents (eg. the chessmaster setting - thats why I asked for your best setting) What you are doing is to say that the default setting is not nessesarily the best setting but you test against a "possibly bad setting" of other engines. Should'nt they be "improved" against your special settings? This seems to be a bit unbalanced! Bye Ingo PS: It seems the default settings are the best overall setting we have right now!
This page took 0.01 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.