Author: Enrique Irazoqui
Date: 04:13:14 01/30/99
Go up one level in this thread
On January 29, 1999 at 22:25:02, Robert Hyatt wrote: >On January 29, 1999 at 05:30:24, Enrique Irazoqui wrote: > >>On January 29, 1999 at 00:34:54, Robert Hyatt wrote: >> >>>On January 28, 1999 at 21:41:53, James T. Walker wrote: >>> >>>>On January 28, 1999 at 20:35:49, Robert Hyatt wrote: >>>> >>>>>I didn't realize it took a mathematical genius to figure out the next term in >>>>>this 'series'. But it goes like this: the most recently released program will >>>>>pop to the top of the SSDF nearly every time. The reason? It is called "auto- >>>>>232"... >>>> >>>>I'm sorry but that's a little too suttle for me. Would you mind explaining what >>>>you mean? >>>>Jim Walker >>> >>> >>>sure... take your new program, lash it up using auto232 to play against the >>>currently available programs. Tune until you win more than you lose. Then >>>release that version... and pop to the top of the SSDF. Until the next >>>programmer has a chance to get your new program, lash it up with auto232, and >>>test/tune against it... An endless circle... Most commercial guys seems to >>>have multiple machines doing this 24 hours a day. Ed once wrote that he had >>>8 machines doing such non-stop. Others are doing the same I'm sure... >> >>Bob, you keep coming with this idea of yours that is simply not true. Rebel, >>Fritz and Hiarcs have been alternating on top of the SSDF list during the last >>few years, but among the leading professional programs, Genius, Mchess, Nimzo, >>Junior and Shredder didn?t. >> >>It is true that some programmers tune their engines in thousands of autoplayed >>games, but others don?t. For instance, Mark Uniacke, author of Hiarcs, has one >>machine, which makes it rather difficult to autoplay. ChessBase has been >>claiming repeatedly that they don?t autoplay either. Others ask me to check >>during late beta testing how well their programs do, because they can not >>autoplay. So you are not sure at all that "others are doing the same". >> >>Enrique > >First, 'chessbase' doesn't impress me with their honesty. They do impress me as being perfectly honest. > Doesn't it seem just >a bit hard to swallow that they have a 'secret' autoplayer that works but which >doesn't get released on the 'for sale' version? why would someone go to the >trouble to write something they don't sell, and something they don't use? I >don't buy that for a minute. They made their reasons public for a full year now. 1 - They implemented the autoplayer in Fritz 5 because they wanted it tested and nowadays no testers play manually. This is how they went to the first place on the SSDF list. 2 – They didn’t want it public because they didn’t want the competition to cook their book. I think they made a wrong political decision, but I don’t see what’s in it that allows you or anyone to suspect dishonesty, let alone to accuse them of not being honest. Anyone can check Fritz 5.32. The CD-ROM version doesn’t learn and doesn’t access tablebases when autoplaying. The patched versions do. Do you think for a second that CB would have released a version of Fritz 5.32 seriously handicapped in auto232 if they would autoplay with it? >Ed has reported before that he (at one time) had 8 machines busily playing 4 >simultaneous matches against other commercial competition. Chris Whittington >said something similar although I don't remember how many machines he says he >has. Sure. Some do autoplay non-stop. Well done. I would too... >I can't speak for the others, but I have seen this 'leap-frog' for several >years. I only offered a reasonable explanation, regardless of who says they >don't use what. I can _guarantee_ you that no one would write auto232 code if >they didn't use it, as it is a touchy piece of software. Shredder is apparently >doing this as Steffen wrote the windows autoplayer stuff. Again I doubt he did >it for the heck of it... You forget the obvious explanation: auto232 is implemented to make possible the testing and rating of their programs. The SSDF autoplays and so do other testers. This is why. >There's plenty of autoplaying going on whether everyone is willing to 'fess up' >or not, IMHO... Or we have some really bright programmers wasting a lot of time >writing code they aren't using. Speaking for myself, I don't do that... Lots of programmers “wasting a lot of time” is not the case. Their willingness to compete is. I think people in general and you in particular overestimate the resources available to some programmers. Quite a few of them don’t even own the pair of machines it takes to autoplay. Some do, of course, and I see nothing wrong with it. I would do too if I were a programmer. But let’s go back to the original point of this discussion: it is not true that new versions of programs go systematically to the top of the SSDF list and it is not true either that this happens because of their tuning with all the opponents. As I said before, commercial programs like Genius, Nimzo, Mchess, Junior and Shredder didn’t manage to the top during the last years, and of the three that did two don’t even autoplay. I accept a more logical explanation: some go to the top because they are improved. Enrique
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.