Author: Bruce Moreland
Date: 18:56:25 09/27/99
Go up one level in this thread
On September 27, 1999 at 17:33:00, Frederic Louguet wrote: >As I created the LCT II test, I believe I can clarify a few points about it. > >1) I created it in 1994. A *very* long time ago. > >2) It was fun at the time to create a test which could give a reasonable >approximation of the capabilities of a chess program, and I enjoyed it. I chose >positions with no mistakes, and I think they are all interesting, in one way or >another. It took me a long time, but I was not as busy then as I am now... > >3) I keep hearing here and there that the LCT II test was made to closely >correlate the SSDF results. It is simply not true. In fact, the LCT II predicted >some SSDF results BEFORE they were published, and I was quite surprised myself. >Some other people too, as they began to overestimate its usefulness. > >4) The LCT II gives an Elo rating (Elo, not elo, see the difference), because it >was funny to translate numbers into something more meaningful. Of course it has >nothing to do with a real elo. > >5) Like any test, the LCT II is a tool to evaluate the capabilities of chess >programs on a few chosen positions. Nothing more, nothing less. > >6) In 1995, I began to write Chess Wizard, and soon I realized that no test >could replace real games. In fact, I think that developing a chess program and >measuring its progress by running test suites is meaningless, and can even lead >to bad decisions regarding its implementation. Since early 1997, I did not use >ANY test to measure the progress of Chess Wizard. > >7) Last year, I had a funny version of Chess Wizard which did 2705 on the LCT II >test (for example, it found Vaisser Cxh7!! in 55 seconds on a Pentium II 300). >But in fact, it was at least 100 real elo points WEAKER than the best version... >Overloaded with extensions ! > >I hope it clarifies the situation. To create the LCT II was an interesting, and >fun, experience. It was never intended to be taken like the Word of God on chess >playing strength. Unfortunately, some people never understood that. I would just >like to make clear to anyone that the LCT II can not be substitued to real >games, and that it was never intended to be. And it is the same for all tests, >because they only measure the capability to exploit winning positions, not to >slowly build them. > >If you want to be brilliant, one move is enough. If you want to be strong, all >of them count. I like it a lot, it's a fun test. I use it as a sanity check of tactical strength, and I like the positional problems as well. bruce
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.