User talk:Tralala
From Wikipedia, the free encyclopedia
[edit] MTD(f) vs. Negascout
I saw that you added to the MTD(f) and Negascout articles that although MTD(f) is theoretically superior to Negascout, it has "practical issues" (which I assume make it suboptimal). I was just wondering what those issues are. Thanks! --Arabani (Talk ∞ Contribs) 19:25, 18 January 2006 (UTC)
- I'm not a programmer but there are numerous discussions on that topic in a chess engine authors board (CCC). The board requires registration therefore I didn't include it as a reference. The post I'm referring to is here: http://talkchess.com/forums/1/message.html?480538 Tralala 23:08, 18 January 2006 (UTC)
-
- I read the post you're referring to, and, quite honestly, it's pretty vague. It seems to me that the poster's reason for saying that Negascout is superior to MTD(f) is that the "proof" that MTD(f) is better relies on the use of weak programs that ignore practical issues that real programs can't ignore. Yet he fails to specify what those practical issues are, which completely defeats his argument. Anyone can make blanket statements like that: for example, I could say that Pentiums perform better than Athlons, but without any real data, that's a meaningless statement. I'm kind of uncomfortable with using that post as a source without any further information to back it up. Even if academic papers do make optimistic assumptions, etc., at least they're backed up by some form of data. The assertion that MTD(f) suffers from practical issues without stating what those issues are just isn't as credible. (Just a note, I'm not nitpicking. I'm quite interested in what the issues are.) --Arabani (Talk ∞ Contribs) 09:24, 19 January 2006 (UTC)
-
-
- I understand your concerns. In fact I agree with you that the one post I was referring to is not a sufficient statement to question the superiority of MTD(f). However the question of MTD(F) versus PVS (Negascout) has been discussed several times in several boards and always was the accepted opinion that it is not superior to PVS. Since the search engine of CCC is not working correctly I can't find the relevant posts right now but I just asked for statements about that what will hopefully give some more data on that. Perhaps it helps if you express your genuine interest in the topic as well. Tralala 20:04, 19 January 2006 (UTC)
-
- The only "practical issue" with MTD(f) is having a working transposition table, which is not an issue at all. 70.108.42.228
