Author: Thorsten Czub
Date: 07:55:21 01/01/98
Go up one level in this thread
>I've been following this thread and am listening to both sides >concerning >how search affects playing style. Good. >Here is a question that may help clarify it: Does pure depth affact >playing style? Does Genius on a 386 at 16 MHZ play a different style >of chess than Genius on a fast Pentium? I believe it will play a >very similar style, but just be MUCH stronger. It has been said on >these threads that Genius has always played the same style, even in >the early days of the Lang programs. Right. This is my opinion. The amount of knowledge in the program increased and also decreased over the years. Also Genius4 was an attempt to let it play more risky. Genius5 refuted this attempt because genius4 got weaker ratings in sweden, and only due to the different book and maybe learning stuff it rated higher. >Sometimes on these threads we can get into arguments and not really >be talking about the same things. What does anyone mean by "playing >style?" Can I have the same playing style as Kasporov or some other >great player without being a good player? Nobody says Genius is a weak program because it has a boring playing style. The term boring I use wants to show that it is not always playing the best moves in the 1,3,5,7,9 iterations meanwhile it is SEEING all kinds of threads in the 2,4,6,8 iterations. This leads into a defending and not into an active playing style. When I sometimes miss a good move although I am in a good position, the game develops like a chewing gum. Although I am better, my moves don't kill the opponent instantly. This is exactly how genius wins it's games. Save, it wins normally, it wins them, but it needs time. Genius can smash weaker opponents easily, but if the opponent is at the same strength it does almost nothing and waits for a mistake of the opponent. "Do nothing, but do it well!". >But despite Genius's pruning it is a powerful tactician. And Genius >also has a significant full width BASE at which it misses nothing at >all. ?? Full width ?? But this is what I want to explain. From my point of view it has no full width. I don't think you can say it has a full width. With genius5 e.g. you could adjust the selective search (whatever this slider does) into ranges of 0,2,4,6,8,10,12. Also you can set up level = ply and let it compute 1 ply. Do combinations of the selectivity with combinations of search-depth's and write down the main lines. In Paris we have done this. Of course you have to knock off hash-tables and learning and opening-book (if you are in a book position). But you will get insights about the search. > I don't really want to contradict anyone, but it really seems >likely to me that very little of its style can be explained by the >search. ?!? When I miss good moves in the 1,3,5,7,9 plies it will not affect the style ?? Of course it will ! I don't see why you don't see how this could affect the style. >But let's consider this possibility too: Asymetrical evaluation! Right. I said that I don't know if what I see is caused by asymetrical evaluation or asymetrical search. If you have an idea how to find out, please give me the position and I will test it with all my genius versions including the old dedicated machine ones. >It's not hard to imagine that a programmer who concieves of the >asymetrical search style of Genius might also do similar things >with evaluation. King safety: weight heavier if computer king >in trouble, passed pawns: opponents more dangerous than computers, >Bad pawns: more serious if belonging to computer, etc. In fact >I think this could be tested with some carefully constructed tests. !! I am open for anything ! >Also, doesn't Genius have a full width mode now in some versions? ?? I don't think so. If you put selectivity to 0 this changes nothing to the asymetrical aproach from my point of view. Maybe Genius would see anything later with selective search put to 0. >Turn this off and do some positional assymetry tests. Whatever >the answer is about positional assymetry, I'll bet Genius will not >suddenly become an aggressive attacking player in full width mode. This was the first thing I wanted to find out. I want to see how genius works when it sees it the 1,3,5,7,9 iterations the same things that it sees obviously in the 2,4,6,8 iterations. Pity to say that the communication between Richard and me stopped always in the moment when I asked for a symetrical genius. Maybe he thought I am crazy and stopped talking. Could be. I don't care. A genius that is in the plies 1,3,5,7,9 as strong as in plies 2,4,6,8 could be a very very strong opponent. This was the thing I wanted to find out. >-- Don Look - I don't want to be RIGHT in this topic ! I mean - not right personally. All I want to know is: how it works !! I had a theory I explained myself the programming and strength progress of genius for years. If somebody shows I am wrong and gives a better explanation, I have no problem with it. The only thing I wanted to know is: how does it work. If you or somebody else invent a better theory, I am pleased to accept this.
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.