Author: Robert Hyatt
Date: 18:49:42 06/12/98
Go up one level in this thread
On June 12, 1998 at 21:05:42, Thorsten Czub wrote: >On June 12, 1998 at 18:37:19, Robert Hyatt wrote: >>I am trying to point out the fallacy in your reasoning. What is >>positional >>to you, is tactical to Cray Blitz and Deep Blue, because they can see >>much >>deeper. > >If cstal would have 1/100 of the hardware and energy-power, it would >search and play as strong. that's easy to test... take on Crafty on ICC... I am doing about 300K nodes per second... Cray Blitz is another 10-20X faster... which means you have more like 1/30th the power of Cray Blitz already. You seem to think that we are all about hardware. You are wrong. Crafty (and CB) is *so much more* than just a fast piece of hardware. In any case, you already have more than 1/100th the power, but you aren't anywhere near as strong... > >> What is positional to CSTal, might well be tactical to fritz or >>Crafty because they search much deeper. But *no* human wins a game >>without >>utilizing tactics. > >The power of the human brain is, that these processes were mainly put >into unconsciousness. When I drive my car, i drive it completely >different than I have done it when I was 18 and I was a learner. >Most of my JOB while driving is done without myself recognizing what I >am doing. I can listen to music in the radio, or to my neighbour telling >me his ideas about communism. And I am still driving. >This is different to my first driving lessons, where I drove the car >very weak, BECAUSE i drove it like a computer: considering about >anything with my will. >Nobody who drives this way, will become a good car-driver. >And nobody playing like a computer, will become a good chess-player/car >driver. >Of course a fast-driver. But still a stupid driver. >Ok - only my idea. We are where we always end. You have your point, I do >have my point. Why not concentrate on the games instead ?! > >> Intuition is not enough. We all "calculate" at >>times >>to see if a move wins or loses. > >If we measure how many hours of the day you are willingly doing >something, and how many hours your unconsciousness drives you, you will >be surprised... but when I play chess, I am not unconscious, nor daydreaming, nor working on my boat... I am *playing chess*. And I am calculating like mad... > >>The question is, when you live by intuition alone, as you suggest, > >No - i don't suggest to ONLY live by intuition. CSTal e.g. has a search >!! >So it does not only trust on evaluation. But it uses the search not for >stupid moves. there you tripped over your own "body part". CSTal looks at *plenty* of stupid moves. The typical GM looks at about 100 moves to make up his mind. How many times that does CSTal look at? I believe it is searching 5K nodes per second... so in 5 minutes that is 1,500,000 nodes... of which almost *every one* is stupid. So that statement won't cut crap here, it is wrong, completely wrong. If you want to say you look at fewer stupid moves than we do, that's a bogus argument. Of course you do. You are *far slower*. I can look at fewer stupid moves too, if I wanted to... a loop here and there and I can get down to the point where I look at only 10% of the stupid moves you look at. Of course, I'd get smashed just like you do then... >I am not saying you should forget to think. I am only saying that you >will never reach your target with your strategy. Because the way you are >doing it is unbalanced. > We'll see. Based on current results, if it takes me 100 years, it will take your approach > 100 years... as I see nothing so far to suggest that CSTal is "closing the gap" in any way. It is a worthwhile approach to experiment with. But I'll bet that as time goes on, it is either going to have to rely more on search and speed, or get left further behind... > >> how >>long before you die by intuition. > >Even creatures who forget anything they have done willingly after 15' >have a good life and have fun... :-)) But computers are much more >primitive. > >> IE I'll bet that for every such >>attack >>against genius that CST wins, there are multiple games that it will >>lose. > >Right. And we normally practise this multiple losing on championships ! >We ask the other participants to look away until I have a game on board >that will work. So we have done in the past, and we will do in the >future. Sometimes we pay the opponent program, or the other >participants. Costs 1000$ to pay the ICCA ! >It is very difficult Bob to play multiple games on championships. >Especially against Genius with black ! I hope you try once !! >In opposite to you I was that fair that I showed Ossi before all the >games we played against Genius, and that the engine won. And no matter >which opening was chosen. He said: nice - so we will chose a quiet >opening. It did not work. >We work on the problem.... :-))) > >>And I don't call that "positional" chess. I call it "suicide". > >So we have a different point of view. We try to win with committing >suicide against the fast-searchers. Who dies first ?!? >Do we get a pacifistic nobel-price for this method ?! at least you get a decent burial. :)
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.