Author: Robert Hyatt
Date: 20:09:54 06/29/98
Go up one level in this thread
On June 29, 1998 at 22:21:49, Don Dailey wrote: >On June 29, 1998 at 19:53:57, Robert Hyatt wrote: > >>On June 29, 1998 at 18:33:36, Bruce Moreland wrote: >> >>> >>>On June 29, 1998 at 13:28:31, Robert Hyatt wrote: >>> >>>>On June 29, 1998 at 12:39:13, Bruce Moreland wrote: >>> >>>>>I am merely stating that it can't be said that null-move is bad just because DB >>>>>doesn't do it. >>> >>>>there I'd agree. They are so much faster/stronger than the rest of us, that >>>>whether they use null-move or not is probably inconsequential. I believe that >>>>null-move would make them even stronger, but Hsu has this "perfection" syndrome >>>>that refuses to accept any error at all unless it is unavoidable. With null- >>>>move, you instantly agree to accept errors, in return for more depth that might >>>>catch errors you didn't agree to accept. >>> >>>You have seen where this attitude gets you on a micro, I am sure, and my point >>>is that once you have learned to jettison this attitude, I don't think you'll >>>take it back if you get the horsepower again. >>> >>>>without a doubt correct, although I probably lean a little backward toward Cray >>>>Blitz now, since parallel processing has boosted my speed enough, I'm now trying >>>>to find time to investigate things that I found useful in CB, like singular >>>>extensions, for one example. I'd like to one day try null-move R=2 on CB, but >>>>it would be *very* difficult to do since all that stuff is in assembly language. >>>>But I'd like to know how that affects the thing since I *never* tried it, never >>>>even tried recursive null-move in fact.. >>> >>>I bet there would be a dramatic improvement. >>> >>>My feeling is based upon running Ferret with and without null-move during its >>>evolution. Null-move is always dramatically better overall on extended tactical >>>tests. >>> >>>>they had a lot of pressure from us. We were never slow, and almost beat them >>>>the first time we played them, but a cute SMP bug made us avoid a outright >>>>winning move that they were expecting, and we were going to play, until the >>>>last minute. >>>> >>>>In fact, they never "overwhelmed" us in speed, since they were doing 2-3M nodes >>>>per second in deep thought II, while we were only doing 1/4 of that or so at >>>>the time. As a result, their stick wasn't "that big", only a factor of 4 or >>>>so. But other things were certainly working for them in our games, like SE >>>>for one. >>> >>>How many times did you play them? >>> >>>bruce >> >>4. first time we could have won. last three we were just busted. They see >>too much. > >I really don't think you are correct to draw these conclusions from >only 4 games. You make it seem like you had no chance whatsoever, >and I have no doubt whatsover that this is how you felt as you watched >your own brainchild lose to them. > >You might very well be right but your last statement isn't exactly >convincing evidence, "first time we could have won, last threee we >were just busted. They see too much." > >Re-think this please. We are talking about Deep Thought, not Deep >Blue. I would prefer to see 30 or 40 games with a lopsided score >before I would agree with you. Cray Blitz was no slouch. > >- Don no it wasn't.. but remember, I had the opportunity to sit across the tabe from them (I have a eyboard bug that has me without the characters between j and m.) and got to see their output with ours. remember the game I posted ast year where they saw winning a piece 10 moves before CB saw it? That happened every game we ran into them. Happened to most everybody e.se as we.., based on resu.ts I watched them wax CB, HiTech, *socrates, *tech, .achex, and near.y everyone they p.ayed during that period of time... I'm convinced of their strength, because probabi.ity on.y ta.es you so far (hate this .eyboard)
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.