Author: Robert Hyatt
Date: 20:12:43 04/24/00
Go up one level in this thread
On April 24, 2000 at 22:02:00, Peter Kappler wrote: >>> >>>I never denied it would hurt Crafty. I'm just waiting for evidence that it >>>hurts Crafty more than others. >>> >> >> >>Why do you need evidence? > >Huh? Because it's not obvious to me that it's true. > > >>>>(2) I am very aggressive with null-move. As you reduce the time per move, >>>>there is a noticable point where Crafty will start getting killed by a program >>>>that doesn't use it as 'carelessly' as I do. This means that (1) above will >>>>cause (2) to happen since time trouble -> reduced search depth. >>>> >>> >>>Then this problem is going to exist in any fast time control game, *regardless* >>>of the ponder state... >> >> >>Have you ever heard me say "Crafty isn't a very good blitz/bullet chess player. >>Because it is an aggresssive null-move program that doesn't restrict the use >>of null-move at all"??? >> >>I have said it often enough... And I gave some games as samples a couple of >>years ago.. At short search depths, crafty has trouble. Just ask any comp >>operator on ICC what time control they prefer. 3 0 blitz is the favorite. They >>do _much_ better at 3 0 than at 5 3. _much_ better. >> >> > >Bob, I never claimed otherwise. I just wanted to point out that agressive >nullmove pruning will cause Crafty problems at fast time controls, *regardless* >of the ponder state. And since this is true, why all the hooplah over >pondering, when in fact the time-control has much more influence on Crafty's >strength relative to other programs? > >--Peter Once you understand the above. you understand why ponder=off is a big problem for crafty. Both from a time utilization point of view, as well as reducing the search time, giving more opportunity for null-move failures... > > >>> >>> >>>>I simply say that to play a match, you play the two opponents at their strongest >>>>(and best tested) settings. Not at some crippled level where we spend the next >>>>year arguing which is affected the most. >>> >>> >>>Great, now I just need access to two *identical* dedicated machines so I can >>>play an "optimal" match. So, should I go spend $2000 for that new machine, or >>>maybe it makes more sense for me to try to get an answer to the question in my >>>previous post? See *my* point? >>> >>>--Peter
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.