Author: fca
Date: 12:11:48 08/06/98
Go up one level in this thread
On August 06, 1998 at 13:52:56, Amir Ban wrote: >On August 06, 1998 at 12:40:05, fca wrote: > >> >>Making two guesses (first being the "when") who will then be that #1? >> >>R10? J6? F6? N99? H7? MCP11? Ferr1?? DiepX??? CCTValX? :-) >> > >You are way ahead of me. I'm not done with J5 yet. And Richard's 32 bit engine isn't done either... I didn't say how far ahead we need to look... >>>Several of us are now beating CG3. >> >>Not by *that* much, though, bearing in mind it was out 4yrs ago (assuming no >>books)? > >That, by the way, was my point. I knew. I was only underlining it. >>>The real trick, however, was to do it in 1994. >>Tardis needed now to do that :-) >>Thorsten is quite right IMO highlighting null-move algorithms as being a >>significant development.... >>So, are the days of clever evaluation function being of paramount (#1) software >>(yes, Thorsten, I know hardware jumps have had a much bigger effect, that's why >>I put 'software' there so you do not jump at me :-)) ) importance now over? >>Views anyone... >> > >I think you have this backward. Maybe not... pl see below :-) >Genius had very good evaluation for its time, but its strong point was >definitively the search. Absolutely agreed! Thorsten had stated "Shit. Shit nullmove ! The nullmove killed Richards search." I agree, as already stated in my previous post. So during the _commanding_ reign of Genius, eval fn was less important. Now with F5 apparently in _commanding_ #1 position (yes, controversy is good :-) ), eval fn also found to be less important. No "commanding" lead by any prog between the times of Genius on top and F5 on top (rather, first among equals). So... is the age of eval fn importance over? Is it going to remain that way "forever"? That was the question... Forwards and backwards. I asked for views; it is not necessarily my opinion... >I didn't notice that the null-movers are having a party at the expense of >Genius. F5? SSDF give it 29.5 : 18.5 over G5D on 200MMXs, cf neck and neck with H6. Could be a party, then? Please do not all shout booked, RAM size, autoplayer etc. at once - IMO these only account for a little of the difference. >It's more the programs who are not null-movers who are doing this >on the basis of better evaluation. In my heart too, I believe this (better eval) must be the right way. But *F5 is on top*. It keeps winning games "by chance". Too many games for it to be chance. So, my heart can be wrong maybe? >Amir Kind regards fca
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.