Author: Dave Gomboc
Date: 20:21:51 05/23/99
Go up one level in this thread
On May 23, 1999 at 22:30:51, Melvin S. Schwartz wrote: > >On May 23, 1999 at 21:25:14, Dave Gomboc wrote: > >>Here's a section of http://www.chessbase.com/products/engines/Hiarcs/index.html: >> >> Playing Characteristics >> >> At first glance, the Hiarcs engine is strikingly slow in terms of nodes >> per second. Hiarcs calculates at roughly a tenth of Fritz' and Junior's >> node speed. However the 'look-and-feel' of Hiarcs in practical analysis is >> that of a very, very fast program. Its speed in solving tactical positions >> is at least in the league of the fastest searchers, if not better. Its by a >> >>This claim, I can live with. People here seem to think Hiarcs is pretty good in >>this department. >> >> far margin the strongest 1min-Blitz player in the world. But on the other >> >>But where does this audacious claim come from? Did somebody decide to test a >>bunch of programs on ICC at 1 0? This claim doesn't even seem to be completely >>testable: Ferret has won what, the last 3 blitz championships at the WMCCC? Two >>of the last three, at least. But it's tough for anyone but Bruce to play a 1 >>minute match between Hiarcs and Ferret, and as far as I know, Bruce hasn't >>reported any results for such a match. >> >>If it really IS "by a far margin the strongest 1min-Blitz player in the world", >>great!, congratulations... but saying it doesn't make it so: where's the >>evidence? >> >> hand you immediately feel its extraordinarily profound chess knowledge in >> the middle- and endgame. >> >> Hiarcs shows a highly attractive, strategical and goal-oriented playing >> style which makes it an enjoyable opponent for humans. Mark Uniackes work >> is more than a valuable enrichment for the world of chess programs since he >> achieves supreme playing strength by a different paradigm. >> >>This is no different from claims made by other chess software professionals. So >>what is the "different paradigm"? How is this testable? Hiarcs is not an >>open-source software project. People blab about how CSTAL does it differently, >>but it's not verifiable by third parties. No difference here. >> >>From the program description given to the ICCA for WCCC 1999: >> >> HIARCS searches around an order of magnitude less positions per second >> (av. 18,000) than most of its competitors. However, it makes up for this >> apparent slow speed by clever searching and accurate evaluation. HIARCS >> uses many selective search extension heuristics to guide the search and >> incorporates a sophisticated tapered search to resolve tactical >> uncertainties while finding positionally beneficial lines. >> >>This description is more useful, thank you. However, I will not hesitate to >>point out that one of the great things about the "sophisticated tapered search" >>sentence is that it is concrete enough to give the impression that some >>disclosure is being made, and abstract enough to allow every individual to have >>their own interpretation of precisely what it means. <grin> >> >>Dave > >Dear Dave, > >I have Hiarcss 7 as well as REBEL 10. I can tell you that yes Hiarcs can be a >bit slow - sometimes by minutes - but in complex middlegame positions it will >make the better move! I personally feel that time is relevant but hold the >proper analysis to be of even more importance. What good is it if a program >plays faster but misses key moves? Therefore, I firmly believe Mark Unkacke is >going in the right direction, for I would rather see a chess program play an >intelligent game and lose on time than a fast program making a blunder and >losing! This is of course just my opinion. > >Regards, >Mel I don't think your response addressed any of the issues I raised in my post. I suppose there's nothing wrong with this, but if your intention was to respond to my concerns, please re-read my original message. Dave
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.