Author: Jonathan Parle
Date: 21:23:50 01/27/02
Go up one level in this thread
On January 27, 2002 at 23:19:23, robert flesher wrote: >Hi everyone! I wondering what everyones thoughts are on Tactic's VS Positions >play. Seems to me that Computers with there every growing tactical skill seems >to be able to Destroy Good position play with awesome, blazing tactical >variations. This is NOT always the case but more and more it is become reality. >Here is a game me and a friend play Gambit Tiger pentium 900mhz Versus >Chessmaster8000 1ghz. This game is wildly tactical with no regard for pawn >structure, just and all out bloodbath. Very Very amusing game. > > >[Event "?"] >[Site "?"] >[Date "27/1/2002"] >[Round "?"] >[White "Gambit Tiger 2.0"] >[Black "Chessmaster 8000"] >[Result "1-0"] >[Opening "B01 Scandinavian Defense"] > >1.e4 d5 2.exd5 Qxd5 3.Nc3 Qa5 4.d4 Nf6 5.Nf3 c6 6.Bc4 Bf5 7.Bd2 e6 8.Nd5 Qd8 >9.Nxf6+ Qxf6 >10.Qe2 Nd7 11.O-O-O Bg4 12.d5 Bxf3 13.gxf3 cxd5 14.Bxd5 O-O-O 15.Be4 Bd6 16.Kb1 >Nb6 17.Bc3 >Qh6 18.a4 Bc7 19.a5 Nd5 20.Bd2 Qh5 21.a6 b5 22.Rhg1 Qe5 23.c4 bxc4 24.Qxc4 f5 >25.Bc2 Rd6 >26.Rge1 Qf6 27.Bb3 Rhd8 28.Rc1 R8d7 29.Ba4 Re7 30.Bc6 Nb6 31.Qc2 Qh4 32.Be3 Rf7 >33.Rg1 >h6 34.Be8 Re7 35.Bb5 Kb8 36.Rgd1 Qxh2 37.Bxb6 axb6 38.a7+ Kxa7 39.Qa4+ Kb8 >40.Qa6 Qf4 41.Bc6 >Qxc1+ 42.Rxc1 Rxc6 43.Rxc6 1-0 I certainly believe there is a computational level at which pure tactical play will defeat positional play nearly all the time, if not all the time. I think we will see this phenonomen happen as an incidental occurance, due to ongoing hardware improvements over the coming years. When this happens, I won't commit to saying, partly because I don't really feel that any programmers feel the need to experiment seriously in this direction anymore. However, tactical improvements are going to come more and more slowly as time goes by, because processor speed improvements would have to be highly exponential to allow tactics to improve at the same rate as they have been thus far. So far, we seem to be doubling processor speeds every year (keeping Moore happy), but this won't be enough to keep the tactical improvements coming at the same rate. You only have to look at the ELO improvement of programs such as Deep Fritz to see only around 60 points improvement by going from a mere K6-2 450Mhz processor to a modern 1200 Mhz Althon. (And arguably, a signficant component of the 60 points could be attributable to the larger memeory available for hash tables). 18 years ago, you would have got 100 points ELO by throwing out your 2 Mhz 6502 machine, and purcahsing a 4Mhz 6502 machine!!! So doubling processor speeds a couple of decades ago brought much greater improvements than even a tripling does today. I predict that from ten years time, a ten fold increase in processing power will only get you around 10 - 20 points ELO improvement, all other things being equal. I still think the key to defeating super Grand Masters on a regular basis requires more hueristics, more knowledge and massive amounts of extremely high speed and stable memory. I don't really think adding 200 or even 300 points to a program's tactical ability will really improve it's playing strength against a 2700+ human, although it would certainly be much more noticeable in computer vs computer competition. And in closing, I would point out to programmers that , imo, the most boring chess programs on the planet are the ones that emphasise tactics. I'd much rather play against a program that searches very few nodes (and does a lot of processing with each one), than one that goes hell for leather through as many iterations as the hardware will conceivably allow, with little regard to anything other than "he goes there...I go here....he goes there...I go here....he goes there....I go here" ad infinitum.
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.