Author: Duncan Roberts
Date: 07:35:15 01/13/05
Go up one level in this thread
First of all thanks for doing this and I believe this has contributed to chess knowledge. To sum up can we say on the assumption that null move finds most stuff, that most likely there is no material gain for white in the first 35 (12 + 23)moves. ? I was rather surprised that as before you had got up to 34 ply, the 6 extra move start only gained you an extra ply or 2. Is this something that you were expecting ? on ply 23 only 8 moves are shown, not the full 23 ply. Why is this ? if you upgrade in the next 2 -3 years, would you be prepared to re-do the experiment to look at 36 ply. duncan On January 11, 2005 at 17:25:34, Andrew Williams wrote: >On January 10, 2005 at 17:03:49, Andrew Williams wrote: > >>On January 10, 2005 at 16:45:43, Ricardo Gibert wrote: >> >>>I don't understand the interest in material only search. You've completed 34 >>>ply, but what does that prove? For instance, how do you know if upon completion >>>of ply 35, your engine might decide that the best line gives up a pawn at move >>>3? It's possible. You're just spinning your wheels here for no reason. And this >>>is independent of whether or not you are using null move pruning or whatever. >>>That issue is a red herring that distracts from the fact that no firm conclusion >>>can drawn in any case. >> >>You're telling the wrong person here! I don't disagree at all. This started in >>2001 when Rudolf Huber searched to ply 30 in 2.5 hours and asked if anyone could >>beat that. My program could (on faster hardware). I was interested in his >>"challenge" at the time because both Rudolf's and my programs are based on a >>technique called MTD(f), which is not commonly used. Somebody raised this again >>and I said I'd do the search. I don't think there's anything we can conclude >>except that postmodernist's search with a material-only eval and null move >>pruning doesn't find any material gain for either side after a 34-ply search. >>Unfortunately, this doesn't say very much about the game of chess, but it was no >>problem to do the "experiment". >> >>I think this might be more interesting: >> >>http://www.talkchess.com/forums/1/message.html?404572 >> >>What do you think? >> >>Andrew >> > >I did Uri's experiment see link below: > >http://www.talkchess.com/forums/1/message.html?404572 > >I got this with on overnight, material only search (I'm sorry, I don't have time >to clean it up): > >COMPLETELY CLEARING HASH TABLES >TTSIZE: 16777216 >PAWNHTSIZE: 524288 >EVALHTSIZE: 262144 > > > >AFTER EDIT > >8 r n b q k b n r >7 o o o o o o o o >6 . . . . . . . . >5 . . . . . . . . >4 . . B P P B . . >3 . . N . . N . . >2 P P P . . P P P >1 R . . Q K . . R > > a b c d e f g h > > > >analyze > >Preparing clock: time left=36000000 allocation=1395000 max >allowed=8370000 failing low max=2790000 halfMoves: 0 remaining moves=25 >clock DIVISOR WAS: 25 >clock startTime WAS: -1129662192 > >MOVES SINCE BOOK: 0 > 1> +0.00 00:00:00.00 2 1.a3 > 1< +0.01 00:00:00.00 56 1.a3 > 1= +0.00 00:00:00.00 56 1.a3 > 2> +0.00 00:00:00.00 58 1.a3 > 2< +0.01 00:00:00.00 112 1.a3 > 2= +0.00 00:00:00.00 112 1.a3 > 3> +0.00 00:00:00.00 135 1.a3 > 3< +0.01 00:00:00.00 288 1.a3 > 3= +0.00 00:00:00.00 288 1.a3 > 4> +0.00 00:00:00.00 334 1.a3 a6 > 4< +0.01 00:00:00.00 894 1.a3 > 4= +0.00 00:00:00.00 894 1.a3 a6 > 5> +0.00 00:00:00.00 1057 1.a3 a6 2.a4 > 5< +0.01 00:00:00.01 2006 1.a3 > 5= +0.00 00:00:00.01 2006 1.a3 a6 2.a4 > 6> +0.00 00:00:00.01 2322 1.a3 a6 2.a4 f6 > 6< +0.01 00:00:00.03 7267 1.a3 > 6= +0.00 00:00:00.03 7267 1.a3 a6 2.a4 f6 > 7> +0.00 00:00:00.04 8266 1.a3 a6 2.a4 f6 3.b4 > 7< +0.01 00:00:00.08 18499 1.a3 > 7= +0.00 00:00:00.08 18499 1.a3 a6 2.a4 f6 3.b4 > 8> +0.00 00:00:00.20 22090 1.a3 a6 2.a4 f6 3.b4 Nc6 4.b5 > 8< +0.01 00:00:00.30 46036 1.a3 > 8= +0.00 00:00:00.30 46036 1.a3 a6 2.a4 f6 3.b4 Nc6 4.b5 > 9> +0.00 00:00:00.33 51238 1.a3 a6 2.a4 f6 3.b4 Nc6 4.b5 > 9< +0.01 00:00:00.49 128517 1.a3 > 9= +0.00 00:00:00.49 128517 1.a3 a6 2.a4 f6 3.b4 Nc6 4.b5 >10> +0.00 00:00:00.53 147555 1.a3 a6 2.a4 f6 3.b4 Nc6 4.b5 axb5 5.axb5 >10< +0.01 00:00:00.78 290343 1.a3 >10= +0.00 00:00:00.78 290343 1.a3 a6 2.a4 f6 3.b4 Nc6 4.b5 axb5 5.axb5 >11> +0.00 00:00:00.82 312345 1.a3 a6 2.a4 f6 3.b4 Nc6 4.b5 axb5 5.axb5 >11< +0.01 00:00:01.74 844194 1.a3 >11= +0.00 00:00:01.74 844194 1.a3 a6 2.a4 f6 3.b4 Nc6 4.b5 axb5 5.axb5 >12> +0.00 00:00:01.91 938867 1.a3 a6 2.a4 f6 3.b4 Nc6 4.b5 axb5 5.axb5 >Rxa1 6.Qxa1 >12< +0.01 00:00:03.20 1684698 1.a3 >12= +0.00 00:00:03.20 1684698 1.a3 a6 2.a4 f6 3.b4 Nc6 4.b5 axb5 5.axb5 >Rxa1 6.Qxa1 >13> +0.00 00:00:03.53 1807771 1.a3 a6 2.a4 f6 3.b4 Nc6 4.b5 axb5 5.axb5 >Rxa1 6.Qxa1 >13b +0.01 00:00:10.36 5702674 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 >13< +0.82 00:00:10.89 6033101 1.Bxf7 >13= +0.80 00:00:10.89 6033198 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nc6 bxc6 >5.Nxc7 Kf7 6.Nxa8 d6 >14< +0.80 00:00:21.35 11937802 1.Bxf7 >14> +0.47 00:00:22.04 12423716 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nc6 Nxd5 >5.Nxd8 Nxf4 6.Qf3 Nxg2 7.Qxg2 Kxd8 >14= +0.50 00:00:22.04 12423813 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nc6 Nxd5 >5.Nxd8 Nxf4 6.Qf3 Nxg2 7.Qxg2 Kxd8 >15> +0.50 00:00:22.13 12466245 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nc6 Nxd5 >5.Nxd8 Nxf4 6.Qf3 Nxg2 7.Qxg2 Kxd8 >15< +0.51 00:01:00.42 34921843 1.Bxf7 >15= +0.50 00:01:00.42 34921843 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nc6 Nxd5 >5.Nxd8 Nxf4 6.Qf3 Nxg2 7.Qxg2 Kxd8 >16> +0.50 00:01:00.72 35115638 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nc6 Nxd5 >5.Nxd8 Nxf4 6.Qf3 Nxg2 7.Qxg2 Kxd8 >16< +0.84 00:02:01.48 69840931 1.Bxf7 >16= +0.80 00:02:01.48 69841027 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 exf6 >5.Qh5 g6 6.Nxg6 Bb4 7.c3 hxg6 8.Qxh8 Bf8 >17> +0.80 00:02:01.73 69971825 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 exf6 >5.Qh5 g6 6.Nxg6 Bb4 7.c3 hxg6 8.Qxh8 Bf8 >17< +0.81 00:04:21.12 153913475 1.Bxf7 >17= +0.80 00:04:21.12 153913475 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 exf6 >5.Qh5 g6 6.Nxg6 Bb4 7.c3 hxg6 8.Qxh8 Bf8 >18> +0.80 00:04:24.85 156009937 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 exf6 >5.Qh5 g6 6.Nxg6 Bb4 7.c3 hxg6 8.Qxh8 Bf8 >18< +0.81 00:11:31.86 391722548 1.Bxf7 >18= +0.80 00:11:31.86 391722548 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 exf6 >5.Qh5 g6 6.Nxg6 Bb4 7.c3 hxg6 8.Qxh8 Bf8 >19> +0.80 00:11:33.08 392367418 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 exf6 >5.Qh5 g6 6.Nxg6 Bb4 7.c3 hxg6 8.Qxh8 Bf8 9.b4 >19< +0.81 00:30:54.601015431731 1.Bxf7 >19= +0.80 00:30:54.601015431731 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 exf6 >5.Qh5 g6 6.Nxg6 Bb4 7.c3 hxg6 8.Qxh8 Bf8 9.b4 >20> +0.80 00:30:59.571017941960 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 exf6 >5.Qh5 g6 6.Nxg6 Bb4 7.c3 hxg6 8.Qxh8 >20< +0.81 01:24:45.21-1514697185 1.Bxf7 >20= +0.80 01:24:45.21-1514697185 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 >exf6 5.Qh5 g6 6.Nxg6 Bb4 7.c3 hxg6 8.Qxh8 >21> +0.80 01:25:00.17-1506404084 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 >exf6 5.Qh5 g6 6.Nxg6 Bb4 7.c3 hxg6 8.Qxh8 Bf8 9.b4 >21< +0.81 03:20:59.86-1970190323 1.Bxf7 >21= +0.80 03:20:59.86-1970190323 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 >exf6 5.Qh5 g6 6.Nxg6 Bb4 7.c3 hxg6 8.Qxh8 Bf8 9.b4 >22> +0.80 03:21:31.17-1953479875 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 >exf6 5.Qh5 g6 6.Nxg6 hxg6 7.Qxh8 d6 8.c3 >22< +0.81 07:54:41.90-1472530151 1.Bxf7 >22= +0.80 07:54:41.94-1472530151 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 >exf6 5.Qh5 g6 6.Nxg6 hxg6 7.Qxh8 d6 8.c3 >23> +0.80 07:56:19.23-1420881993 1.Bxf7 Kxf7 2.Ne5 Ke8 3.Nd5 Nf6 4.Nxf6 >exf6 5.Qh5 g6 6.Nxg6 hxg6 7.Qxh8 d6 8.c3 Nc6 > > >As you can see, the final position and evaluation is nonsensical, but that's >what the search revealed. Incidentally, I switched it off while it was still >deciding what the value for ply 23 was. The fact that it says "23>" doesn't mean >that the score was necessarily going to go any higher; "23>" should really be >"23>=", but that doesn't look as neat. > >Cheers > >Andrew
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.