Author: John Merlino
Date: 15:13:59 04/12/05
Go up one level in this thread
On April 12, 2005 at 15:20:35, chandler yergin wrote: >On April 12, 2005 at 11:54:35, John Merlino wrote: > >>On April 12, 2005 at 05:15:43, chandler yergin wrote: >> >>>On April 12, 2005 at 00:23:30, John Merlino wrote: >>> >>>>On April 11, 2005 at 23:36:27, chandler yergin wrote: >>>> >>>>>On April 11, 2005 at 13:00:10, Rob Basham wrote: >>>>> >>>>>>On April 11, 2005 at 12:24:18, John Merlino wrote: >>>> >>>>>>>>New game, >>>>>>>>[D]2r1r2k/4R1pp/pb6/1p6/2nqN1QB/P5PP/1PB4K/5R2 w - - 0 1 >>>>>>>> >>>>>>>>Analysis by Shredder 8: >> >><Some Analysis Snipped> >> >>>>>>>>1.Bf6 Qxf6 2.Nxf6 Bg1+ 3.Kxg1 Rg8 4.Nxg8 Nxb2 5.Qxg7# >>>>>>>> +- (#152) Depth: 12/42 00:00:22 7641kN >>>>>>>>1.Bf6 Rxe7 2.Bxd4 Rd8 3.Qh4 Ree8 4.Nc3 Re2+ 5.Nxe2 h6 6.Kh1 Rxd4 7.Rf8# >>>>>>>> +- (#152) Depth: 13/18 00:00:23 7840kN >>>>>>>>1.Bf6 Rxe7 2.Bxd4 Rd8 3.Qh4 Bxd4 4.Qxe7 Rg8 5.Rf8 Bg1+ 6.Kxg1 Nd6 7.Qxd6 b4 >>>>>>>>8.Rxg8+ Kxg8 9.Qd8+ Kf7 >>>>>>>> +- (#150) Depth: 14/40 00:00:43 15586kN >>>>>>>>1.Bf6 Rxe7 2.Bxd4 Rd8 3.Qh4 Bxd4 4.Qxe7 Rg8 5.Rf8 Bc5 6.Nxc5 Nd6 7.Qxd6 b4 >>>>>>>>8.Rxg8+ Kxg8 9.Qe6+ Kf8 10.Nd7# >>>>>>>> +- (#150) Depth: 15/36 00:00:43 15751kN >>>>>>>>1.Bf6 Rxe7 2.Bxd4 Rd8 3.Qh4 h5 4.Qxe7 Ne5 5.Qxe5 Bxd4 6.Qxh5+ Kg8 7.Bb3+ Rd5 >>>>>>>>8.Bxd5# >>>>>>>> +- (#150) Depth: 16/38 00:00:47 17089kN >>>>>>>>1.Bf6 Rxe7 2.Bxd4 Rce8 3.Qf5 Kg8 4.Ng5 g6 5.Qd5+ Re6 6.Rf8+ Kxf8 7.Qf3+ Ke7 >>>>>>>>8.Qb7+ Bc7 9.Qxc7+ Kf8 10.Qg7# >>>>>>>> +- (#150) Depth: 17/36 00:00:49 17858kN >>>>>>>> >>>>>>>>(, 11.04.2005) >>>>>>> >>>>>>>Shredder's output is incredibly bizarre here. The position is a Mate in 9, but >>>>>>>not a single PV shows mate in 9 moves (although there are mates shown in 10, 7 >>>>>>>and even 5 moves!??). >>>>>>> >>>>>>>Conclusion: Shredder found the best move, but did NOT conclusively find the mate >>>>>>>in 9. >>>>> >>>>>THe Mate in 9 is not forced; Shredder gave all the variations leading to >>>>>Mate after the Key Move. >>>>> >>>>>With "Best PLay" by Black it's Mate in 9. >>>>> >>>>>If Black errs.. he gets Mated sooner. >>>> >>>>I understand that, but Shredder's output still makes no sense at all. What do >>>>you mean by "Shredder gave all the variations leading to Mate after the Key >>>>Move"? >>> A Program evaluates every possible move at each ply depth; >>>that is one complete iteration. >>>Shredder found a Mate at ply 12, another at ply 13, another at ply 14, >>>another at ply 15, another at play 16 etc. >>> >>>What don't you understand? >> >>Simple -- take this PV from Shredder (which is the first one listed above): >> >>1.Bf6 Qxf6 2.Nxf6 Bg1+ 3.Kxg1 Rg8 4.Nxg8 Nxb2 5.Qxg7# >> +- (#152) Depth: 12/42 00:00:22 7641kN >> >>A program should not announce a Mate for the side to move that is simply not >>possible with best play. If I were to stop analysis at this point and post my >>findings here, it would be soundly refuted by the simple fact that there is no >>Mate in 5 for White. >> >>It seems bizarre to me that, at depth 12, the Shredder thinks that there is a >>forced mate in 5 (only 9 plies away). >> >>jm > The Program did what a Program is supposed to do! > >I "clipped" the analysis. >Do you have any doubt, that if I had let it run,. that it would stabalize >on the Mate in 9 with a deeper search? No, I have no doubt about that at all. But there are two problems with that: 1) Anybody could say that about any program with regards to a forced mate of reasonably short distance, and 2) It's not the ability to find the mate that I have called into question. It is the PV output that makes no sense. (Notice that I haven't even mentioned the "#150" that Shredder regularly puts out for no reason....) jm
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.