Author: Vincent Diepeveen
Date: 16:56:23 12/24/00
Go up one level in this thread
On December 23, 2000 at 18:45:02, Joshua Lee wrote: >Which programs solve positions 2,3,9,17,18, and 23? > >These are the only ones Fritz , Gandalf , and Hiarcs can't touch....or did i not >sit them on the positions long enough??? you gotta be kidding that no prog finds these: #2 d5 - no way diep ever finds that one within a time that i have patience #3 f5 - diep always had probs finding this one as the line for it to find it is representing quite a huge search depth. However to my big amazement commercials in past always found this position at lightspeed. That includes several of the above progs, you kidding that they nowadays don't find it? Idem for all other pos you name especially i remember position h7h5 which diep in past found at depth=1 like all other progs, nowadays this testset not so popular anymore which explains probably why they do bad. Note there are good reasons to do bad at these positions. Actually these positions should be forbidden to find at low depth, but should be all found at big depths. However if progs find all pos for the right reason, then they going to be world champ with induction to superior knowledge compared to mankind as the pos are pretty tough for mankind already as it's all about opening (half)closed positions where we talk about and that at the right time... #9 f4-f5 Always was a peanut for diep, but nowadays it gives away less happily pawns to just create a messy pawn structure with black. New lemma of mine: get a pawn, keep it till endgame, and win game. Going to work brilliant against Gambit Tiger :) [D]2kr1bnr/pbpq4/2n1pp2/3p3p/3P1P1B/2N2N1Q/PPP3PP/2KR1B1R w Note that basically the reason that i researched this pos recently gives why it no longer finds it. To see the pawn sac to be good you need to also see the manoeuvre Ne2-f4 which together with some development moves is needing quite a bit of depth. Note that i wonder how black plans to draw this anyway without f5 sacrafice. I'm at +1.6 at 10 ply already real soon for Bd3 and f4-f5 sac a few moves later. So i tuned it to not give away a pawn that soon, how about that, tuning an engine to not solve a trick in a testpos real soon! Probably not many dudes doing that walking around! Don't need to mention most progs in past didn't find this pos anyhow as giving away pawns at that time was not so common. No big deal if nowadays no engine finds this of course, idem for a lot of positions as the strategic pos all share that they're above the knowledge of any program anyway... #17 h5 now this always was a peanut for nearly any engine i know, note the move is strategical sound breaking open the position so i have no idea why most progs might no longer solve it. Perhaps they tuned king safety a bit higher so they fear h5 gxh5 nowadays, just in order to create more aggressive play and solving some nolot pos like #1 a bit sooner, who knows? Anyway i remember many engines including mine to solve this pos at plydepth=1 when this testset was a bit more popular... :) [D]r2q1rk1/1ppnbppp/p2p1nb1/3Pp3/2P1P1P1/2N2N1P/PPB1QP2/R1B2RK1 b But to conclude: in past some of the above progs solved these pos at lightspeed. Only Ed had worked a bit more obvious as the others and his child also solved #2 Greetings, Vincent
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.