On 27/03/16 09:28, Derek Profts wrote:
1q6/1Pr2k2/1R2p3/3pPp1p/3P2pP/4P1P1/1R3K2/8 b - - 14 59
I believe that this position leads to a draw with correct play.
Stockfish 7 64 bit has been crunching away at it for 48 hours or so
and assesses it as -1.58
I think you're right. But this is one of those positions where
the sides can shuffle for 49 moves, then, seeing the draw imminent,
Black can sacrifice the f-pawn and shuffle for another 49 moves before
giving up R for P and then another long sequence before the Q vs 2R
ending is established drawn. [I'm not claiming this to be best play,
just the play that delays the inevitable the longest.] So you could
use another 48h and still not reach (49+49+49)*2 == 294 ply!
I had a position OTB some years ago where, in a bad position,
I could have [but didn't, out of cowardice in time trouble] sacrificed
a rook to reach one of those positions where Black's extra R was
pinned and attacked enough that Black could make progress only by
sacrificing it back, into a drawn opposite-coloured B ending. It
didn't surprise me that the computer didn't find the sacrifice,
and mis-evaluated it analysing forwards. But when I'd moved it on
a bit, it realised that it wasn't making progress, found the draw,
and then found it again when wound back to the start, presumably
because the draw was cached. Possibly something similar might be
possible here? If Stockfish can be persuaded to see the draw after
the f-pawn moves, it might then see it from the start. [I don't
have Stockfish, so can't try this myself, and in any case don't
want my computer tied up for 96 hours!]
--
Andy Walker,
Nottingham.
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)