OpenChess

I have no idea where I got this position but found it in some old files I was getting rid of. It's amazing that of all the engines I tested all except Critter think that White's position is hopeless while obviously it is a dead draw.

6kr/5b1p/2p3pP/rpPp1pP1/pP1PpP2/P3P3/1K6/8 w - -

biscuit1953 wrote:I have no idea where I got this position but found it in some old files I was getting rid of. It's amazing that of all the engines I tested all except Critter think that White's position is hopeless while obviously it is a dead draw.

6kr/5b1p/2p3pP/rpPp1pP1/pP1PpP2/P3P3/1K6/8 w - -

There are versions of Crafty that find this instantly. By evaluation. The problem is, how often have you seen such a position? How much will recognizing that improve a program? And most importantly, how much time does that waste in the evaluation and what is the cost in the positions where such locked pawns don't occur? Answer to the second question. It won't improve the Elo at all. The answer to the third question is "it costs quite a few Elo because it slows the NPS down and reduces the depth slightly." If you look at crafty 19.10 and a few versions after that, there is specific code (not written by myself, contributed by someone else. If you compile with -DDETECT_DRAWS it should find that instantly. Unfortunately the cost was too high and it was removed in later versions...

biscuit1953 wrote:I have no idea where I got this position but found it in some old files I was getting rid of. It's amazing that of all the engines I tested all except Critter think that White's position is hopeless while obviously it is a dead draw.

6kr/5b1p/2p3pP/rpPp1pP1/pP1PpP2/P3P3/1K6/8 w - -

The evaluation itself is irrelevant, as long as the engine avoids taking the rook.

The last version of Crafty that had Jeremiah Pennery's DETECTDRAW code in it was 20.0.
It is, without a doubt, my favorite engine for analyzing closed positions.
It may be a problem in game play, but it is essential for analysis of certain positions. Of course, for these positions, we humans can instantly recognize them with our eyeballs (bunch of locked pawns that protect each other).

These positions do come up in real games, especially anti-computer games (I remember one by Christiansen that was really beautiful).

This famous position is clobbered by Crafty 20.0: I keep a crafty 20 program handy for analysis of closed positions, as it is the best program on earth for that, without question.

Here is the analysis for the original locked position:
epdpfga q:\locked.epd locked.out
PFGA: EPD record: 1
17 1 13 1007907 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
17 1 13 1015652 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
18 1 44 1511143 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
18 1 45 1519925 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
19 1 86 2097112 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
19 1 86 2119022 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
20 1 127 3019503 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
20 1 127 3043175 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
21 1 389 4411066 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
21 1 392 4670069 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
22 1 471 7198720 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
22 1 580 7652501 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
23 1 697 14353467 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6
23 1 864 15843569 1. Kc3 Be6 2. Kd2 Bd7 3. Kc3 Be6

zullil wrote:

biscuit1953 wrote:I have no idea where I got this position but found it in some old files I was getting rid of. It's amazing that of all the engines I tested all except Critter think that White's position is hopeless while obviously it is a dead draw.

6kr/5b1p/2p3pP/rpPp1pP1/pP1PpP2/P3P3/1K6/8 w - -

The evaluation itself is irrelevant, as long as the engine avoids taking the rook.

If your evaluation does not understand this, you will ALWAYS take a free rook when offered. Basic minimax guarantees that. So you have to have something in the evaluation that says that +5 gain should be ignored...

User923005 wrote:The last version of Crafty that had Jeremiah Pennery's DETECTDRAW code in it was 20.0.
It is, without a doubt, my favorite engine for analyzing closed positions.
It may be a problem in game play, but it is essential for analysis of certain positions. Of course, for these positions, we humans can instantly recognize them with our eyeballs (bunch of locked pawns that protect each other).

These positions do come up in real games, especially anti-computer games (I remember one by Christiansen that was really beautiful).

This famous position is clobbered by Crafty 20.0:

Code: Select all

2b5/1r6/2kBp1p1/p2pP1P1/2pP4/1pP3K1/1R3P2/8 b - - bm Rb4; id "WAC.230";
c:\project\dcorbit\crafty-20.0\crafty\x64\Release>crafty

Initializing multiple threads.
System is SMP, not NUMA.
EPD Kit revision date: 1996.04.21
unable to open book file [./book.bin].
book is disabled
unable to open book file [./books.bin].

Crafty v20.0 (1 cpus)

White(1): xboard

tellicsnoalias set 1 Crafty v20.0 (1 cpus)
tellicsnoalias kibitz Hello from Crafty v20.0! (1 cpus)
new
tellicsnoalias set 1 Crafty v20.0 (1 cpus)
post
st 99
mt 4
epdpfga q:\wac230.epd wac230.out
PFGA: EPD record: 1   ID: WAC.230
        12     -1      56 1472717  1. ... Rh7 2. Rb1 Kb5 3. Ba3 Bd7 4. Bb2 Rh5 5. Kg4 Kc6 6. Ba1 Be8 <HT>
        12   -170     282 1271218  1. ... Rb4 2. cxb4 a4 3. Rb1 c3 4. Rc1 b2 5. Rxc3+ Kd7 6. Rc7+ Kd8 7. Ra7 b1=Q 8. Rxa4
        12   -170     282 4383656  1. ... Rb4 2. cxb4 a4 3. Rb1 c3 4. Rc1 b2 5. Rxc3+ Kd7 6. Rc7+ Kd8 7. Ra7 b1=Q 8. Rxa4
        13   -188     354 6532710  1. ... Rb4 2. cxb4 a4 3. Rd2 Kb5 4. f4 c3 5. Rd3 Kc4 6. Rd1 c2 7. Rc1 a3
        13   -188     362 7716372  1. ... Rb4 2. cxb4 a4 3. Rd2 Kb5 4. f4 c3 5. Rd3 Kc4 6. Rd1 c2 7. Rc1 a3
        14   -228     432 11480861  1. ... Rb4!!
        14   -247     703 22819240  1. ... Rb4 2. Kf3 Ra4 3. Rb1 Ra2 4. Rc1 a4 5. Ke3 Bd7 6. f3 Rh2 7. Ra1 Kb6 8. Ra3
        14   -247     742 26079115  1. ... Rb4 2. Kf3 Ra4 3. Rb1 Ra2 4. Rc1 a4 5. Ke3 Bd7 6. f3 Rh2 7. Ra1 Kb6 8. Ra3
        15   -232    1154 50832810  1. ... Rb4 2. cxb4 a4 3. b5+ Kxb5 4. Rc2 bxc2 5. Ba3 c3 6. Kf4 Kc4 7. Ke3 c1=Q+ 8. Bxc1 Kb3 9. Kf4 a3
        15   -232    1394 58945262  1. ... Rb4 2. cxb4 a4 3. b5+ Kxb5 4. Rc2 bxc2 5. Ba3 c3 6. Kf4 Kc4 7. Ke3 c1=Q+ 8. Bxc1 Kb3 9. Kf4 a3
        16   -227    1720 74870558  1. ... Rb4 2. cxb4 a4 3. b5+ Kxb5 4. Rc2 bxc2 5. Ba3 c3 6. Kf4 Kc4 7. Ke3 c1=Q+ 8. Bxc1 Kb3 9. Kf4 a3 10. Be3
        16   -227    2053 107473782  1. ... Rb4 2. cxb4 a4 3. b5+ Kxb5 4. Rc2 bxc2 5. Ba3 c3 6. Kf4 Kc4 7. Ke3 c1=Q+ 8. Bxc1 Kb3 9. Kf4 a3 10. Be3

I keep a crafty 20 program handy for analysis of closed positions, as it is the best program on earth for that, without question.

It was good code. And today might not be so harmful. I will resurrect this in a version of Crafty (where it can be compiled out if it does hurt too much)...

hyatt wrote:

zullil wrote:

biscuit1953 wrote:I have no idea where I got this position but found it in some old files I was getting rid of. It's amazing that of all the engines I tested all except Critter think that White's position is hopeless while obviously it is a dead draw.

6kr/5b1p/2p3pP/rpPp1pP1/pP1PpP2/P3P3/1K6/8 w - -

The evaluation itself is irrelevant, as long as the engine avoids taking the rook.

If your evaluation does not understand this, you will ALWAYS take a free rook when offered. Basic minimax guarantees that. So you have to have something in the evaluation that says that +5 gain should be ignored...

The position that arises if White takes the rook evaluates as -INFINITY for White. Crafty-23.4 prefers Ka1, which evaluates as -15 or so. Even if Ka1 evaluated as -400, Crafty would still prefer it to bxa5. That's what I meant. No special draw detection code is needed in this position, as long as you have a decent search.

Yes, but how will engines do on this one:
k2b1q2/8/2p1p3/1pPpPp2/pP1B1Pp1/P5P1/8/4K2R w - -
c:\project\dcorbit\crafty-20.0\crafty\x64\Release>crafty

Other engines try their best for k2b1q2/8/2p1p3/1pPpPp2/pP1B1Pp1/P5P1/8/4K2R w - -
Has anyone seen my clue-by-four?
Analysis of Q:\semi-locked.epd
5/24/2012 12:08:31 PM Level: 60 Seconds
Analyzing engine: Stockfish Glinscott Pinned {3T}

1
Avoid move:
Best move (Stockfish Glinscott Pinned {3T}): Ke1-f2
Not found in: 01:00
2/3 00:00 205 0 -6.22 Rh7 Kb8
3/3 00:00 391 0 -6.06 Rh7 Kb8 Kd2
4/4 00:00 591 0 -6.06 Rh7 Kb8 Kd2 Qg8
5/6 00:00 1,123 0 -6.14 Rh7 Qg8 Rd7 Qe8 Rh7
6/6 00:00 1,703 1,703,000 -6.30 Rh7 Qg8 Rd7 Qe8 Rh7 Kb8
7/10 00:00 2,969 2,969,000 -6.30 Rh7 Qg8 Rd7 Kb8 Ke2 Kc8 Ra7 Bc7 Kd3
8/13 00:00 9,318 2,329,500 -6.66 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3
9/13 00:00 20,707 796,423 -6.86 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Qg7
10/13 00:00 29,579 518,929 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Be7 Rh1
11/13 00:00 35,962 599,366 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Be7 Rh2
12/14 00:00 43,295 698,306 -6.78 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Be7 Rh2 Qg7 Rh1
13/15 00:00 69,288 989,828 -6.82 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qf7
14/15 00:00 76,662 851,800 -6.82 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qf7
15/17 00:00 112,654 1,052,841 -6.94 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qf7 Rh1 Qg8 Rh6 Qg7
16/17 00:00 123,369 1,045,500 -6.94 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Bd8 Rh1 Be7 Rh2
17/17 00:00 137,371 1,048,633 -6.94 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Bd8 Rh1 Be7 Rh2
18/19 00:00 157,559 1,094,159 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg6
19/19 00:00 173,799 1,158,660 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Qf7 Rh1 Qg8
20/23 00:00 197,700 1,235,625 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Bd8 Rh1 Bc7
21/23 00:00 226,943 1,274,960 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Bd8 Rh1 Qg6 Rh2 Bc7 Rh8 Qg7 Rh1
22/23 00:00 334,177 1,591,319 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Bd8 Rh1 Qg6 Rh2 Bc7 Rh8 Qg7 Rh1
23/24 00:00 401,557 1,753,524 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Bd8 Rh1 Qg6 Rh2 Bc7 Rh8 Qg7 Rh1
24/25 00:00 656,178 2,262,682 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Bd8 Rh1 Qg6 Rh2 Kc7 Rh8 Qg7 Rh1 Kb7 Rh2 Kb8 Rh1 Kc8 Ke3 Kd7 Kd3 Bc7 Kd2 Bd8 Rh2 Ke7 Kd3 Bc7 Rh1 Kd7
25/25 00:00 810,216 2,447,782 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Bd8 Rh1 Qg6 Rh2 Bc7 Rh8 Qf7 Rh6 Qg8
26/26 00:00 1,432,829 2,942,154 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Bd8 Rh1 Bc7
27/27 00:00 1,863,366 3,240,636 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Bd8 Rh1 Qf7 Rh6 Be7 Rh2
28/29 00:00 2,110,999 3,257,714 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Ke7 Rh1 Qg6 Rh2 Qg7 Rh1 Kd7 Rh2 Ke7
29/29 00:00 3,452,411 3,836,012 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Ke7 Rh1 Qg6 Rh2 Qg7 Rh1 Kd7 Rh2 Qg6 Rh8 Qg7 Rh1
30/30 00:01 4,500,900 4,099,180 -6.90 Rh7 Qg8 Rh6 Kb8 Ke2 Kc8 Kd2 Kd7 Kd3 Bc7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Ke7 Rh1 Qg6 Rh2 Qg7 Rh1 Qg8 Rh6 Qg7 Rh1
31/51 00:06 34,078,659 5,272,069 -6.90 Rh7 Qg8 Rh5 Kb7 Ke2 Bc7 Rh2 Qg7 Kd2 Kc8 Rh1 Kd7 Kd3 Qg8 Rh2 Qf7 Rh6 Ke7 Rh2 Qg7 Rh1 Qg8 Rh2 Qf7 Rh1 Qg6 Rh2 Qg7
32/51 00:06 35,472,370 5,277,841 -6.90 Rh7 Qg8 Rh5 Kb7 Ke2 Bc7 Rh2 Qg7 Kd2 Kc8 Rh1 Kd7 Kd3 Qg8 Rh2 Qf7 Rh6 Ke7 Rh2 Qg7 Rh1 Qg8 Rh2 Qf7 Rh1 Qg6 Rh2 Qg7
33/51 00:06 36,151,565 5,285,316 -6.90 Rh7 Qg8 Rh5 Kb7 Ke2 Bc7 Rh2 Qg7 Kd2 Kc8 Rh1 Kd7 Kd3 Qg8 Rh2 Qf7 Rh6 Ke7 Rh2 Qg7 Rh1 Qg8 Rh2 Qf7 Rh1 Qg6 Rh2 Qg7
34/51 00:07 38,960,237 5,303,598 -6.90 Rh7 Qg8 Rh5 Kb7 Ke2 Bc7 Rh2 Qg7 Kd2 Kc8 Rh1 Kd7 Kd3 Qg8 Rh2 Qf7 Rh6 Ke7 Rh2 Qg7 Rh1 Qg8 Rh2 Qf7 Rh1 Qg6 Rh2 Kd7 Rh1 Qg8
35/51 00:12 67,601,655 5,446,914 -6.90 Rh7 Qg8 Rh5 Kb7 Ke2 Bc7 Rh2 Qg7 Kd2 Kc8 Rh1 Kd7 Kd3 Qg8 Rh2 Qf7 Rh6 Ke7 Rh2 Qg7 Rh1 Qg8 Rh2 Qf7 Rh1 Qg6 Rh8 Bd8 Rh1 Kd7 Rh8 Be7 Rh1 Qf7 Rh2 Qg8 Kd2 Qg7 Kd3 Qg6 Ke2 Bd8 Kd3 Be7
36/55 00:59 327,752,735 5,498,007 -6.90 Kf2 Qg8 Ke3 Kb7 Ke2 Kc8 Kd3 Bc7 Kd2 Kd7 Kd3 Qg6 Rh2 Qf7 Rh6 Ke7 Rh2 Qg7 Rh1 Qg8 Rh2 Qf7 Rh1 Bd8 Rh8 Qg7 Rh1 Kd7 Rh2 Bc7 Rh1 Qg8 Rh2 Qf7
5/24/2012 12:09:34 PM, Time for this analysis: 00:01:00, Rated time: 01:00

0 of 1 matching moves
5/24/2012 12:09:35 PM, Total time: 12:01:03 AM
Rated time: 01:00 = 60 Seconds