83 nice Testpositions (Long and big post).

[Uri Blass]

Analysis from c:\epd\tough.epd
Analyzing engine: Hiarcs11.2SPUCI
8/11/2007 12:26:19 AM Level: 720 Seconds
11) Ra4xa8 g7-g8N

illegal move
It seems that hiarcs tried to solve the previous position here.

[Uri Blass]

After enough time movei admits that Qh6 is better than Rbd1 but both moves seem to win
edit:the fail high is wrong fail high and movei go back to Rd1.

I can add that based on looking at the positions I find more than one case when movei is faster than hiarcs(hiarcs fail to solve position number 1 when movei does not fail to do it if you give it 12 minutes)

New game
r1b1r1k1/p2q1p1p/np2np2/3p1N1Q/8/6PB/P4P1P/1R2R1K1 w - - 0 1

Analysis by Movei00_8_438:
1.Qh5-h6
+- (4.26) Depth: 17 00:37:51 841703kN
1.Qh5-h6
+- (4.55) Depth: 17 00:41:06 910233kN
1.Rb1-d1 Ne6-f8 2.Nf5-h6+ Kg8-g7 3.Bh3xd7 Bc8xd7 4.Re1xe8 Ra8xe8 5.Qh5xf7+ Kg7xh6 6.Rd1xd5 Re8-e1+ 7.Kg1-g2 Bd7-c6 8.Qf7xf6+ Nf8-g6 9.Qf6xc6 Na6-b4 10.Qc6-c3 Nb4xd5 11.Qc3xe1 Kh6-g7
+- (4.25) Depth: 17 00:45:38 1002291kN
1.Rb1-d1 Ne6-f8 2.Nf5-h6+ Kg8-g7 3.Bh3xd7 Bc8xd7 4.Re1xe8 Ra8xe8 5.Qh5xf7+ Kg7xh6 6.Rd1xd5 Re8-e1+ 7.Kg1-g2 Bd7-c6 8.Qf7xf6+ Nf8-g6 9.Qf6xc6 Na6-b4 10.Qc6-c3 Nb4xd5 11.Qc3xe1 Kh6-g7
+- (4.25) Depth: 17 00:45:42 1004169kN

(, 14.08.2007)

[Dann Corbit]

There are many mistakes in the test positions.
Here is another mistake:

The following solution is not the only solution and Rbd1 wins the game.
Note that you need to change
fputs("less_reduce 0",evalfile); to
fputs("less_reduce 1",evalfile); in order to get stronger movei

It is possible to get this personality also by changes files
movei_changes438.ini
that include
less_reduce 1

Here is analysis by this personality of movei

r1b1r1k1/p2q1p1p/np2np2/3p1N1Q/8/6PB/P4P1P/1R2R1K1 w - - 0 2 bm Qh6;

New game
[D]r1b1r1k1/p2q1p1p/np2np2/3p1N1Q/8/6PB/P4P1P/1R2R1K1 w - - 0 1

Analysis by Movei00_8_438:

1.Bh3-g2
-+ (-2.57) Depth: 1 00:00:00
1.Bh3-g2 Bc8-b7
-+ (-2.82) Depth: 2 00:00:00
1.Nf5-h6+
-+ (-2.81) Depth: 2 00:00:00
1.Nf5-h6+ Kg8-f8 2.Nh6-f5
-+ (-2.55) Depth: 2 00:00:00
1.Nf5-h6+ Kg8-f8 2.Nh6-f5
-+ (-2.55) Depth: 2 00:00:00
1.Nf5-h6+ Kg8-f8 2.Nh6-g4 Ne6-g7
-+ (-2.59) Depth: 3 00:00:00
1.Nf5-h6+ Kg8-f8 2.Nh6-g4 Ne6-g7
-+ (-2.59) Depth: 3 00:00:00
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 Qd7-a4 3.Nf5-d6
-+ (-2.55) Depth: 4 00:00:00
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 Qd7-a4 3.Nf5-d6
-+ (-2.55) Depth: 4 00:00:00
1.Nf5-h6+
-+ (-2.25) Depth: 5 00:00:00
1.Nf5-h6+ Kg8-g7 2.Bh3-f5 Kg7-f8 3.Qh5-g4 Kf8-e7
-+ (-2.28) Depth: 5 00:00:00
1.Nf5-h6+ Kg8-g7 2.Bh3-f5 Kg7-f8 3.Qh5-g4 Kf8-e7
-+ (-2.28) Depth: 5 00:00:00
1.Nf5-h6+
-+ (-1.98) Depth: 6 00:00:00 36kN
1.Nf5-h6+ Kg8-g7 2.Bh3-f5 Re8-h8 3.Qh5-h4 Na6-c5 4.Rb1-b4
-+ (-1.89) Depth: 6 00:00:00 48kN
1.Nf5-h6+ Kg8-g7 2.Bh3-f5 Re8-h8 3.Qh5-h4 Na6-c5 4.Rb1-b4
-+ (-1.89) Depth: 6 00:00:00 56kN
1.Nf5-h6+
-+ (-1.59) Depth: 7 00:00:00 88kN
1.Nf5-h6+ Kg8-h8 2.Nh6xf7+ Kh8-g7 3.Nf7-d6 Qd7xd6 4.Qh5xe8 Na6-c7 5.Qe8-h5
µ (-1.13) Depth: 7 00:00:00 117kN
1.Nf5-h6+ Kg8-h8 2.Nh6xf7+ Kh8-g7 3.Nf7-d6 Qd7xd6 4.Qh5xe8 Na6-c7 5.Qe8-h5
µ (-1.13) Depth: 7 00:00:01 138kN
1.Nf5-h6+
-+ (-1.43) Depth: 8 00:00:01 280kN
1.Nf5-h6+
µ (-0.83) Depth: 8 00:00:01 307kN
1.Nf5-h6+
µ (-0.83) Depth: 8 00:00:02 340kN
1.Nf5-h6+
³ (-0.53) Depth: 9 00:00:02 522kN
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 h7-h6 3.Qh5xh6+ Kf8-g8 4.Qh6xf6 Na6-c5 5.Nf5-h6+ Kg8-f8 6.Nh6-f5 Kf8-g8
= (0.00) Depth: 9 00:00:02 734kN
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 h7-h6 3.Qh5xh6+ Kf8-g8 4.Qh6xf6 Na6-c5 5.Nf5-h6+ Kg8-f8 6.Nh6-f5 Kf8-g8
= (0.00) Depth: 9 00:00:04 801kN
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 Kf8-g8
= (0.00) Depth: 10 00:00:04 1026kN
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 Kf8-g8
= (0.00) Depth: 10 00:00:06 1162kN
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 Kf8-g8
= (0.00) Depth: 11 00:00:06 1621kN
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 Kf8-g8
= (0.00) Depth: 11 00:00:07 1966kN
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 Kf8-g8
= (0.00) Depth: 12 00:00:08 2870kN
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 Kf8-g8
= (0.00) Depth: 12 00:00:14 5490kN
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 Kf8-g8
= (0.00) Depth: 13 00:00:20 7803kN
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 Kf8-g8
= (0.00) Depth: 13 00:01:02 23462kN
1.Nf5-h6+ Kg8-f8 2.Nh6-f5 Kf8-g8
= (0.00) Depth: 14 00:01:14 28132kN
1.Rb1-d1
= (0.01) Depth: 14 00:03:23 76439kN
1.Rb1-d1
² (0.30) Depth: 14 00:04:10 93609kN
1.Rb1-d1 Qd7-a4 2.Qh5-h6 Na6-c7 3.Qh6xf6 Qa4-d7
+- (2.95) Depth: 14 00:08:56 186925kN
1.Rb1-d1 Qd7-a4 2.Qh5-h6 Na6-c7 3.Qh6xf6 Qa4-d7
+- (2.95) Depth: 14 00:08:58 187726kN
1.Rb1-d1
+- (3.25) Depth: 15 00:10:21 214375kN
1.Rb1-d1 Na6-c5 2.Qh5-h6 Re8-d8 3.Qh6xf6 Qd7-c7 4.Nf5-h6+ Kg8-f8 5.Qf6-h8+ Kf8-e7 6.Nh6-f5+ Ke7-d7 7.Rd1xd5+ Kd7-c6 8.Rd5xd8 Ne6xd8 9.Bh3-g2+ Kc6-b5 10.Qh8-b2+ Kb5-a4 11.Qb2-c2+ Ka4-a5 12.Bg2xa8 Bc8-b7 13.Qc2-d2+ Ka5-a6 14.Ba8xb7+ Nd8xb7
+- (3.38) Depth: 15 00:12:01 247276kN
1.Rb1-d1 Na6-c5 2.Qh5-h6 Re8-d8 3.Qh6xf6 Qd7-c7 4.Nf5-h6+ Kg8-f8 5.Qf6-h8+ Kf8-e7 6.Nh6-f5+ Ke7-d7 7.Rd1xd5+ Kd7-c6 8.Rd5xd8 Ne6xd8 9.Bh3-g2+ Kc6-b5 10.Qh8-b2+ Kb5-a4 11.Qb2-c2+ Ka4-a5 12.Bg2xa8 Bc8-b7 13.Qc2-d2+ Ka5-a6 14.Ba8xb7+ Nd8xb7
+- (3.38) Depth: 15 00:12:14 252730kN
1.Rb1-d1
+- (3.68) Depth: 16 00:14:22 295162kN
1.Rb1-d1 Na6-c5 2.Qh5-h6 Qd7-c7 3.Qh6xf6 Re8-d8
+- (3.72) Depth: 16 00:16:23 335895kN
1.Rb1-d1 Na6-c5 2.Qh5-h6 Qd7-c7 3.Qh6xf6 Re8-d8
+- (3.72) Depth: 16 00:16:53 348681kN
1.Rb1-d1
+- (4.02) Depth: 17 00:21:30 442898kN
1.Rb1-d1 Ne6-f8 2.Nf5-h6+ Kg8-g7 3.Bh3xd7 Bc8xd7 4.Re1xe8 Ra8xe8 5.Qh5xf7+ Kg7xh6 6.Rd1xd5 Re8-e1+ 7.Kg1-g2 Bd7-c6 8.Qf7xf6+ Nf8-g6 9.Qf6xc6 Na6-b4 10.Qc6-c3 Nb4xd5 11.Qc3xe1 Kh6-g7
+- (4.25) Depth: 17 00:27:43 588857kN

(, 14.08.2007)

The scores appear to be similar. Hiarcs also waffled between those two moves:
59) Qh6;
Searching move: Qh5-h6
Best move (Hiarcs11.2SPUCI): Qh5-h6
identical moves! Found in: 03:41
9/21 00:01 216.723 346.202 0.00 Nf5h6+ Kg8f8 Nh6f5 Kf8g8
10/24 00:01 301.239 350.277 0.00 Nf5h6+ Kg8f8 Nh6f5 Kf8g8
11/26 00:02 629.918 350.538 0.00 Nf5h6+ Kg8f8 Nh6f5 Kf8g8
12/29 00:07 2.190.384 325.224 0.00 Nf5h6+ Kg8f8 Nh6f5 Kf8g8
13/32 00:19 6.054.004 328.059 0.00 Nf5h6+ Kg8f8 Nh6f5 Kf8g8
13/34 00:43 13.506.956 318.042 0.00 Rb1d1
13/34+ 00:44 14.033.863 318.040 +0.25 Rb1d1
13/35 00:55 17.339.617 317.883 +1.90 Rb1d1 Na6c5 Qh5h6 Ra8b8 Qh6xf6 Rb8b7 Nf5h6+ Kg8f8 Bh3xe6 Qd7xe6 Qf6h8+ Kf8e7 Re1xe6+ Bc8xe6 Qh8xh7 Ke7d6 Nh6f5+ Kd6c7 Nf5d4 Kc7b8
14/35 01:08 21.839.579 322.874 +1.97 Rb1d1 Na6c5 Qh5h6 Ra8b8 Qh6xf6 Rb8b7 Nf5h6+ Kg8f8 Bh3xe6 Qd7xe6 Qf6h8+ Kf8e7 Re1xe6+ Bc8xe6 Qh8xh7 Rb7b8 Nh6f5+ Ke7d7 Qh7h6
15/39+ 01:46 34.253.098 324.480 +2.22 Rb1d1
15/43 02:59 60.043.238 335.992 +4.89 Rb1d1 Ne6f8 Nf5h6+ Kg8g7 Bh3xd7 Bc8xd7 Qh5xf7+ Kg7xh6 Re1xe8 Ra8xe8 Rd1xd5 Re8e1+ Kg1g2 Bd7c6 Qf7xf6+ Nf8g6 Qf6xc6 Na6b4 Qc6c3 Nb4xd5 Qc3xe1 Kh6g7 f2f4
15/43 03:41 74.528.858 337.448 +4.90 Qh5h6
15/43 04:20 88.210.624 339.904 +6.71 Qh5h6 Na6c5 Rb1d1 Qd7c7 Qh6xf6 Re8d8 Nf5h6+ Kg8f8 Qf6h8+ Kf8e7 Nh6f5+ Ke7d7 Rd1xd5+ Kd7c6 Rd5xd8 f7f6 Qh8xf6 Qc7xd8 Bh3g2+ Kc6b5 Qf6b2+ Kb5a5 Qb2c3+ Ka5a4
16/43 05:31 113.974.643 344.806 +6.95 Qh5h6 Na6c5 Rb1d1 Qd7c7 Qh6xf6 Re8d8 Nf5h6+ Kg8f8 Qf6h8+ Kf8e7 Nh6f5+ Ke7d7 Rd1xd5+ Kd7c6 Rd5xd8
17/43 08:16 173.659.247 350.030 +7.18 Qh5h6 Qd7e7 Nf5xe7+ Re8xe7 Qh6xf6 Re7e8 Bh3f5 Na6c7 Rb1c1 Ne6g7 Rc1xc7 Re8xe1+ Kg1g2 Bc8d7 Rc7xd7 Ra8f8 Rd7xd5 Ng7xf5 Rd5xf5
8/11/2007 11:51:29 AM, Time for this analysis: 00:12:00, Rated time: 6:31:04

[Dann Corbit]

There are many mistakes in the test positions.
{snip}

There are also quite a few positions where Hiarcs chose the suggested position and retained it for quite a while but then discarded it for something better.

There are also positions for which Hiarcs never found the key move but had a high, positive score.

In any case, it is a very interesting test set.

I am puzzled by the first position, because it seems so obvious:
[D]8/6p1/P1b1pp2/2p1p3/1k4P1/3PP3/1PK5/5B2 w - - bm Bg2;
Clearly, if black takes white's bishop, then white can promote the pawn to queen with ease. So black must advance his king to B5. It seems that this does not prevent queening either, and so I am mystified that computer programs can't see it. I guess that there might be a very good defense that I did not think about and that is why some programs never choose this obvious move.

[Uri Blass]

There are many mistakes in the test positions.
{snip}

There are also quite a few positions where Hiarcs chose the suggested position and retained it for quite a while but then discarded it for something better.

There are also positions for which Hiarcs never found the key move but had a high, positive score.

In any case, it is a very interesting test set.

I am puzzled by the first position, because it seems so obvious:
[D]8/6p1/P1b1pp2/2p1p3/1k4P1/3PP3/1PK5/5B2 w - - bm Bg2;
Clearly, if black takes white's bishop, then white can promote the pawn to queen with ease. So black must advance his king to B5. It seems that this does not prevent queening either, and so I am mystified that computer programs can't see it. I guess that there might be a very good defense that I did not think about and that is why some programs never choose this obvious move.

The problem is that black can capture the queen,

1.Bg2 Bxg2 2.e4 f5 3.a7? fxe4 4.a8Q exd3+ and Bxa8 next move is what you probably miss

white needs to play 3.gxf5 exf5 4.a7 fxe4 5.d4 e3 6.dxe5 to win when one of the pawns is unstoppable

Uri

[Dann Corbit]

There are many mistakes in the test positions.
{snip}

There are also quite a few positions where Hiarcs chose the suggested position and retained it for quite a while but then discarded it for something better.

There are also positions for which Hiarcs never found the key move but had a high, positive score.

In any case, it is a very interesting test set.

I am puzzled by the first position, because it seems so obvious:
[D]8/6p1/P1b1pp2/2p1p3/1k4P1/3PP3/1PK5/5B2 w - - bm Bg2;
Clearly, if black takes white's bishop, then white can promote the pawn to queen with ease. So black must advance his king to B5. It seems that this does not prevent queening either, and so I am mystified that computer programs can't see it. I guess that there might be a very good defense that I did not think about and that is why some programs never choose this obvious move.

The problem is that black can capture the queen,

1.Bg2 Bxg2 2.e4 f5 3.a7? fxe4 4.a8Q exd3+ and Bxa8 next move is what you probably miss

white needs to play 3.gxf5 exf5 4.a7 fxe4 5.d4 e3 6.dxe5 to win when one of the pawns is unstoppable

Uri

Here is the sequence that Hiarcs thought after a good long think:

Code: Select all

1) Bg2;                 
    Searching move: Bf1-g2
    Best move (Hiarcs11.2SPUCI): Bf1-g2
    identical moves! Found in: 15:18
    11/22	00:01	     248.343	428.917	+0.51	a6a7 f6f5 Kc2c1 Kb4b5 Bf1g2 e5e4
    12/24-	00:02	     569.362	498.565	+0.26	a6a7
    12/24	00:02	     676.239	514.641	+0.02	a6a7 e5e4 g4g5 e4xd3+ Bf1xd3 f6xg5
    13/25	00:03	   1.260.842	527.107	-0.07	a6a7 e5e4 g4g5 Kb4a5 g5xf6 g7xf6 d3xe4 Ka5b6 Bf1c4 Kb6xa7 Bc4xe6 Bc6xe4+ Kc2d2 Ka7b6
    14/27	00:04	   2.114.928	541.317	-0.26	a6a7 e5e4 d3xe4 Bc6xe4+ Bf1d3 Be4c6
    15/27	00:06	   3.092.811	539.287	-0.27	a6a7 e5e4 d3xe4 Bc6xe4+ Bf1d3 Be4c6 b2b3 Kb4a5 Bd3c4 Ka5b6 Bc4xe6 Kb6xa7 Kc2d3 Ka7b6 Kd3c4 Bc6e4
    16/30	00:09	   4.801.329	560.640	-0.33	a6a7 e5e4 d3xe4 Bc6xe4+ Bf1d3 Be4c6 Bd3a6 Kb4a5 Ba6c8 Ka5b6 Kc2c3 Kb6xa7 Bc8xe6 Ka7b6 Kc3d3 Kb6b5 Be6c4+ Kb5b4
    17/30	00:14	   7.397.644	551.117	-0.30	a6a7 e5e4 d3xe4 Bc6xe4+ Kc2d2 Kb4a5 Bf1c4 Ka5b6 Bc4xe6 Kb6xa7 Be6f5 Be4c6 Kd2d3 Bc6b5+ Kd3e4 Bb5c4 Bf5g6 Ka7b6
    18/32	00:19	  10.706.361	563.937	-0.32	a6a7 e5e4 d3xe4 Bc6xe4+ Bf1d3 Be4c6 Bd3a6 Kb4a5 Ba6c8 Ka5b6 Bc8xe6 Kb6xa7 Kc2d3 Ka7b6 Kd3c4 Bc6f3 Be6d7
    19/35	00:34	  18.607.479	552.085	-0.43	a6a7 e5e4 d3xe4 Bc6xe4+ Kc2d2 Kb4a5 Bf1c4 Ka5b6 Bc4xe6 Kb6xa7 Be6f5 Be4g2 g4g5 f6xg5 e3e4 Ka7b6 e4e5 Kb6c6 e5e6 Kc6d6 Kd2e3
    20/41	00:59	  30.965.329	527.617	-0.43	a6a7 e5e4 d3xe4 Bc6xe4+ Kc2d2 Kb4a5 Bf1c4 Ka5b6 Bc4xe6 Kb6xa7 Be6f5 Be4g2 g4g5 f6xg5 e3e4 Ka7b6 e4e5 Kb6c6 Kd2e3 Kc6d5 e5e6
    21/41	01:55	  54.661.764	478.695	-0.24	a6a7 e5e4 d3xe4 Bc6xe4+ Bf1d3 Be4c6 Bd3a6 Kb4a5 Ba6c8 Ka5b6 Bc8xe6 Kb6xa7 Kc2d3 Ka7b7 g4g5
    22/41	02:51	  81.602.995	479.264	-0.20	a6a7 e5e4 d3xe4 Bc6xe4+ Bf1d3 Be4c6 Bd3g6 Kb4a5 Kc2d3 Ka5b6 Bg6f7 Kb6xa7 Bf7xe6 Ka7b6 g4g5 f6xg5 e3e4 Bc6b5+ Kd3e3 Kb6c7 e4e5 Bb5c6 Ke3d3
    23/47	07:06	 186.468.135	438.231	-0.19	a6a7 e5e4 g4g5 f6f5 g5g6 e4xd3+ Bf1xd3 c5c4
    24/48	10:24	 285.260.611	457.640	-0.19	a6a7 e5e4 g4g5 f6f5 Kc2d2 e4xd3 Bf1xd3 Kb4a5
    24/53	15:18	 414.089.194	451.269	-0.18	Bf1g2
    24/53+	16:57	 452.116.915	444.620	+0.05	Bf1g2
    24/53	23:45	 615.900.565	432.457	+2.98	Bf1g2 Bc6xg2 e3e4 f6f5 g4xf5 e6xf5 a6a7 f5xe4 d3d4 e5xd4 a7a8Q d4d3+ Kc2c1 Bg2f3 Qa8b7+ Kb4a5 Qb7xg7 Ka5b5 b2b3 Kb5c6 Qg7e5
    25/53+	30:44	 758.640.673	411.417	+3.23	Bf1g2
    25/53	31:14	 771.147.274	411.690	+3.23	Bf1g2 Bc6xg2 e3e4 f6f5 g4xf5 e6xf5 a6a7 f5xe4 d3d4 e5xd4 a7a8Q d4d3+ Kc2c1 Bg2f3 Qa8b7+ Kb4a5 Qb7xg7 Ka5b5 Qg7d7+ Kb5b4 Kc1d2
    26/53+	45:49	1.059.168.120	385.346	+3.48	Bf1g2
    26/53	1:02:27	1.449.745.166	386.930	+3.85	Bf1g2 Bc6xg2 e3e4 f6f5 g4xf5 e6xf5 a6a7 f5xe4 d3d4 e5xd4 a7a8Q d4d3+ Kc2c1 Bg2f3 Qa8b7+ Kb4a5 Qb7c7+ Ka5b4 Kc1d2 g7g6 b2b3
    27/54	1:38:56	2.181.029,829	367.464	+4.09	Bf1g2 Bc6xg2 e3e4 f6f5 g4xf5 e6xf5 a6a7 f5xe4 d3d4 e5xd4 a7a8Q d4d3+ Kc2c1 Bg2f3 Qa8b7+ Kb4a5 Qb7xg7 Ka5b5 Qg7d7+ Kb5a5 Kc1d2 Ka5b4 Qd7b7+
   8/14/2007 1:17:34 AM, Time for this analysis: 02:00:00, Rated time: 15:18

[ardee]

Can somebody please explain how 1. Rg4+ wins here? I can only find a draw. For example, 1. Rg4+ fxg4 2. f4+ Kh5 3. Bf7 b3 4. Bxb3 g5 5. Bf7+ Rg6 6. f5 a2

[D]8/4K2p/6pr/5pk1/1pB2R2/p4PP1/8/8 w - - 0 1

Thanks!

[ardee]

Another one whose "solution" I am not yet convinced of:

[D]r1b1r1k1/1p1nnpp1/pq1Bp2p/8/4N3/6Q1/2PRB1PP/5R1K w - - 0 1

I don't see White's advantage after, for example, 1. Bc7 Qc6 2. Rxf7 Kxf7 3. Nd6+ (3. Bh5+ Kg8 4. Bxe8 Qxe4) 3... Kg8 4. Nxe8 Nf5 5. Qg6 Qc3 6. Qxe6+ Kh8

Anybody?

[ardee]

Please forgive my ignorance, but here's one more from that list that I cannot get, even with help from programs:

[D]2b1r3/r2ppN2/8/1p1p1k2/pP1P4/2P3R1/PP3PP1/2K5 w - - 0 1

White can reportedly draw via 1. Nd6+, but after 1. ... exd6 I don't see the draw. (Don't see a way to keep checking or to block Black's other Rook in or to promote a pawn, even beginning with 2. a3).

Got more analysis for this one? For some of the other tactical ones that strong programs like Rybka cannot get?

[ardee]

Now THIS one I can get (with computer help, of course) and I think it's beautiful:

[D]4kr2/5p1K/3p1Q2/1p4P1/4P3/1PP5/7b/8 w - - 0 1

1. Kh6! Be5 (1... Bg3? 2. Kh5 Be5 3. Qf1 and the King escapes) 2. Kg7! Bh2 (2... Bxf6+? 3. gxf6 wins) 3. c4! bxc4 4. e5! Bxe5 5. bxc4 Bh2 6. c5! wins