Is it actually a draw?

[Houdini]

The position is a win but the question is if it is reachable withing 50 moves in realistic conditions. And this question cannot be answered by any engine since it would require at least 7-men TBs.

If Houdini sees a +20 score it is beyond doubt that a win is definitely reachable within the 50 move rule. Syzygy TB take into account the 50-move rule.

[Uri Blass]

It is a big jump from "Houdini cannot produce a stable PV" to your claim that "the position was actually draw despite extremely high evaluation score by SF and H4".
As Houdini and SF are telling you, the position most definitely is a win.

The instability of the PV and score is probably due to insufficient hash memory with respect to the search depth. Feel free to add hash memory, Houdini 4 can use up to 256 GB.

The position is a win but the question is if it is reachable withing 50 moves in realistic conditions. And this question cannot be answered by any engine since it would require at least 7-men TBs. Even 256GB of hash is far from sufficient and I seriously doubt there is anyone here who could afford it. I'm already analyzing with 8GB hash (I could maybe go to 16GB but I don't think there is point).
The only resolution is to actually play at LTC H4 with 6-men vs. SF without TBs and see how it develops.

I did not analyze the game but saying that it is a draw without giving a drawing line in a practical game against stockfish is not convincing.

If you show that fire or another engine can draw at Long Time control against Stockfish or Houdini then it is going to be more convincing.

If you do not show it then my guess is simply that white wins because I choose to believe very high score of chess programs unless I have a good reason to think that the high score is wrong.

Stockfish does not understand KNN vs KP and evaluate it as win for the knights is an example for a good reason.

There is a win but it is practically a draw by the 50 move rule is not a good reason without seeing a practical line in a chess game that leads to draw by the 50 move rule.

[Uri Blass]

The position is a win but the question is if it is reachable withing 50 moves in realistic conditions. And this question cannot be answered by any engine since it would require at least 7-men TBs.

If Houdini sees a +20 score it is beyond doubt that a win is definitely reachable within the 50 move rule. Syzygy TB take into account the 50-move rule.

I believe that it is a win but not 100% sure and the question is if the 50 move rule draw cannot happen before getting to TB.

If you use hash tables to prune then you may remember that position A is mate in 5 and later prune based on this mate when there is no mate because of the 50 move rule.

It is possible to prevent it by not pruning near the 50 move rule even based on hash tables but the question is if houdini does it.

[Milos]

The position is a win but the question is if it is reachable withing 50 moves in realistic conditions. And this question cannot be answered by any engine since it would require at least 7-men TBs.

If Houdini sees a +20 score it is beyond doubt that a win is definitely reachable within the 50 move rule. Syzygy TB take into account the 50-move rule.

After 800 billion nodes and couple of fail-lows the score is back to 10 and is back to root move. How can you actually be sure that the score won't continue decreasing when depth 32 is resolved?

Code: Select all

FEN: 8/Q5k1/5r2/1p3q2/5B2/5R2/7K/8 b - - 17 58 

Houdini 4 Pro x64:
  4/14	00:00	       1'099	0	-3.09	Kg6 Rg3+ Kh5 Rh3+ Qxh3+ Kxh3 Rxf4 Qc5+ Kg6 Qxb5 Kf7
  5/14	00:00	       1'296	0	-3.09	Kg6 Rg3+ Kh5 Rh3+ Qxh3+ Kxh3 Rxf4 Qc5+ Kg6 Qxb5 Kf7
  6/14	00:00	       1'634	0	-3.09	Kg6 Rg3+ Kh5 Rh3+ Qxh3+ Kxh3 Rxf4 Qc5+ Kg6 Qxb5 Kf7
  7/17	00:00	       2'839	2'839'000	-3.22	Kg6 Rg3+ Kh5 Rh3+ Qxh3+ Kxh3 Rxf4 Qc5+ Kg6 Qxb5 Rf5 Qc6+ Kg5
  8/19	00:00	       5'161	469'000	-3.33	Kg6 Rg3+ Kh5 Rh3+ Qxh3+ Kxh3 Rxf4 Qc5+ Kg6 Qxb5 Rf5 Qd3
  9/25	00:00	      36'961	1'155'000	-3.77	Kg6 Rg3+ Kh5 Rh3+ Qxh3+ Kxh3 Rxf4 Qc5+ Kg6 Qxb5 Rf5 Qd3 Kf6 Kg4 Re5 Qd6+ Re6 Qd4+ Re5
 10/25	00:00	      99'064	2'153'000	-3.82	Kg6 Rg3+ Kh5 Rh3+ Qxh3+ Kxh3 Rxf4 Qc5+ Kh6 Kg3 Rf6 Qxb5 Rd6 Qc5 Rg6+ Kf4 Rf6+ Ke5 Kg6 Qc2+ Kg7
 11/25-	00:00	     104'332	1'932'000	-3.98	Kg6 Rg3+
 11/25-	00:00	     105'862	1'764'000	-4.30	Kg6 Rg3+
 11/25-	00:00	     109'239	1'316'000	-7.40	Kg6 Rg3+
 11/29	00:00	     441'950	3'809'000	-10.09	Kg6 Rg3+ Kh5 Rh3+ Qxh3+ Kxh3 Kg6 Bc1 Rf5 Qb6+ Kf7 Bb2 Rf3+ Kg4 Rf1 Qxb5 Rg1+ Kf5 Re1 Qd5+ Ke7 Bf6+ Kf8
 12/30-	00:00	     464'434	2'884'000	-10.50	Kg6 Rg3+
 12/32-	00:00	     513'295	2'804'000	-11.33	Kg6 Rg3+
 12/32	00:00	     557'003	2'785'000	-11.15	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Rh4 Rf5 Qa8+ Kf7 Qb7+ Kf6 Qb6+ Kf7 Qc7+ Kg8 Qc4+ Kg7 Rg4 Rh5+ Kg1 Rg5 Qd4+ Kf7 Qf4+ Qf5 Rxg5 Qxf4 Bxf4 Ke6 Rb5 Kd7 Rxb4 Kc6 Kf2 Kc5 Bd2 Kc6 Kf3 Kd5 Kf4 Kd6 Ke4 Kc7 Rb3 Kd7 Rb6 Kc7 Be3 Kd7 Bf4 Ke7 Rc6 Ke8 Kf5 Ke7 Rc7+ Ke8
 13/78	00:00	     557'021	2'771'000	-11.15	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Rh4 Rf5 Qa8+ Kf7 Qb7+ Kf6 Qb6+ Kf7 Qc7+ Kg8 Qc4+ Kg7 Rg4 Rh5+ Kg1 Rg5 Qd4+ Kf7 Qf4+ Qf5 Rxg5 Qxf4 Bxf4 Ke6 Rb5 Kd7 Rxb4 Kc6 Kf2 Kc5 Bd2 Kc6 Kf3 Kd5 Kf4 Kd6 Ke4 Kc7 Rb3 Kd7 Rb6 Kc7 Be3 Kd7 Bf4 Ke7 Rc6 Ke8 Kf5 Ke7 Rc7+ Ke8
 14/78	00:00	     557'484	2'654'000	-11.15	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Rh4 Rf5 Qa8+ Kf7 Qb7+ Kf6 Qb6+ Kf7 Qc7+ Kg8 Qc4+ Kg7 Rg4 Rh5+ Kg1 Rg5 Qd4+ Kf7 Qf4+ Qf5 Rxg5 Qxf4 Bxf4 Ke6 Rb5 Kd7 Rxb4 Kc6 Kf2 Kc5 Bd2 Kc6 Kf3 Kd5 Kf4 Kd6 Ke4 Kc7 Rb3 Kd7 Rb6 Kc7 Be3 Kd7 Bf4 Ke7 Rc6 Ke8 Kf5 Ke7 Rc7+ Ke8
 15/78	00:00	     562'266	2'392'000	-11.15	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Rh4 Rf5 Qa8+ Kf7 Qb7+ Kf6 Qb6+ Kf7 Qc7+ Kg8 Qc4+ Kg7 Rg4 Rh5+ Kg1 Rg5 Qd4+ Kf7 Qf4+ Qf5 Rxg5 Qxf4 Bxf4 Ke6 Rb5 Kd7 Rxb4 Kc6 Kf2 Kc5 Bd2 Kc6 Kf3 Kd5 Kf4 Kd6 Ke4 Kc7 Rb3 Kd7 Rb6 Kc7 Be3 Kd7 Bf4 Ke7 Rc6 Ke8 Kf5 Ke7 Rc7+ Ke8
 16/78	00:00	     573'821	2'332'000	-11.15	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Rh4 Rf5 Qa8+ Kf7 Qb7+ Kf6 Qb6+ Kf7 Qc7+ Kg8 Qc4+ Kg7 Rg4 Rh5+ Kg1 Rg5 Qd4+ Kf7 Qf4+ Qf5 Rxg5 Qxf4 Bxf4 Ke6 Rb5 Kd7 Rxb4 Kc6 Kf2 Kc5 Bd2 Kc6 Kf3 Kd5 Kf4 Kd6 Ke4 Kc7 Rb3 Kd7 Rb6 Kc7 Be3 Kd7 Bf4 Ke7 Rc6 Ke8 Kf5 Ke7 Rc7+ Ke8
 17/78	00:00	     615'170	2'375'000	-11.15	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Rh4 Rf5 Qa8+ Kf7 Qb7+ Kf6 Qb6+ Kf7 Qc7+ Kg8 Qc4+ Kg7 Rg4 Rh5+ Kg1 Rg5 Qd4+ Kf7 Qf4+ Qf5 Rxg5 Qxf4 Bxf4 Ke6 Rb5 Kd7 Rxb4 Kc6 Kf2 Kc5 Bd2 Kc6 Kf3 Kd5 Kf4 Kd6 Ke4 Kc7 Rb3 Kd7 Rb6 Kc7 Be3 Kd7 Bf4 Ke7 Rc6 Ke8 Kf5 Ke7 Rc7+ Ke8
 18/78	00:00	     617'146	2'219'000	-11.15	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Rh4 Rf5 Qa8+ Kf7 Qb7+ Kf6 Qb6+ Kf7 Qc7+ Kg8 Qc4+ Kg7 Rg4 Rh5+ Kg1 Rg5 Qd4+ Kf7 Qf4+ Qf5 Rxg5 Qxf4 Bxf4 Ke6 Rb5 Kd7 Rxb4 Kc6 Kf2 Kc5 Bd2 Kc6 Kf3 Kd5 Kf4 Kd6 Ke4 Kc7 Rb3 Kd7 Rb6 Kc7 Be3 Kd7 Bf4 Ke7 Rc6 Ke8 Kf5 Ke7 Rc7+ Ke8
 19/78	00:00	     619'572	2'151'000	-11.15	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Rh4 Rf5 Qa8+ Kf7 Qb7+ Kf6 Qb6+ Kf7 Qc7+ Kg8 Qc4+ Kg7 Rg4 Rh5+ Kg1 Rg5 Qd4+ Kf7 Qf4+ Qf5 Rxg5 Qxf4 Bxf4 Ke6 Rb5 Kd7 Rxb4 Kc6 Kf2 Kc5 Bd2 Kc6 Kf3 Kd5 Kf4 Kd6 Ke4 Kc7 Rb3 Kd7 Rb6 Kc7 Be3 Kd7 Bf4 Ke7 Rc6 Ke8 Kf5 Ke7 Rc7+ Ke8
 20/78	00:00	     623'111	2'090'000	-11.15	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Rh4 Rf5 Qa8+ Kf7 Qb7+ Kf6 Qb6+ Kf7 Qc7+ Kg8 Qc4+ Kg7 Rg4 Rh5+ Kg1 Rg5 Qd4+ Kf7 Qf4+ Qf5 Rxg5 Qxf4 Bxf4 Ke6 Rb5 Kd7 Rxb4 Kc6 Kf2 Kc5 Bd2 Kc6 Kf3 Kd5 Kf4 Kd6 Ke4 Kc7 Rb3 Kd7 Rb6 Kc7 Be3 Kd7 Bf4 Ke7 Rc6 Ke8 Kf5 Ke7 Rc7+ Ke8
 21/78	00:00	     626'873	2'035'000	-11.15	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Rh4 Rf5 Qa8+ Kf7 Qb7+ Kf6 Qb6+ Kf7 Qc7+ Kg8 Qc4+ Kg7 Rg4 Rh5+ Kg1 Rg5 Qd4+ Kf7 Qf4+ Qf5 Rxg5 Qxf4 Bxf4 Ke6 Rb5 Kd7 Rxb4 Kc6 Kf2 Kc5 Bd2 Kc6 Kf3 Kd5 Kf4 Kd6 Ke4 Kc7 Rb3 Kd7 Rb6 Kc7 Be3 Kd7 Bf4 Ke7 Rc6 Ke8 Kf5 Ke7 Rc7+ Ke8
 22/78	00:01	  21'131'850	11'092'000	-11.24	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Rh4 Rf5 Qa8+ Kf7 Qb7+ Kf6 Qb6+ Kf7 Qc7+ Kg8 Qc4+ Kg7 Rg4 Rh5+ Kg1 Rg5 Qd4+ Kf7 Qf4+ Qf5 Rxg5 Qxf4 Bxf4 Ke6 Rb5 Kd7 Rxb4 Kc6 Kf2 Kc5 Bd2 Kc6 Kf3 Kd5 Be3 Kd6 Ke4 Kc7 Rb6 Kd7 Bf4 Ke7 Rc6 Ke8 Kf5 Ke7 Rc7+ Ke8
 23/79-	00:02	  29'139'389	11'012'000	-11.70	Kg8 Rg3+
 23/86-	00:03	  44'892'818	11'798'000	-12.63	Kg8 Rg3+
 23/103-	00:28	 436'188'532	15'467'000	-23.15	Kg8 Rg3+
 23/103	00:37	 574'807'229	15'340'000	-13.35	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Kf7 Rh4 Ke6 Qc6+ Kf7 Rf4 Rh8+ Kg2 Rh5 Rxf5+ Rxf5 Qc4+ Kf6 Qxb4 Kf7 Qc4+ Kf6 Bh4+ Ke5 Qc3+ Kd5 Bg3 Ke6 Qc4+ Kf6 Bh4+
 24/103-	00:37	 578'339'319	15'306'000	-13.91	Kg8 Rg3+
 24/103-	00:38	 580'789'715	15'275'000	-15.02	Kg8 Rg3+
 24/103-	00:44	 674'210'197	15'291'000	-27.69	Kg8 Rg3+
 24/103	00:49	 757'422'262	15'228'000	-16.53	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Kf7 Rh4 Kg6 Be5+ Kf7 Bb2 Ke6 Rh6+ Kf7 Qb7+ Ke8 Qc6+ Kf7 Rf6+ Qxf6 Qxf6+ Ke8 Qe6+ Kd8 Bd4 Kc7 Qe7+ Kc6 Qc5+ Kb7 Qxb4+ Kc6 Qc5+ Kd7 Qxf8 Ke6 Kg3 Kd5 Qc5+ Ke4 Qe5+ Kd3 Qe3+ Kc4
 25/103-	00:50	 769'423'991	15'123'000	-17.22	Kg8 Rg3+
 25/103-	00:53	 799'721'957	15'049'000	-18.59	Kg8 Rg3+
 25/103-	00:53	 803'112'161	15'034'000	-22.56	Kg8 Rg3+
 25/103+	01:46	1'645'920'427	15'397'000	-15.85	Kg8
 25/103+	02:22	2'189'464'774	15'373'000	-14.48	Kg8
 25/103	02:57	2'654'937'516	14'971'000	-13.53	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qa7+ Kg8 Qa2+ Kg7 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg5 Qa2+ b3 Qxb3+ Kg7 Qb7+ Rf7 Qb2+ Kg6 Be5 Qg4 Qc2+ Kg5 Qc1+ Rf4 Rg3 Kf5 Qc8+ Kxe5
 26/103-	02:58	2'667'806'417	14'951'000	-14.09	Kg8 Rg3+
 26/103-	02:58	2'670'031'900	14'939'000	-15.22	Kg8 Rg3+
 26/103-	03:01	2'717'522'952	14'933'000	-28.02	Kg8 Rg3+
 26/103	03:30	3'149'789'880	14'978'000	-16.58	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qb7+ Kg8 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg5 Qa2+ b3 Qxb3+ Rf7 Qa2 Qf6 Rh4 Qg6 Qc4 Qg5 Rg4 Qxg4
 27/103-	03:32	3'173'932'683	14'959'000	-17.27	Kg8 Rg3+
 27/103-	03:32	3'183'644'191	14'950'000	-18.65	Kg8 Rg3+
 27/103-	03:33	3'192'014'831	14'942'000	-22.66	Kg8 Rg3+
 27/103	06:39	6'053'374'180	15'144'000	-16.43	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qb7+ Kg8 Qg2+ Kf7 Bg3 Qg6 Qa2+ Re6 Rh4 Kg7 Qb2+ Qf6 Qxb5 Qh6
 28/103-	07:16	6'639'777'306	15'222'000	-17.10	Kg8 Rg3+
 28/103-	07:17	6'657'142'788	15'215'000	-18.46	Kg8 Rg3+
 28/103-	07:18	6'671'271'321	15'209'000	-22.41	Kg8 Rg3+
 28/103	10:36	9'748'161'347	15'315'000	-16.67	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qb7+ Kg8 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Qd5+ Rf7 Be1 Kg7 Qe5+ Rf6 Rg3 Kh7 Rxg6 Kxg6 Bxb4 Rf2+ Kg3 Rf1 Qe6+ Kg7 Qg4+ Kf6 Bc3+ Ke7 Qe2+ Kd6 Qxf1 Kc5 Kg4 Kd5 Qd3+ Kc6 Qc4+ Kb6
 29/103-	10:41	9'808'243'574	15'301'000	-17.36	Kg8 Rg3+
 29/103-	10:58	10'014'220'736	15'211'000	-18.74	Kg8 Rg3+
 29/103-	11:01	10'052'754'271	15'207'000	-22.75	Kg8 Rg3+
 29/103	27:19	26'036'608'142	15'883'000	-318.00	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qb7+ Kg8 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Qd5+ Rf7 Be1 Kg7 Qe5+ Rf6 Rg3 Kh7 Rxg6 Kxg6
 30/103	1:47:23	104'732'525'413	16'254'000	-318.00	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qb7+ Kg8 Qg2+ Kf7 Bg3 b4 Qb7+ Kg8 Qa8+ Rf8 Qg2 Qg6 Qd5+ Rf7 Be1 Kg7 Qe5+ Rf6 Rg3 Kh7 Rxg6 Kxg6
 31/103+	2:13:13	130'954'577'895	16'382'000	-168.44	Kg8
 31/103+	2:16:14	133'844'945'759	16'373'000	-154.40	Kg8
 31/103+	2:20:28	137'851'184'263	16'355'000	-112.42	Kg8
 31/112	2:44:53	161'103'567'505	16'283'000	-16.28	Kg8 Rg3+ Rg6 Qb8+ Kf7 Qc7+ Kg8 Rh3 Rf6 Qb8+ Kf7 Qb7+ Kg8 Qg2+ Kf7 Bg3 Re6 Bd6 Qe4 Qf2+ Rf6 Qa7+ Kg8 Qb8+ Kf7 Qf8+ Ke6 Qe7+ Kd5 Qb7+ Kd4 Qb6+ Kd5 Rh5+ Rf5 Qc5+ Ke6 Rh6+ Rf6 Qc8+ Kd5 Rh5+ Kd4 Qc5+ Kd3 Rd5+ Qxd5 Qxd5+ Kc2 Qe4+ Kc3 Qe5+ Kc4 Qxf6 Kb3 Qf4 Kc2 Qf5+ Kc3 Qxb5 Kc2 Kg3
 32/112-	2:51:55	167'507'140'767	16'238'000	-16.95	Kg8 Rg3+
 32/112-	2:53:39	169'021'960'831	16'221'000	-18.30	Kg8 Rg3+
 32/112-	2:53:52	169'219'895'430	16'221'000	-22.21	Kg8 Rg3+
 32/112+	11:53:41	712'914'514'811	16'649'000	-15.61	Kg8
 32/112+	13:34:05	806'714'263'148	16'515'000	-10.35	Kg8

[Milos]

I did not analyze the game but saying that it is a draw without giving a drawing line in a practical game against stockfish is not convincing.

If you show that fire or another engine can draw at Long Time control against Stockfish or Houdini then it is going to be more convincing.

If you do not show it then my guess is simply that white wins because I choose to believe very high score of chess programs unless I have a good reason to think that the high score is wrong.

Stockfish does not understand KNN vs KP and evaluate it as win for the knights is an example for a good reason.

There is a win but it is practically a draw by the 50 move rule is not a good reason without seeing a practical line in a chess game that leads to draw by the 50 move rule.

How can you show 50 move rule draw line for god sake????
You need depth 100 for that...

[Houdini]

After 800 billion nodes and couple of fail-lows the score is back to 10 and is back to root move. How can you actually be sure that the score won't continue decreasing when depth 32 is resolved?

Play the PV moves to get more insight.
From your initial diagram: 1...Kg7 2.Qa7+ Kg8 3.Rg3+ Rg6 4.Qb8+ Kf7 5.Qc7+ Kg8 6.Rh3 Rf6 7.Qb8+ Kf7 8.Qb7+ Kg8 9.Qg2+ Kf7
[d]8/5k2/5r2/1p3q2/5B2/7R/6QK/8 w - -
You will find a "M1000" score which indicates a Syzygy win.
Both Bg3 and Rh4 appear to be winning, quite a remarkable ending.

[zullil]

After 800 billion nodes and couple of fail-lows the score is back to 10 and is back to root move. How can you actually be sure that the score won't continue decreasing when depth 32 is resolved?

Play the PV moves to get more insight.
From your initial diagram: 1...Kg7 2.Qa7+ Kg8 3.Rg3+ Rg6 4.Qb8+ Kf7 5.Qc7+ Kg8 6.Rh3 Rf6 7.Qb8+ Kf7 8.Qb7+ Kg8 9.Qg2+ Kf7
[d]8/5k2/5r2/1p3q2/5B2/7R/6QK/8 w - -
You will find a "M1000" score which indicates a Syzygy win.
Both Bg3 and Rh4 appear to be winning, quite a remarkable ending.

But does the Syzygy win happen in few enough moves to avoid a 50-move draw, given that some number of moves were needed to reach this position?

[syzygy]

But does the Syzygy win happen in few enough moves to avoid a 50-move draw, given that some number of moves were needed to reach this position?

I cannot really speak for H4 here, but it is quite likely that H4 during its search treats TB wins very similar to how SF treats them.

When SF finds a TB win during its search, it keeps track of the number of ply from the root to that TB win. Through hashtable interactions it can happen that it finds a TB win in x ply where x is much larger than necessary, but it cannot happen that it finds a TB win in x ply where x is smaller than necessary.

So whenever SF sees a TB win in x ply with x <= 100, there is definitely a TB win and it will also find it unless strange things happen (like not enough time to find a win-in-29 ply even though it already found a win-in-31 ply at the previous move).

Compare this to engines reporting a mate. Did you ever see a non-buggy engine report a mate-in-x and then not be able to play out the mate in x moves or less? Often it will need less than x moves, but it should never need more than x moves.

In theory it is possible that SF finds a TB win in x ply with x > 100 with the last part of it in the hashtable (say 70 ply in search where it finds a win-in-40 ply in the hash), so that SF will not know whether this win stays within the 50-move rule. SF will treat this as a win and it might really be a draw. But this can happen as well with real mates, at least in theory. In practice it does not happen.

[zullil]

But does the Syzygy win happen in few enough moves to avoid a 50-move draw, given that some number of moves were needed to reach this position?

I cannot really speak for H4 here, but it is quite likely that H4 during its search treats TB wins very similar to how SF treats them.

When SF finds a TB win during its search, it keeps track of the number of ply from the root to that TB win. Through hashtable interactions it can happen that it finds a TB win in x ply where x is much larger than necessary, but it cannot happen that it finds a TB win in x ply where x is smaller than necessary.

So whenever SF sees a TB win in x ply with x < 100, there is definitely a TB win and it will also find it unless strange things happen (like not enough time to find a win-in-29 ply even though it already found a win-in-31 ply at the previous move).

Compare this to engines reporting a mate. Did you ever see a non-buggy engine report a mate-in-x and then not be able to play out the mate in x moves or less? Often it will need less than x moves, but it should never need more than x moves.

OK. But if the shortest win here needs say 49 moves, then this line is valueless in the position as originally posted. Since I have no idea what M1000 from H4 specifically means, I'm not sure that seeing it guarantees a win in the position as originally posted.

What does SF+Syzygy give as evaluation here?

[d]8/5k2/5r2/1p3q2/5B2/7R/6QK/8 w - -

[Uri Blass]

I did not analyze the game but saying that it is a draw without giving a drawing line in a practical game against stockfish is not convincing.

If you show that fire or another engine can draw at Long Time control against Stockfish or Houdini then it is going to be more convincing.

If you do not show it then my guess is simply that white wins because I choose to believe very high score of chess programs unless I have a good reason to think that the high score is wrong.

Stockfish does not understand KNN vs KP and evaluate it as win for the knights is an example for a good reason.

There is a win but it is practically a draw by the 50 move rule is not a good reason without seeing a practical line in a chess game that leads to draw by the 50 move rule.

How can you show 50 move rule draw line for god sake????
You need depth 100 for that...

very simple
play against stockfish(when you can use other programs to help you).
draw by the 50 move rule and post the game.

If you can do it then it may be a convincing evidence to support the claim that the position is a draw.
Otherwise you have no evidence.

Uri