Why is Houdini 4 so clueless?

[M ANSARI]

I really don't understand this post ??!! Houdini lost a game against SF ... big deal and it happens both ways. The game posted seems like an example where Houdini resigned too early as even at SF depth of 107 !!! the evaluation is still less than +2. The resign setting was probably set too low and even though there might be a white win, it sure is not very clear and certainly worth playing it out. I hate games like that where one engine resigns too early.

Having said that ... it is a well known fact that SF can go deep into a position incredibly fast ... much faster than any engine out there. I remember a pawn push in TCEC that completely busted Houdini and that was on big hardware and LTC, yet Houdini never saw it coming. This does not mean that Houdini is a bad engine but is more a testament to the amazing job the SF team have done in improving and tuning the SF engine. As for SF being a free engine ... well I assure you that "free" is not due to lack of time spent by a group of very talented and dedicated hardcore engine specialists working on the SF project. They are giving their time free of charge, but that does not mean that the engine is of poor quality. Think of it as a "gift" rather than "free".

[shrapnel]

Having said that ... it is a well known fact that SF can go deep into a position incredibly fast ... much faster than any engine out there. I remember a pawn push in TCEC that completely busted Houdini and that was on big hardware and LTC, yet Houdini never saw it coming

True, Ansari bro ; as I've posted elsewhere, Stockfish does have the rare brilliant win against Houdini thanks to the great Depth it reaches, but I've found that Stockfish lacks CONSISTENCY.
Very often, I've found that the superior depths reached by Stockfish rarely translates into victories. Houdini can't reach the Depths reached by Stockfish, but it is rarely beaten by Stockfish and even gets victories at even lower Depths !
Overall, I've found H4 to be a consistent, reliable, well-rounded Engine with a sort of under-stated brilliance ; none of which terms can be applied to Stockfish !
Why, even the proper Settings for various Stockfish parameters seems to be a matter of controversy, with even the Stockfish authors seeming to be reluctant to advise proper settings to people who ask them !?!
No such problems with Houdini, where Houdart very clearly explains what the various settings of Houdini will do !
Overall, H4 seems a more well-rounded, POLISHED Engine than the erratically brilliant Stockfish.
These are some of the reasons I prefer Houdini/ Komodo over Stockfish anyday !

[Michel]

A single game says nothing about the relative strength of SF vs Houdini.

In this game Houdini seems to have beaten itself by resigning in a position where it (correctly) saw it was lost. At that point SF's score wasn't a winning score. I assume SF's positive score reflects its material advantage and the fact that it has the good bishop.

Since the win shown by Finalgen depends on zugzwang SF _might_ have had difficulties actually winning. It would be interesting to have SF and Houdini play this endgame again without the option of resigining.

[ernst]

My post isn't about a single game, about winning or loosing. It's about a pattern I see on a regular basis and that is that Houdini gets outplayed by Stockfish and doesn't have a clue until it's too late. The game posted is merely an example.

What surprises me most is the immaturity of a certain person in this thread. I did not attack Houdini in any way, just expressing my observations and thoughts.

[beram]

Dear Ernst,

I dare you to point out the regularity of your observed "clueless" play phenonomon by Houdini

I would be surprised if you can point out more than 3 games out of 82
In the Richard B Riddick LTC match between Stockfish and Houdini 4
(H4 ct=0 lost only 7 out of 82 games and won 10)
http://www.talkchess.com/forum/viewtopic.php?t=50943

Sofar also rather to my surprise Houdini 4 (ct=0) is leading convincingly here.

But besides that, I observe no "clueless" Houdini 4 in my H4-SF matches neither.
But you must let H4 play with ct=0(!) that really does make a difference in matchplay against Stockfish (and Komodo), see for instance the blitz time control matches of Tom Casanovas: http://www.talkchess.com/forum/viewtopic.php?t=47512

grts Bram

[shrapnel]

Dear Ernst,

I dare you to point out the regularity of your observed "clueless" play phenonomon by Houdini

I would be surprised if you can point out more than 3 games out of 82
In the Richard B Riddick LTC match between Stockfish and Houdini 4
(H4 ct=0 lost only 7 out of 82 games and won 10)
http://www.talkchess.com/forum/viewtopic.php?t=50943

Sofar also rather to my surprise Houdini 4 (ct=0) is leading convincingly here.

But besides that, I observe no "clueless" Houdini 4 in my H4-SF matches neither.
But you must let H4 play with ct=0(!) that really does make a difference in matchplay against Stockfish (and Komodo), see for instance the blitz time control matches of Tom Casanovas: http://www.talkchess.com/forum/viewtopic.php?t=47512

grts Bram

+ 1

[ernst]

Fourth game of the evening, hardware with 10% difference, 6 piece sysygy:

Code: Select all

[Event "Rated game, 5 min"]
[Site "Engine Room"]
[Date "2014.03.11"]
[Round "?"]
[White "Houdini 4 Pro x64"]
[Black "Stockfish 100314 64"]
[Result "0-1"]
[ECO "C89"]
[WhiteElo "2569"]
[BlackElo "2615"]
[Annotator "0.00;0.00"]
[PlyCount "107"]
[EventDate "2014.03.11"]
[TimeControl "300"]


1. e4 {B 0} e5 {B 0} 2. Nf3 {B 0} Nc6 {B 0} 3. Bb5 {B 0} a6 {B 0} 4. Ba4 {
B 0} Nf6 {B 0} 5. O-O {B 0} Be7 {B 0} 6. Re1 {B 0} b5 {B 0} 7. Bb3 {B 0} O-O {
B 0} 8. c3 {B 0} d5 {B 0} 9. exd5 {B 0} Nxd5 {B 0} 10. Nxe5 {B 0} Nxe5 {B 0}
11. Rxe5 {B 0} c6 {B 0} 12. d4 {B 0} Bd6 {B 0} 13. Re1 {B 0} Qh4 {B 0} 14. g3 {
B 0} Qh3 {B 0} 15. Be3 {B 0} Bg4 {B 0} 16. Qd3 {B 0} Rae8 {B 0} 17. Nd2 {B 0}
Re6 {B 0} 18. a4 {B 0} Qh5 {B 0} 19. axb5 {B 0} axb5 {B 0} 20. Nf1 {B 0} Bf5 {
B 0} 21. Qd2 {B 0} Bh3 {B 0} 22. Bd1 {B 0} Qg6 {B 0} 23. Bf3 {B 0} Qf5 {B 0}
24. Bh1 {B 0} h5 {B 0} 25. Bg5 {B 0} Rfe8 {B 0} 26. Rxe6 {B 0} Rxe6 {B 0} 27.
Be3 {B 0} h4 {0.00/26 9} 28. Qd1 {B 0} Rf6 {B 0} 29. Qe1 {B 0} Re6 {B 0} 30.
Ra6 {B 0} Be7 {0.00/27 11} 31. Ra1 {0.00/23 6 (Ta8+)} Bd6 {0.00/31 11} 32. Qc1
{0.00/21 15 (Db1)} Qg6 {B 0 (Rf6)} 33. Qd2 {0.00/21 10} Qh5 {0.00/27 9 (Qf5)}
34. Qd1 {0.00/22 4} Bg4 {0.00/25 1} 35. Qc2 {0.00/23 5 (Dd2)} Bh3 {0.00/29 7 
(Bf5)} 36. Qb1 {-0.13/20 4 (Dd1)} f5 {-1.29/26 23} 37. Qd1 {-0.13/23 0} Bg4 {
-1.58/26 6} 38. Qd2 {-0.13/22 0 (Dc2)} f4 {-1.92/26 7 (Bh3)} 39. Ra8+ {-1.66/
21 13 (Fxd5)} Kh7 {-2.36/26 7} 40. Bxd5 {-2.02/21 4} fxe3 {-2.58/28 2} 41. Nxe3
{-2.10/23 10} cxd5 {-2.60/30 0} 42. Nxg4 {-2.25/23 23} Qxg4 {-2.80/32 0} 43.
Qd3+ {-2.21/23 0} g6 {-2.85/27 8} 44. Kg2 {-2.61/23 16 (Ta7+)} Re7 {-3.41/28 7
(h3+)} 45. b4 {-2.78/23 12 (Ta1)} hxg3 {-6.79/32 17 (h3+)} 46. fxg3 {-4.47/23 7
} Re2+ {-7.23/32 0} 47. Kg1 {-4.47/22 1} Qe6 {-7.31/32 3} 48. Ra7+ {-5.50/24 12
} Kh6 {-8.12/34 0} 49. Rf7 {-6.24/24 12} Rxh2 {-9.68/34 2} 50. Kxh2 {-6.33/23 0
} Qxf7 {-9.81/35 4} 51. Kg2 {-6.35/25 5 (Dd2+)} Qe6 {-12.11/32 8 (Qf5)} 52.
Qd2+ {-7.26/22 4} Kg7 {-21.27/33 0} 53. Qa2 {-5.24/20 1 (Dd3)} Qe4+ {-63.79/39
5} 54. Kf2 {-15.02/28 4 Houdini 4 Pro x64 resigns  (Lag: Av=0.62s,
max=13.6s)} 0-1

[shrapnel]

Big Deal...Here is an online game I played as Black using Houdini 4 and my opponent was using the latest Stockfish. THIS is the real thing, not the 5 minute joke-of-a-match you post
[Event "120s/Move"]
[Site "www.come2play.com"]
[Date "2014.03.12"]
[Round "1"]
[White "Stockfish 100314 64 SSE4.2"]
[Black "Houdini 4 Pro x64"]
[Result "0-1"]
[ECO "D47"]
[PlyCount "154"]

{8192MB, Hiarcs14hBook.ctg, i7 3930k} 1. Nf3 {0} Nf6 {B 0} 2. d4 {39} d5 {B 0}
3. c4 {22} e6 {B 0} 4. Nc3 {47} c6 {B 0} 5. e3 {34} Nbd7 {B 0} 6. Bd3 {35} dxc4
{B 0} 7. Bxc4 {9} b5 {B 0} 8. Bd3 {27} Bb7 {B 0} 9. Bd2 {27} a6 {B 0} 10. a4 {
26} b4 {B 0} 11. Ne4 {9} c5 {B 0} 12. Nxf6+ {34} Nxf6 {B 0} 13. O-O Be7 {0.00/
25 93} 14. Qe2 {0.00/25 19} cxd4 {0.00/25 82} 15. Nxd4 {0.00/25 42} O-O {0.00/
25 62} 16. Rfd1 a5 {-0.12/26 102} 17. Rac1 Qb6 {-0.02/27 90} 18. Be1 e5 {-0.18/
26 75} 19. Nb3 {-0.18/25 9} Rfc8 {-0.19/24 60} 20. Bc4 e4 {-0.14/25 94} 21. h3
g6 {-0.27/27 71} 22. Nd2 Ne8 {-0.47/27 53} 23. Bb3 {-0.53/26 38} Rxc1 {-0.46/
26 45} 24. Rxc1 {-0.49/27 1} Bf6 {-0.47/28 75} 25. Rc2 {-0.47/27 46} Kg7 {-0.
47/29 56} 26. Nc4 {-0.47/28 18} Qc7 {-0.75/35 72} 27. Nd2 Qd8 {-1.06/36 72} 28.
Nc4 {-1.03/34 25} Ba6 {-1.11/38 50} 29. Qg4 {-1.19/32 20} Bc8 {-1.34/35 71} 30.
Qe2 {-1.31/31 9} Be6 {-1.03/35 56} 31. Rd2 Qe7 {-1.06/36 55} 32. Bc2 Qc5 {-1.
82/35 62} 33. Bb3 {-1.90/35 24} Rc8 {-2.10/39 78} 34. Rc2 {-2.09/26 1} Nd6 {-2.
20/39 55} 35. Nxd6 Qxd6 {-2.19/38 21} 36. Rxc8 {-2.19/39 22} Bxc8 {-2.19/39 1}
37. Bc4 {-2.27/39 27} Qe5 {-2.21/40 65} 38. b3 {-2.21/40 25} Qa1 {-2.65/36 43}
39. Qf1 {-1.67/27 26} Qc1 {-1.76/28 22} 40. g4 g5 {-2.11/27 24} 41. Kg2 Qd1 {
-2.73/26 17} 42. Qe2 {-3.01/26 28} Qxe2 {-3.12/27 21} 43. Bxe2 {-2.76/24 1} Bc3
{-2.66/28 21} 44. Kf1 {-2.63/27 21} Be6 {-2.82/26 16} 45. Bd1 {-2.92/27 25} Kf6
{-2.92/27 16} 46. Bc2 {-2.80/27 24} Ke5 {-3.07/25 15} 47. f4+ gxf4 {-3.14/25 17
} 48. Bf2 {-3.36/26 25} h5 {-3.36/26 9} 49. gxh5 {-3.42/26 16} Bxh3+ {-3.42/26
3} 50. Kg1 {-3.42/26 16} Bd2 {-3.41/28 13} 51. h6 {-3.41/25 15} Bf5 {-3.51/27
12} 52. exf4+ {-3.36/27 19} Bxf4 {-3.52/23 4} 53. Bb6 Bxh6 {-3.71/25 15} 54.
Bxa5 Bd2 {-3.89/24 10} 55. Kf1 {-4.10/25 18} Kd4 {-4.35/25 9} 56. Bb6+ Kc3 {-4.
66/24 11} 57. Bd1 {-5.22/24 16} Bg5 {-5.26/23 8} 58. Bc5 e3 {-6.10/22 9} 59.
Bd6 Bd3+ {-12.07/24 6} 60. Be2 Bxe2+ {-136.21/24 8} 61. Kxe2 {-#999/28 6} Bd8 {
-#1000/28 9} 62. Kxe3 Kxb3 {-#1000/33 7} 63. Be5 Kxa4 {-#1000/27 3} 64. Kd2 b3
{-#1000/99 1} 65. Bb2 {-77.58/30 30} f5 {-#1000/99 1} 66. Ke3 {-76.58/29 22}
f4+ {-#1000/99 8} 67. Kxf4 {-75.83/28 25} Be7 {-#1000/99 1} 68. Ke4 {-76.19/29
28} Ba3 {-#1000/99 1} 69. Ba1 Bc1 {-#1000/99 3} 70. Kf5 Ka3 {-#1000/99 6} 71.
Bc3 Bb2 {-#1000/99 2} 72. Bd2 {-76.19/27 27} Ka2 {-#1000/99 1} 73. Kg6 Be5 {
-#1000/99 1} 74. Bc1 Bf4 {-#1000/99 2} 75. Bxf4 {-73.93/24 19} b2 {-#1000/99 1}
76. Kf6 b1=Q {-#1000/99 1} 77. Ke5 Qb5+ {-#1000/99 3} 0-1

[ernst]

You show no understanding of the real situation. Your game was won by black almost right out of book and both engines knew it looking at the evaluation.

Though seeing, they do not see; though hearing, they do not hear or understand

Have fun.

[shrapnel]

Can't help it if my opponent's book sucks...used Book only for first 10-12 moves anyway.....Houdini4 dominated throughout...will post more examples later...