Can Houdini 3 be beaten?

i'm not sure about 0% but it is very unlikely. still, keep trying

Nope, it's 0%, because having checked the game with Houdini it's an absolute certainty that he's not playing even a decent 2000 strength player, let alone something with 3300 that has had no official recorded human victory.

I don't know if the OP is trying to fool us, or is just a little deluded, but either way it's getting tiring 

Houdini-Moody August 2013 Match

I was never in this game:

1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. Nf3 O-O 6. Be2 c5 7. d5 Na6 8. O-O Nc7 9. Bf4 Nh5 10. Bg5 f6 11. Be3 Bh8 12. h3 Ng7 13. Qd2 e5 14. dxe6 Ngxe6 15.Rad1 Ne8 16. h4 Bd7 17. h5 g5 18. h6 Bc6 19. Qd3 Qe7 20. Nd5 1-0

Umpf, the religious troll is spreading.

If someone wants to feed the religious troll, do it there : http://www.chess.com/forum/view/general/david-vs-goliath-impossible-win

...because his posts will disappear from here as soon as he is banned (which will be soon, I saw it in my crystal ball).

Houdini played the Ulvestad the very next White, so I refused to play the opening just to prove I can draw and promptly played a horrible Caro-Kann.

I hate the French because in the 20-30 French Defenses  I've played I've never gotten  a decent game. This is no exception.

Moody-Houdini 3  Match 2013

1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 Nfd7 5. Nce2 c5 6. c3 Nc6 7. f4 b5 8. Nf3 b4 9. Bd2 Rb8 10. Nc1 Qb6 11. Be3 bxc3 12. bxc3 Qb2 13. Ne2 Ba6 14. Qc1 Qb6 15.Ng3 cxd4 16. cxd4 Be7 17. Bxa6? Qxa6 18. Kf2 Nb4 19. Ne1 O-O 20. a3 Nc6 21. Nc2 Na5 22. Nb4 Qb5 23. Rb1 Nc4 24. Nc2 Qa4 25. Ne2 Rxb1 26. Qxb1 f6 0-1

Never play an opposite wing attack in the Sicilian against a strong program; the c-file always wins.

Moody-Houdini  August 2013 Match

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Bg5 e6 7. f4 Nbd7 8. Qf3 Qc7 9. O-O-O b5 10. Bd3 Bb7 11. Nce2 Be7 12. Ng3 O-O 13. h4 Rfc8 14. Kb1 h6 15.f5 e5 16. Nde2 b4 17. Bc1 a5 18. c3 Rab8 19. Ka1 Bc6 20. c4 Qb7 21. Rhe1 Nc5 22. Ng1 Ba4 23. b3 Bxb3 24. axb3 Nxb3+ 0-1

I think a super-GM has excellent chances in correspondance, since obviously blunders can be reduced by alot, and their strategic vision will be worth alot. But Houdini is Houdini.

The only way Houdini can be beaten is if it is forced to play human "book". Two openings it would never play thinking for itself---either the Fritz or its book move in the Ulvestad leading to a pawn down middlegame---there the computer can be beaten.

However, once you remove all book, subtle positional errors will creep in, but they are probably too small to be exploited by human players. My Houdini seems to play the white side of the King's Indian poorly, for example.

I can succeed occasionally because I just count tempos, coordinate my pieces, avoid pawn weaknesses but don't calculate more than 2-3 moves ahead. In the last thirty games I played Houdini, I was only hit with one tactical shot more than five moves deep. What you will find with good piece coordination is that it seriously reduces the depth of the tactical  shots.

I hope to be able to play in just one or two tournaments a year---the World  Open or some other big money tournament where Class prizes are like $15,000. The only tournament I played in last year, the New York State Championship I tied for first in the U1800.

Moody-Houdini  Match 2013

After six boring losses I think White missed a draw here in the heavy piece ending.

1. e4 d5 2. exd5 Qxd5 3. Nf3 Bg4 4. Be2 Nf6 5. d4 e6 6. c4 Qd7 7. h3 Bxf3 8.Bxf3 Nc6 9. Be3 Rd8 10. O-O Nxd4 11. Bxb7 Bc5 12. Nc3 O-O 13. Nb5 e5 14. Nxd4 exd4 15. Bg5 c6 16. Bxf6 gxf6 17. Qh5 Qxb7 18. Qxc5 d3 19. b3 Qd7 20. Qe3 Rfe8 21. Qf3 Re6 22. Rfd1 d2 23. Kf1 Rde8 24. Qc3 Re2 0-1

One draw in the Ulvestad---prior to the tournament I was winning one win to four losses---that was playing White only. Playing both colors I have only three draws out of 30 games. That is---three draws out of 15 games with White. Prior to the tournament it was three games out of 12.

In thirty games I've only missed only two tactical shots. The losses are maddeningly similar. Houdini gets a slight plus, builds on its position, shuts down counterplay and then simply strangles you. Most of my games I resign are with equal material, or at most, one pawn down.

Bishops of opposite color endings are tricky. With  a Rook on the board and mobile pawns Houdini was able to convert its advantage easily. In a heavy piece ending often drawn with an extra pawn, with equal material, a far advanced passed pawn gives the player with the initiative real winning chances.

Moody-Houdini 3  Match August 2013

Houdini makes the mistake of playing book in the main line of the Two Knights Defense and White scored an easy draw.

1. e4 e5 2. Nf3 Nc6 3. Bc4 Nf6 4. Ng5 d5 5. exd5 Na5 6. Bb5+ c6 7. dxc6 bxc6 8.Qf3 Be7 9. Bxc6+ Nxc6 10. Qxc6+ Bd7 11. Qf3 O-O?! (Roman and I believe that 11...Rc8 is more accurate) 12. Nc3 Rc8 13. O-O Bc6 14. Qe2 Nd5 15. Nf3 Bd6 16. d3 Nxc3 17. bxc3 Bxf3 18. Qxf3 Rxc3 19. Be3 Rxc2 20. Rfc1 Rxc1+ 21. Rxc1 Ba3 22. Rd1 Qa5 23. d4 exd4 24. Bxd4 Rd8 25. Be3 Rxd1+ 26. Qxd1 h6 27. g3 Bc5 28. Bxc5 Qxc5 29. Qd2 Qf5 30. Qe2 1/2-1/2

Why dont you play an unrated cc game against me I will be allowed to use houdini?

I have a few questions and points for sloughter,
1) What are the time controls for your games?
2) I sometimes fail to even draw Houdini with knight odds in my favour so how exactly are you beating it or drawing it, assuming it's the full strength Houdini? which leads me to my next question
3) are you sure you're playing the full strength Houdini?
4) How is it that Houdini is drawing games on move 30 here? it just says 30.Qe2 1/2 - 1/2, no engine ever agrees to draw, if it wants a draw it repeats the position, so how is it that that game ended in a draw?
5) Usually it won't matter from what opening you play Houdini, it will win, I remember a game between Dzindz and Rybka 3 (about 2900-3000 elo engine) where Dzind had f-pawn odds and first move and still lost ( http://www.youtube.com/watch?v=xejb35JFpvQ ), Houdini 3 is MUCH better than Rybka 3 (by about 300 elo), so you're saying that you're far stronger than a GM? because you seem to be effortlessly drawing some of these games without odds.
6) you say you are using an opening book, this leads Houdini into losing or drawing positions, (which btw if that's true, you're not really beating Houdini), but having examined some of the positions, I'm not seeing any major disadvantages in the eval, meaning Houdini 3 at full strength would probably still beat you (or anyone) from those positions. When I manage to get into an endgame with it on equal terms, I've found that I'll still lose from equal positions because its accuracy begins to compound on my positions.

I'll try to answer your questions:

1) 120'40 60'20 (But most of my games are over by move 20; that is why I can play so many games),

2) If you saw some of the games Houdini has played there would be little doubt that it is an extremely strong program; I doubt it is 3300, but it has found mates in 45,

3) According to Roman my program is "good".

4) I could not have been more surprised when Houdini agreed to a draw; apparently there is some algorithm that says if it is hopelessly drawn, it will accept a draw offer. Like you I thought it would decline the draw offer and I would have to shuffle wood for fifty moves.

5) It depends a lot on what book moves Houdini chooses. If it plays 6...h6? in the Ulvestad, an A player could draw effortlessly at least three different ways.

6)Fifteen years ago when I had my first encounter with "competitive  intuition" I was able to draw GM Lev Alburt on the black side of the Blumenfeld Counter Gambit where Lev is an expert, so even 15 years ago I could occasionally play at GM level. On the other hand, when my game is off, I can lose 20 games in a row to weak players.

Once on the Black side of a Sicilian Defense against Fritz 8 I tripled majors on the fifth rank and allowed Fritz to Queen a pawn just because passive defense was hopeless; I had seen, intuitively that I had a draw by repetition 15 moves from the point where I decided to attack rather than blockade the passed pawn. Twice Fritz 8 refused to play out a draw by repetition and lost on time.

When I was testing variations, I would force Houdini to accept book positions knowing they were winning for White just to challenge existing theory. Now I allow Houdini to pick its own openings.

You are right about equal endings---Houdini is brutal when it has a slight plus in the endgame.

I'll give it a shot---contact me and I'll play---provided I get White :)

Will you play comp4 online in the same time control? Challenge it to 60|30 or something, so we can see the results against a weaker engine?

Sure---how do i do it?

I just resigned a game with Black against Houdini simply because Houdini played my own universal chess against me: 1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.e4 d6 5.Nf3 O-O 6.Be2 c5 7.O-O Na6 8.d5 Nc7 9.h3 e5? 10.a3 Nce8 11.Bg5 a5 12.Ne1! h6 13.Be3 +/- 

You will note that all of White's pieces are in back of his pawns, there are fewer than two half open files or one open file so White has a universal position and is simply winning by just pushing pawns, creating more interior space, pinning Black down defensively and just strangling him.

Black has not one shred of counterplay and is dead lost. He can try Nd7 with a universal position, but the White universal position is vastly superior to the Black universal position.