Can Houdini 3 be beaten?

You have some major problems with your program there, bub. I don't know if I can help. Maybe you can answer a few questions?

Is this Chess King that you're using? (Could you mention the details of your hardware and chess program?)

What was the time control of the game?

Were Black's moves 6 and 10 opening book moves?

What is the opening book that you're using? Was it supplied by Houdart or the Chess King people?

(I don't know why I keep coming back to this thread. I feel a little like Charlie Brown trying to kick the football that Lucy was holding. )

It is Chess King 3 with no details. All I know about my hardware is that have an i7 CPU Socket 1155  4GB memory  500GB hard drive. Both Roman and my computer guy say this is a pretty good system. There is no other information that came with the chess program.

The time control was 60'40  30'20. In three games my program made 39 moves, ran through the second time control and lost on time. In the last game, I forced the computer to move and it continued the game to checkmate.

Its opening book is "junk" sometimes e.g. it will play: 1.e4 e5 2.Nf3 Nc6 3.Bc4 h6?! 4.d4 exd4 5.Nxd4 Nxd4 6.Qxd4 Qf6? 7.Be3 +/=.

I think the programmers are deliberately handicapping its program to give us poor humans a fair chance!

I have been in regular contact with Diego, the representative of Chess King; he is going to relay my concerns to Hourdard. I assume it was Hourdard who provided the starting moves.

No, Houdart does not provide opening books- just the engine.

Most respectable opening books (free or not) come in Fritz-compatible format (ctg). I don't know if your Kosteniukian product can handle ctg books, though.

Clearly there are problems; here it froze after leaving book; the last move in book was 10...h6

1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 6.Bg5 e6 7.f4 Be7 8.Qf3 Nbd7 9.O-O-O Qc7 10.Bd3 h6 11.Bh4 Rh8 (The computer refused to move so after 12 minutes I forced it to move; usual here, I guess, is b5/Bb7.) 12.Rhe1 g5 Now I tried the unsound sacrifice 13.Bg3? gxf4 14.Bxf4 e5 15.Nf5 exf4 16.Qxf4 Ne5 17.Nxe7 Qxe7 18.Nd5 Nxd5 19.exd5 I missed Qg5 -+

In this game and many others the computer makes the positional mistake in book of playing Nxd4 when Qxd4 strongly centralizes the Queen at an outpost where she is perfectly safe. Here is a fairly typical loss:

Well, you have two basic problems.

1) The engine and/or the opening book playing crap moves. All I can say is check your settings to verify that Houdini is playing at full strength. (From the recent games you've been showing us, Houdini is definitely playing like a patzer at times.) The Chess King (CK) info that I've seen on the web doesn't say which opening book is used, but it does say that the opening book can't be changed. So you're stuck with the one that came with CK.

http://chess-king.com/wp_super_faq/can-you-add-an-opening-book-to-chess-king/

2) The engine just refusing to move at times. Something's broken, but I have no idea what the cause is. Does CK provide any debugging mode or log output so you can see what the engine is doing? At the very least, could you look at the Windows Task Manager when the engine stops to see if the engine/program is using cpu resources? This would at least tell you whether the engine is still calculating or whether it's just idle.

I would almost be tempted to download a free GUI like Arena, Winboard, Scid vs. PC, etc, copy Houdini over to the free GUI, and see if Houdini operates OK in the free GUI. At least at that point you'd have something that's useful to you.

Or why don't you just download Critter 1.6a, (a free downloadable and extremely high strength engine), then download Arena 3.0 and use that. If you do that you will find that you cannot beat or draw it, and we can stop this pointless thread.

I prefer to use the engine I already have; I am waiting for a response from the manufacturer. If Houdart doesn't fix the problem, then I'll go to my computer guy and try your suggestion; I am dialup and will need my computer guy to download the software.  

Right now there appears to be a real problem with the software. Houdini 3 Pro should have an "anti-simplification" algorithm to keep as many pieces on the board as possible, even if means sacrificing material. Then it can use its formidable crunching power to find tactics in the late middlegame and endgame. In the last game I played I used less than four minutes to get out to move 20. Even though I had a two pawn advantage for no counter play, I got sloppy in a winning position and lost. Any Class A player could have won efforlessly.

The Pro will simplify into losing positions for reasons I don't understand. It seems that there are real  problems with positional play. Maybe Houdart rushed this into commercialization before getting the bugs worked out.

Dialup is a royal pain, but it's doable. All of these free chess GUI downloads are 20 MB or less. If you use the free DownThemAll plugin for the Firefox browser, downloading these programs is a snap. You can pause a download and finish it at a later time if you so desire. Even if your internet connection drops out during a download, the partial download up to that point is still saved. The only problem is if the download server doesn't support restarts - If that's the case, then you're screwed, but most download servers support restarts.

Link to DownThemAll plugin:  http://www.downthemall.net/

I seriously doubt that Houdart has done anything wrong. I'm using the free Houdini 1.5a, and it's a monster. I'm sure Houdini 3 is even better. I haven't heard anyone else on the various chess forums complain about Houdini 3, so I'm sure that whatever the problem is, it's not with the design of the engine. It has to be either with your setup or the Chess King program.

Of course, you also need to make sure your engine comes from a reliable outlet...

I will wait a week; if Houdart doesn't get in touch I'll try the suggestion of downloading the free software; my computer guy has broadband and has assured me he will download the software for me. I bought the computer from Chess King.

Here is a cookbook recipe to beat Houdini 3 Pro that anyone who is Class A strength can draw easily or even win.

Never calculate tactics for the first 30 moves. I typically get out to move 20 in five minutes. This is money in the bank. You want as much time in the endgame to win won games and draw equal or inferior positions. If a human pawn leaves a pawn en prise, you might want to calculate to see if is really hanging. You know, beyond a reasonable doubt, that the computer doesn't hang material, so don't calculate the win of material except, of course, if the computer plays the Two Knights Defense (my computer plays the Two Knights Defense at least 1/3 of the time with Black.) or other known gambits.

Second of all, with White always play as aggressively as possible, never defend, always attack. The purpose of the attack except in gambit situations is to acquire controlled space. This is the region in back of your pawns that cannot be accessed by enemy piece.

9 times out of 10 in the opening and early middlegame exchange Bishop for Knight. In universal positions Knights are more valuable than Bishops because they can gain time whereas Bishops can never gain more that two tempos (according to how I define development)

Always try to get as many pawns as  possible on the  third or fourth rank. Five at a minimum, six is better, seven is just winning. The purpose of getting seven pawns on the third (but not in the opening; in the  opening you want to acquire space, or, if Black sacs material,  consolidate the way I have done above.) because this "blinds" the computer. You have to think like a computer. When you have six pawns on the third, the computer is assuming that you will push the pawns rapidly, leap into the center where you engage in hand-to-hand combat.

Push a pawn with your extra interior space, churn your Knights and your other pieces and then push another pawn. The Knights can be used as a modified Hedgehog Attack (This is what I will call a position with six pawns on the third, where two or more sets of minor pieces are exchanged.). The Knights can dance in the center (you will note that I posted a Knight on c4 from which it could not be dislodged) or maneuver in back of your lines)

Finally---the whole point of the Hedgehog attack is to gain a winning advantage that you can convert into a typical Classical Attack. You gain an advantage with Universal chess and prosecute the win with Classical chess.

The whole  point of universal structures is to deny the computer tactics. If you have space, excellent piece coordination, and no weaknesses, there are no tactics more than 2-3 moves deep so you have to trust the structures and never calculate.

One of the basic suppositions of post members is that I am playing a defective machine or that I am "delusional" or a "liar". I provided a cookbook recipe

that will cause chess engines to lose 500 rating points within a year of the general dissemination of these simple rules e.g. Carlsen, using these simple rules will crush Houdini 3 Pro.

In this game you will see the Pro exhibit its 3200 rating. I predict that not one Grandmaster out of 100 will find the Pro's 13th move, c4!! sacing a second pawn and giving me a mobile 4 to 1 pawn majority on the Queenside. I am a 1700 tactician and the Pro took away my strategy and forced me to play tactics.

Ok I'm untracking, I'm finding it laughable that you are telling me to not calculate tactics until past move 30 and everything else you're saying, it's all just very stupid.

"Here is a cookbook recipe to beat Houdini 3 Pro that anyone who is Class A strength can draw easily or even win."

Ffs, draw or win easily? Thanks for the advice on how to beat Houdini, I'm sure it will all work out wonderfully for you when you ACTUALLY play against a full strength Houdini. Until then have fun pretending you're doing something amazing.

Forteen years ago I codified the principles of universal play in my book Universal Chess: The Search for Truth and Beauty. Here are some of the most important findings:

1)A side that cannot be attacked cannot lose,

2)Tactics cannot ensue from the absence of pieces. Passive openings are those that force simplification but don't yield an advantage,

3)Universal play favors the better tactician. The better player rather than the better memorizer will win,

4)The growth of interior space (space behind your pawn wall) facilitates a players attacking prospects by providing great wing to wing mobility without having to pass through the center,

5)In universal play it is possible to avoid most pawn weaknesses,

6)Pawn chains never occur with universal play,

7)Delayed aggression is not passivity,

8)Keeping several pawns abreast assures you of future space i.e. the unstoppable ability of your pawns to advance and create additional interior  space and hence greater mobility,

9)Universal play is not an end unto itself; it is designed to transpose favorably to Classical chess,

10)Move order is flexible in universal chess; the universal player is only concerned where his pieces and pawns belong,

11)A basic principle in universal chess is that once you snatch a pawn, try to return to a universal structure because they are most solid defensive  positions you can obtain.

Here is the latest attempt that I will annotate more thoroughly:

In this game the Pro gave me a 3 to 1 pawn majority on the Queenside but stood better.

I've been trying to beat Houdini 3, but somehow it always manages to draw against me. Darn it!

One indication that I think humans can beat computers is that you have to relearn positional chess. If you think in terms of tactics leaping into the center and engaging in hand-to-hand combat i.e. sharp, tactical chess you will always lose in the middlegame. The key with White is to grab space, develop rapidly, gain some advantage e.g. if your opponent with Black plays a gambit, grab the material and return to a universal position.

I predict computers will repudiate Hypermodern chess, because you cannot cede the center in the opening; the Houdini Pro will convert an advantage in space by "churning" its pieces until it finds a way to break through. It is critical to gain interior space in the middlegame to increase your maneuvability. 

I predict that if you look at a lot of computer/human games what you will see is that the computer will maneuver its Knights very effectively. According to my theories to beat a computer, you must always make either a developing move or a "useful move". Always engage in good piece coordination. If you have good piece coordination, you will never be hit with tactical shots more than two-three moves deep.

Where computers are most vulnerable is the transition from human book to the computer's program as it begins to play chess not book lines. It may not like the position it has and will rearange its pieces to what it thinks is the  best piece layout. That might be a loss in time.

My Houdini 3 would play losing book lines; now Houdini 3 Pro plays even worse in the opening. One line it played led to a dead lost position in ten moves.

While I have never studied Queen pawn openings in any depth, I think I'll try out my theories to see what happens.

While many players regard the Houdini 3 Pro as unbeatable, I just won my fourth game when in a drawn position the computer lost on time. It refused to allow a draw by repetition. This is a chronic problem with the engine. The reason I know this is a serious problem is that the computer, once it lost on time blitzed out 30 moves and I blitzed out 30 moves and the computer's screen read, "Engine lost by time. You can end the game at any moment---you'll be victorious."

Let me just say to my critics---Carlsen thinks the computer plays "perfect" chess is that it understands how to go from Universal chess to Classical chess back to Universal chess. I theorized 14 years ago that this is the way humans would beat computers!

It is my belief that Houdini 3 Pro has massive problems with opening theory that I can exploit. In the 10 game match in the Two Knights' Defense I will present in subsequent posts, I predict that I will win 2 draw 2 and lose 6 for a "performance rating" of over 3000. It is an illusion. The Pro plays the Black side of the Two Knights' Defense at about 2200 level. This is why I think humans can beat computers; if I can spot positional errors and convert them into a win, I suspect a Carlsen, Kramnik or Anand, once they learn the principles of universal theory, will begin to outplay the Pro in middlegames and convert their advantage in the endgame. While I can't  prove it, I suspect that the Pro won't play simple endgames better than 2600-2800  level.

In this 10 game match I lost the first game because I blundered my King. I was 3 pawns up, thought that anything won, and stepped into a mate. In this game I won a pawn in the opening with a huge attack and decided to simplify into a pawn up ending. That was a mistake. The simple ending proved beyond my skill level so I decided to force the draw. I could have claimed a win on time because, once again, the Pro reached the first time control and refused to move so I forced it to move so that it wouldn't lose on time.