Houdini Open Theory (HOT)

Sort:
sloughterchess

As a theoretician for the past 30 years, it seems clear that we are an entirely new phase in opening theory. Even a strong chess engine like Fritz 12 is not in the same league as Houdini 3.

What this means is that Houdini could be used by post members to completely redo opening theory in an open source book on chess theory developed by post members using Houdini 3 to create what I call HOT. Every main line in MCO 13, 14 and 15, ECO, BCO and and relevant organizations/publications like New In Chess or chesscafe, etc can be analyzed and published.

The next step is for post members to play and post their innovations in their own games. This would be huge jump in our understanding of opening theory.

sloughterchess
Moses2792796 wrote:

Not really, top players are already using engines to create theoretical novelties, and no doubt engine assisted analysis is published frequently, but even the best engines are still too weak in the opening phase to really revolutionise our basic understanding of the major openings.

 

From personal experience dating back 30 years any pre-2008 literature in sharp openings needs to be flushed through Houdini's brain. An extreme example is MCO 15 where several main lines are completely winning for White even though MCO 15 regards them as "playable".

sloughterchess

But we don't find out most innovations by top professionals until they are played. Most players on the circuit today don't publish their best theory, and, in the good old days, it wasn't at all uncommon for the pros to publish faulty analysis just to trip up an unsuspecting opponent!

What I am proposing is a LINUX style site for analysis and play by players who have vetted their analysis with the Houdini engines. This would be an open source site whose only restriction is that the published games must contain openings vetted by Houdini. Perhaps a further restriction would be that all theory to be published reach a minimum depth of say 25.

Scottrf

What you're probably looking for is a database of centaur style high level correspondence games. I'm sure you can find it if you look.

sloughterchess

This is the first game under the format. Several of my choices differed from those chosen by Houdini. This game has relevance to opening theory because 8...Be7 is regarded by other post members as giving White the greatest difficulty securing a plus.

This game is the first attempt at this new format. It was played against Houdini 3 at 480/30 (Play started on move 9. Typically Houdini was getting out to a Depth of 26 or 27 in the early middle game) 120/10  Game in 30 [But I will not allow a time forfeit by Houdini]. Unlike other variations of the Two Knights’ Defense I have played, this is the main line. This followed my book out to move 17 when Black’s Bxh4 was not my choice).

 

1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Ng5 d5 5.exd5 Na5 6.Bb5ch c6 7.dxc6 bxc6

8.Qf3 Be7 (One of the main lines)

9.Bd3 (The general consensus is that 9.Bxc6ch Nxc6 10.Qxc6ch Bd7 is adequate for equality in key lines and an advantage Black in many lines) O-O

10.Nc3 h6

11.Nge4 (Nigel Short got a good game with 11.h4, but I think it is equal; eventually the Knight will be driven back) Nd5

12.Ng3 Nb4

13.Nf5 Bg5 (Bc5 is another choice)

14.h4 Nxd3ch

15.cxd3 Be7

16.b3! (White wants Black to capture on f5 so that the Rook lift Rh3 is possible) Bxf5

17.Qxf5 (Houdini has this as +.36 or +/=) Bxh4!? (Houdini is saying, “There is no Kingside attack.”)

18.Bb2 Qf6! (Houdini obviously saw this based on its prior move, but it doesn’t solve Black’s long-term pawn structure which is slightly more compromised than the White pawn structure.)

19.Qxf6 Bxf6

20.Ne4 (Houdini still has this as +/=) Be7?!

21.Bxe5 c5

22.Rc1 (Houdini prefers Bc3 and now says it’s equal) Nc6

23.Bc3 (Draw offer---declined---Houdini really doesn’t like to accept draw offers---Houdini is now getting out to a depth of 28 or 29) Rad8

24.Rh3 f5

25.Ng3 Bg5

26.Ne2 Rd5

27.f4 Be7

28.Rc2 (There didn’t appear any way to win this so I thought that with Rc2/Nc1 should make the draw obvious) R8d8

29.Nc1 Bd6

30.Rf3 Nb4

31.Bxb4 cxb4

32.Rc4 & I adjudicated this a draw.

 

sloughterchess

The Fritz main line has been viewed by many on line posters as being equal. HOT disagrees and confers advantage White. In the final position this was a depth of 27/69.

1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Ng5 d5 5.exd5 Nd4 6.c3 b5 7.Bf1 Nxd5 8.Ne4 Ne6 9.Bxb5ch Bd7 10.Bc4 Nb6 11.Bxe6 Bxe6 12.O-O Be713.d4! exd4 (Qd5 14.Re1 O-O 15.dxe5 +/-) 14.cxd4 O-O 15.Nbc3 Nd5

16.Nxd5 Bxd5 17.Nc3 Rb8

18.Qc2 + 0.85 (+/-)White will follow with Be3/Rfe1/Rad1


10.Bc4 Bc6 (In this variation Black can try to restrain d4 for a long time; Houdini prefers an early d3 with a “big” advantage, but play is “anti-positional”; I'd prefer to play “positional” chess with a lesser advantage)

11.O-O Be7 (Nb6 12.Bxe6 Bxe4 13.Bg4 Bd3 14.Be2 +/-)

12.Re1 O-O 13.d4! exd4

14.cxd4 Bb4 (Rb8 15.Nbc3 Re8 16.h3 Rb4 17.b3 Nxc3 18.Nxc3 Qxd4 19.Bxe6 fxe6 20.Rxe6 Qxd1ch 21.Nxd1 Bf6 22.Rxe8ch Bxe8 23.Bb2 Re4 24.Nc3 Re6 25.Rd1 +/=)

15.Nbc3 Nxc3 16.bxc3 Bxe4 17.cxb4 Bd5 18.Bf1 Qd6

19.a3 +.95 at a depth of 27 (+/-)White has an extra pawn but the Knight on e6 temporarily compensates for the Bishop pair.