Opening ideas for TCEC 9

[MikeB]

Engines should choose there own opening lines.

This is a good idea too - perhaps alternate between between given lines and letting computers play their own book.

[Vinvin]

I'd prefer very short lines :
1.e4 d5
1.e4 c6
1.e4 e6
1.e4 e5 2.Nf3 Nc6 3.Bc4
1.e4 e5 2.Nf3 Nc6 3.Bb5
1.d4 Nf6
1.d4 d5
1.d4 d6
1.c4 c5
1.c4 e5
1.g3
1.b3
...
There are probably enough here : https://en.wikipedia.org/wiki/Chess_ope ... s_openings

[Leto]

Not that I have a vote , but I actually liked the openings selected this year. I like the process where they take the most favored openings current openings been played by GMs. I also ok with a limited number of openings ( say no more than 15%) that explore gambits where one side side +.8. Also believe some lines should come through the play by the current chess correspondent leaders.

I like that idea about recent openings from top correspondence leaders. I know that at FICGS it is easy to get these openings because the database is freely available and you can even view current active games. Unfortunately many of these games end drawn, the champion at FICGS, Eros Riccio, hasn't lost a game at FICGS since 2010 and his last win was from 2012.

Riccio's next challenger will be Alan Ludgate and this is the win that got him to the championship (well it actually tied the match but since his entry rating was higher than Wosch's he gets the win) :

[Event "FICGS__CHESS__WCH_CANDIDATES_FINAL__000012"]
[Site "FICGS"]
[Date "2015.07.05"]
[Round "1"]
[White "Ludgate,Alan"]
[Black "Wosch,Arkadiusz"]
[Result "1-0"]
[WhiteElo "2442"]
[BlackElo "2284"]

1.c4 Nf6 2.Nc3 e5 3.g3 Bb4 4.Bg2 O-O 5.e4 Bxc3 6.bxc3 d6 7.Ne2 Nbd7 8.O-O a6 9.a4 a5 10.d3 h6 11.f4 b6 12.h3 Nh7 13.g4 Re8 14.Ng3 Ndf8 15.Qf3 f6 16.f5 Bb7 17.h4 Re7 18.Qd1 Nd7 19.d4 Ba6 20.Qb3 Kh8 21.Rd1 Bb7 22.Be3 Qg8 23.Bf3 Qf7 24.Rd2 Ba6 25.d5 Rg8 26.Rh2 Nc5 27.Qa2 R7e8 28.g5 Qe7 29.g6 Nf8 30.Be2 Ra8 31.Qc2 Ra7 32.Rg2 c6 33.Nh5 Rc7 34.Bxh6 Bb7 35.Be3 cxd5 36.cxd5 Nfd7 37.Nxg7 Kxg7 38.h5 Rd8 39.h6+ Kg8 40.g7 Nf8 41.gxf8Q+ Kxf8 42.Rg7 Qxg7+ 43.hxg7+ Rxg7+ 44.Kf2 Kg8 45.Bxc5 bxc5 46.Qb2 c4 47.Qb6 R8d7 48.Bxc4 Bc8 49.Rh1 1-0

It looks like Ludgate used anti-computer tactics here.

Here you can see all the games from that candidates final:
http://www.ficgs.com/user_page.php?page ... AL__000012

[whereagles]

you can also try unusual 1-2 ply openings, e.g.

1. a3, b4, c3, d3, e3, Na3, Nc3, Nh3.
1. Nf3 a6, e4 Nc6...

etc.

[Graham Banks]

you can also try unusual 1-2 ply openings, e.g.

1. a3, b4, c3, d3, e3, Na3, Nc3, Nh3.
1. Nf3 a6, e4 Nc6...

etc.

That would cheapen TCEC in my opinion, as most chessplayers don't embarrrass themselves by playing most of those moves.

[Zenmastur]

The real alternatives are
1) play without opening books and allow learning.
2) play with truly short opening books (2 or 3 moves).

Unfortunately, 1 is not going to happen.

Miguel

An alternative is to play with own opening books and allow learning.

Point 1.

In the super final almost every game (there were a few exceptions) was seeing EGTB hits on the first move out of book. This occurred even though they were only using 5-man table bases. If 6 or 7 man EGTB's are used, the number of hits on the first move out of book would have been much greater. Like wise, if the machine these games were played on was a 40-core machine as originally intended, every single game would have seen large numbers of TB hits on the first move out of book.

Point 2.

Computers are going to get faster, the programs will get better, and it's likely that the EBFs will get smaller. This means that in the future EGTB's could have a huge / potentially decisive influence on the games. The combination of these three factors will make predetermined opening play a thing of the past, unless you want to predetermine the outcome of match.

Point 3.

It would appear that the operators of TCEC already have enough influence to predetermine the out come of any particular match. This would be a “hard” problem if there were greater disparity in the strength of the programs in question or the programs were less deterministic in their play. But since neither of these are the case it would be a simple, although some what tedious process, to find the openings that would allow one program to defeat any particular opponent of similar strength. In the case of SF and Komodo you would need to find less than 10 openings to change the out come of the match.

So my conclusions are these:

1) Hand picked opening that are kept secret should be dis-allowed in any future competition.
2) Not only should learning programs be allowed, they should be preferred over non learning programs when selecting programs to participate.
3) In the super finale a major fraction of the games should be played beginning to end by the programs. IE no outside interference by humans. A lesser fraction of games can be played from predetermined starting positions of shallow depth that are published in advance.

Regards,

Forrest

[jdart]

Here is another one that is very interesting. I think the position after 11. .. b5 is drawn with best play (even though computers eval this as White being better), but finding all the best moves is not trivial.

--Jon

[Event "MT-Muir/A (USA)"]
[Site "ICCF"]
[Date "2012.02.15"]
[Round "?"]
[White "Glazman, Moshe"]
[Black "Stalmach, Kamil"]
[Result "1/2-1/2"]
[ECO "B99"]
[WhiteElo "2577"]
[BlackElo "2567"]
[PlyCount "73"]
[EventDate "2012.??.??"]

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Bg5 e6 7. f4 h6 8. Bh4
Be7 9. Qf3 Nbd7 10. O-O-O Qc7 11. Be2 b5 12. Bxf6 Nxf6 13. e5 Bb7 14. Qg3 dxe5
15. fxe5 Nd5 16. Nxe6 fxe6 17. Qg6+ Kd7 18. Bg4 Qxe5 19. Nxd5 Qg5+ 20. Qxg5
Bxg5+ 21. Kb1 Bxd5 22. Rxd5+ Kc7 23. Rc5+ Kb6 24. Re5 Rhe8 25. Rhe1 Rac8 26.
Bf3 Rc7 27. Rxe6+ Rxe6 28. Rxe6+ Ka7 29. a3 a5 30. c3 Bf6 31. Ka2 a4 32. Be2
Rc5 33. h3 Rg5 34. g4 Rc5 35. Kb1 Kb7 36. Kc2 Kc7 37. Bf1 1/2-1/2

[jdart]

I also like this one. Was played in a recent engine game and also in Tikkan-Perunovic, EU-chT Warsaw 2013. Again probably a draw but complex:

1. d4 d5 2. c4 c6 3. Nf3 Nf6 4. Nc3 e6 5. Bg5 h6 6. Bxf6 Qxf6 7. Qb3 a5 8. e3 a4

[jdart]

Ok, then maybe the 11. Be2 line would be better (this is the main line).

--Jon