Page 1 of 1

Very big bug in LG311_revision14?

PostPosted: 11 Oct 2004, 18:08
by Guenther Simon
[Event "RWBC_Class_A_4th_Edition"]
[Site "ESPRESSO"]
[Date "2004.10.11"]
[Round "17"]
[White "LittleGoliathRevival_311r14"]
[Black "FrancescaMAD_009"]
[Result "0-1"]
[ECO "B08"]
[PlyCount "47"]
[EventDate "2004.??.??"]

1. Nf3 g6 2. d4 Bg7 3. e4 d6 4. Nc3 Nf6 5. Be2 O-O 6. O-O Nc6 7. d5 Nb8 8. a4
c6 9. Bg5 Nbd7 10. Bc4 Nc5 11. dxc6 Ncxe4 12. Bxf6 Nxf6 13. cxb7 Bxb7 14. Qe2
Nh5 15. Bb3 Bxc3 16. bxc3 e5 17. Qe3 Qf6 18. Nd2 Nf4 19. Ne4 Bxe4 20. Qxe4 Rac8
21. g3 Nh3+ 22. Kg2 Ng5 23. h3?? Nxe4 24. c4 {White resigns} 0-1

[diag]2r2rk1/p4p1p/3p1qp1/4p1n1/P3Q3/1BP3P1/2P2PKP/R4R2[/diag]

2r2rk1/p4p1p/3p1qp1/4p1n1/P3Q3/1BP3P1/2P2PKP/R4R2 w - - 0 1

In that position LG always plays h3?? without any thinking and leaves
the whole Queen en prise!
This is reproducable on all my machines under WB and Arena(as UCI).

Code: Select all
4547 >first : force
4547 >first : new
random
4547 >first : ics -
4547 >first : level 40 15 0
4547 >first : post
4547 >first : hard
4547 >first : easy
4547 >first : ping 2
4547 >first : force
4547 >first : setboard 2r2rk1/p4p1p/3p1qp1/4p1n1/P3Q3/1BP3P1/2P2PKP/R4R2 w - - 0 1
4557 <first : pong 2
9063 >first : name Guenther Simon
9063 >first : time 90000
otim 90000
9063 >first : go
9303 <first : move h2h3

Re: Very big bug in LG311_revision14?

PostPosted: 11 Oct 2004, 18:44
by F.Huber
Hi G?nther,

I?ve just tried this here, because I really couldn?t believe it -
but you?re absolutely right: always suddenly this terrible move!

I had the idea, that it might have to do something with Goliath?s book -
and my idea was correct:
After temporarily renaming the book (so LG Revival couldn?t find it any more), this bad move had gone away! :-)

So it seems, that _exactly_ this position is stored somewhere in LG?s book
together with this nonsense move h3, because also any other minimal changes of this position (e.g. put any other piece to another field) wouldn?t give this h3-move!

The good news from my discovery are of course, that not LG _itself_ has the bug but only its book! ;-)

Best regards,
Franz.