On Jan 29, 4:42*pm, Sanny wrote:

White -- Black
(beginner) -- (Stinky Garlnoot)


So far so good.

15. g2-g4{6} c6-d5{68}

In heavens name why? This is a horrible losing move!
It wrecks a promising position by giving away the d pawn for no
compensation.
Strongest was 15. d6 with equality but Bxf6, c4 or dxc6 might be
tolerable.

Why didn't the engine see the value in making a passed pawn (if only
to tie down his opponents pieces)?

16. b2-b3{12} d5-d4{70}
17. Kg1-h1{6} h7-h5{328}

And again the king goes to cower in the corner whilst the attack
mounts. If he was to move then Kg2 would be the best choice of square,
but 17. exd4 would keep the position more nearly together.

19. f2-f3{6} d4-d3{252}

19. f3 is terminally stupid and demostrates a serious problem with the
programs look ahead.

After f3 the B has no escape squares and so is doomed. A quiessence
search should see this at ply 3.

19. f3 d3 20. cxd3 cxf3 21. Bxf3 hxg4 22. Bxf6 Bxf6 23. Qxf3
(longest sequence of captures favourable or equal to find quiessent
position at ply 3 search)

later on

24. Qg2 looks surprisingly to be playable (but the position is already
lost)

27. Nf4-g6{6} Bf5-g6{110}

Moving the N gives away the d pawn. Better is Bxf6 (although it
doesn't help much)

Whites d pawns seem to be very unlucky in this game.

Regards,
Martin Brown

So far so good.

15. g2-g4{6} c6-d5{68}

In heavens name why? This is a horrible losing move!
It wrecks a promising position by giving away the d pawn for no
compensation.
Strongest was 15. d6 *with equality but Bxf6, c4 or dxc6 might be
tolerable.

It played 15. g4 as BxN(h3) Then gxh3 Would made Double pawn at h
file. and playing d6 the pawn would not be defendable what if Black
plays R-d8 followed by Ne8. How will we defend the pawn at d6 then?

And one pawn was always going. See 15. Qxb2. So there was no way to
save one Pawn.

If we save d5 then b2 will go. If we save b2 then d5 would go. And
BxN(h3) would have made double pawn at h3 and h2. So to avoid double
pawn formation weakening the King it played 15. g2-g4

19. f2-f3{6} d4-d3{252}

19. f3 is terminally stupid and demostrates a serious problem with the
programs look ahead.

Beginner Level sees 8-10 depth deep and this move was 12 depth long so
it was unable to see the threath. I hope Easy level would have seen
such threath.

Bye
Sanny

Play Chess at: http://www.GetClub.com/Chess.html

On Jan 30, 6:27*am, Sanny wrote:
So far so good.

15. g2-g4{6} c6-d5{68}

In heavens name why? This is a horrible losing move!
It wrecks a promising position by giving away the d pawn for no
compensation.
Strongest was 15. d6 *with equality but Bxf6, c4 or dxc6 might be
tolerable.

It played 15. g4 as BxN(h3) Then gxh3 Would made Double pawn at h
file. and playing d6 the pawn would not be defendable what if Black
plays R-d8 followed by Ne8. How will we defend the pawn at d6 then?

15. d6 Rd8 (expected) 16. Bxf6 Bxf6 (the N can't go to e8 now) 17. Qd2

I am sorry Sanny but g4 is just horrible. Far better to end up with
doubled pawns on the h file than waste the valuable d pawn without any
compensation at all.

Even better would be to put some kind of kludge in your evaluation
function to encourage N to develop onto c & f and discourage them from
a & h. A 20cp bonus for Nc3/f3/c6/f6 ought to do it when move no 20.

6. Be2, 7. Bf3 then 8. Nh3 is pretty bad opening style. 6. Be2 7. Nf3
would be *much* better.
Playing Nh3 allowed black to open your kingside castled position with
a tin opener BxN.

Also Be2, Bf3 when there is no clear threat to the B on e2 should be
penalised in the evaluation function. Two consecutive moves of the
same minor piece when one minor piece is still not developed.

And one pawn was always going. See 15. Qxb2. So there was no way to
save one Pawn.

If we save d5 then b2 will go. If we save b2 then d5 would go. And
BxN(h3) would have made double pawn at h3 and h2. So to avoid double
pawn formation weakening the King it played 15. g2-g4

b2 is not under threat. It is a poisonned pawn. You win the queen
after 15. .... Qxb2
That pattern is well known in certain Sicilian opening theory.

eg. 15. d6 Qxb2 16. Rfb1 (queen has no safe squares) Qxa1 17. Rxa1
is about the best black can hope for and he is dead in the water.

The queen is so hopelessly stuck on b2 you even have time to fit in
16. Bxf6 Bxf6 before Rfb1

Black still has a pretty good game with correct play even with whites
best move, but at least white is not losing.
Black has a wide choice of good continuations of which h6 probably the
best, Bxh3, a5, c5, Be6, h5 also good and amazingly Bf8 is just about
playable. The N isn't quite en prise eg.

15. .... Bf8 16. Bxf6 Rxd6! 17. Qc3 Rxf6

19. f2-f3{6} d4-d3{252}

19. f3 is terminally stupid and demostrates a serious problem with the
programs look ahead.

Beginner Level sees 8-10 depth deep and this move was 12 depth long so
it was unable to see the threath. I hope Easy level would have seen
such threath.

It was nothing like that depth. Ply 8 with or without a proper
quiessence swapoff evaluation should still see it.

And all the other moves beyond that are captures. Your program should
never stop part way through a series of exchanges without giving the
other side the option of capturing any higher or equal value piece
until all such captures are exhausted. It would explain why it mostly
plays OK but then makes crass mistakes in these types of positions.
You have a horizon effect feeding bad values into terminal nodes
somewhere.

Regards,
Martin Brown

With little more improvents GetClub played much better today but
Beginner Level lost to Jester.

Here is the game Below. You can see the moves by Beginner is now much
better. But still it lost 2 Pawns and White was able to Queen with
Extra Pawns.

Was there any way to save those Pawns?

Game Played between Jester and beginner at GetClub.com

--------------------------------------------------------------------------------
Jester: (White)
beginner: (Black)
Game Played at: http://www.GetClub.com/Chess.html
View Recorded Game: http://www.getclub.com/playgame.php?...313&game=Chess
--------------------------------------------------------------------------------

White -- Black
(Jester) -- (beginner)

1. e2-e4{6} e7-e5{2}
2. Nb1-c3{6} Ng8-f6{0}
3. Ng1-f3{6} Nb8-c6{0}
4. g2-g3{6} Bf8-c5{10}
5. Bf1-g2{8} Ke8-g8{6}
6. d2-d3{12} d7-d6{14}
7. Nc3-a4{10} Bc5-b6{18}
8. Na4-b6{10} a7-b6{10}
9. h2-h3{10} d6-d5{22}
10. e4-d5{10} Nf6-d5{20}
11. Ke1-g1{10} h7-h6{10}
12. Rf1-e1{12} Rf8-e8{12}
13. Bc1-d2{10} Nd5-f6{22}
14. Bd2-c3{12} Qd8-d6{22}
15. Nf3-d2{10} Bc8-e6{22}
16. a2-a3{14} Re8-d8{6}
17. Qd1-e2{14} Nc6-d4{8}
18. Bc3-d4{12} e5-d4{10}
19. Nd2-e4{18} Nf6-e4{6}
20. Qe2-e4{10} Be6-d7{6}
21. Ra1-d1{32} Bd7-c6{6}
22. Qe4-e7{28} Bc6-g2{8}
23. Qe7-d6{10} Rd8-d6{10}
24. Kg1-g2{10} f7-f5{18} {Is this move Wrong?}
25. Re1-e7{22} Rd6-c6{16}
26. Rd1-d2{18} Kg8-h8{22}
27. Rd2-e2{10} Ra8-a5{16}
28. Re7-f7{12} Kh8-g8{6}
29. Re2-e7{10} g7-g5{6}
30. Rf7-g7{14} Kg8-h8{18}
31. Rg7-h7{14} Kh8-g8{18}
32. Re7-g7{12} Kg8-f8{10}
33. Rg7-c7{10} Rc6-c7{6}
34. Rh7-c7{10} Kf8-g8{16}
35. Rc7-b7{10} Ra5-c5{12}
36. Rb7-b6{10} Kg8-h7{22}
37. a3-a4{12} Rc5-c2{20}
38. a4-a5{8} f5-f4{10}
39. g3-f4{10} g5-f4{10}
40. a5-a6{12} Rc2-e2{22}
41. a6-a7{10} Re2-e5{22}
42. Qa7-a8{Q}{12} Re5-e7{22}
43. Kg2-f3{12} Re7-f7{16}
44. Qa8-c6{14} Kh7-g8{10}
45. Qc6-h6{8} Rf7-f5{22}
46. Rb6-b7{10} Rf5-e5{12}
--------------------------------------------------------------------------------
Jester: (White)
beginner: (Black)
Game Played at: http://www.GetClub.com/Chess.html
View Recorded Game: http://www.getclub.com/playgame.php?...313&game=Chess

Bye
Sanny

24. Kg1-g2{10} f7-f5{18} {Is this move Wrong?} Because after that
White took 2 pawns. Is there any way by which Two pawns could be
saved. As after 16 moves White gains 2 Pawns.

Play Chess at: http://www.GetClub.com/Chess.html

On Jan 30, 12:31*pm, Sanny wrote:
With little more improvents GetClub played much better today but
Beginner Level lost to Jester.

There are definite signs of improvement here. But it still has a thing
about the king cowering in a corner to no good end.

Here is the game Below. You can see the moves by Beginner is now much
better. But still it lost 2 Pawns and White was able to Queen with
Extra Pawns.

Was there any way to save those Pawns?

No obvious way to save the pawns as such, but it could have snaffled
some of the opponents pawns to hold on a bit longer. I'll annotate the
defects as I see them but I am not so good at rook & pawn endgame
tricks.

24. Kg1-g2{10} f7-f5{18} {Is this move Wrong?}

It isn't the best move for black. But it isn't losing either. I reckon
Kf8 is better.

26. Kh8 isn't good either - it loses momentum. The king is not really
in danger on g8.

29. Re2-e7{10} g7-g5{6}

This is where it all goes to pot. g5 doesn't do anything useful for
black. This is a critical point in the game.
AFAICT 29. .... Rac5 is the only sensible move here doubling your own
rooks. And might still hold on for a draw
29. .... g6 or 29. .... Kh8 are both better than the move played.

The evalution should give a bonus for having a rook pair (especially
on the 7th rank).

34. Rh7-c7{10} Kf8-g8{16}

Again another futile wasted K move.
Any of Rd5, Re5, Rb5, Rc5 would be OK (still losing but by much less)

40. a5-a6{12} Rc2-e2{22}

This is a disastrous move. The pawn on a6 can queen in 2 moves unless
you can get the rook to it in time.
Only Rc1 or Rc5 are adequate (and material promotion to queen should
be seen at 3-4 ply)

It is a standard trick to put a rook behind your own pawn to force it
along or to attack the opponents pawn from behind if at all possible.
Again the evaluation function could use a tweak to encourage this for
a passed pawn.

24. Kg1-g2{10} f7-f5{18} {Is this move Wrong?} Because after that
White took 2 pawns. Is there any way by which Two pawns could be
saved. As after 16 moves White gains 2 Pawns.

The pawns being taken are a symptom of the program making pointless K
moves and losing a tempo. ISTR that chess engines have to be
artificially discouraged from moving the king unnecessarily by a tweak
in the evaluation function. It is something of a balancing act to get
good behaviour.

BTW Any chance you could put a "+" on the end of moves resulting in a
check?

Regards,
Martin Brown