My chess program's first victory against TSCP

[ZirconiumX]

I'm so proud.

[pgn]
[Event "Computer Chess Game"]
[Site "THUNDERBIRD"]
[Date "2016.06.27"]
[Round "-"]
[White "Dorpsgek alpha"]
[Black "tscp181"]
[Result "1-0"]
[TimeControl "300+8"]
[Annotator "9. +0.39 9... +0.11"]

1. d4 Nf6 2. Nf3 e6 3. c4 d5 4. Nc3 dxc4 5. e4 Bb4 6. Bg5 c5 7. Bxc4 cxd4
8. Nxd4 Bxc3+ 9. bxc3 {+0.39/8 8} Qa5 {+0.11/7 12} 10. Nb5 {+0.26/7 7} Nxe4
{-0.21/7 12} 11. Qd4 {+0.63/7 9} Nxg5 {-0.49/6 12} 12. Qxg7 {+0.94/8 16}
Rf8 {-0.68/7 12} 13. Qxg5 {+2.81/8 16} Nc6 {-1.46/6 12} 14. Rd1 {+2.87/7 5}
Qb6 {-2.64/6 11} 15. Nd6+ {+4.18/7 6} Kd7 {-5.32/5 11} 16. Bb5 {+4.24/7 15}
h6 {-6.43/6 11} 17. Qxh6 {+7.06/7 6} Kc7 {-7.13/6 11} 18. Qxf8 {+8.23/7 4}
Qc5 {-7.62/6 11} 19. Qxf7+ {+7.74/8 8} Kb8 {-7.86/6 11} 20. Qg7
{+7.75/8 15} Ne5 {-7.91/6 11} 21. Nxc8 {+9.39/8 14} Qxc3+ {-10.26/7 11} 22.
Kf1 {+10.69/9 5} Qc7 {-10.31/7 11} 23. Qg8 {+11.48/8 6} Qxc8 {-12.16/7 11}
24. Rd8 {+12.57/9 12} Kc7 {-12.30/8 10} 25. Rxc8+ {+12.45/9 8} Rxc8
{-12.50/7 10} 26. Qxe6 {+12.46/9 6} Nc6 {-12.50/7 10} 27. Bxc6 {+12.31/8 4}
bxc6 {-12.59/7 10} 28. f4 {+12.32/8 4} Rd8 {-12.76/7 10} 29. Ke2
{+13.15/9 23} Rd6 {-13.48/7 10} 30. Qe5 {+13.65/9 9} Kd7 {-13.59/7 10} 31.
Rb1 {+14.18/9 6} Re6 {-12.79/8 10} 32. Rb7+ {+20.07/11 9} Kc8 {-19.32/7 10}
33. Qxe6+ {+19.86/10 8} Kxb7 {-19.32/7 10} 34. f5 {+20.33/10 10} Kb6
{-20.13/8 10} 35. f6 {+21.44/10 13} Kb5 {-21.42/8 10} 36. Qb3+
{+1000.09/9 6} Kc5 {-21.18/6 10} 37. f7 {+1000.07/10 8} Kd4 {-99.94/6 0.3}
38. f8=Q {+1000.05/9 7} a5 {-99.96/4 0.1} 39. Qh8+ {+1000.03/8 5} Ke4
{-99.98/2 0.1} 40. Qf3# {+1000.01/9 16}
{Xboard adjudication: Checkmate} 1-0
[/pgn]

Ironically, this happened after I scrapped most of the search pruning, so I probably had a massive regression there. Still, it's a win.

[Daniel Anulliero]

It's not Zirconium the name of your engine ?
Anyway, congratulations ! Beating TSCP was my main goal too ... during à looong time !
Bye

[flok]

Congratulations!

I hope one day to reach that as well.

Love the name of your program ("dorpsgek alpha", briliant!)

[ZirconiumX]

Congratulations!

I hope one day to reach that as well.

Love the name of your program ("dorpsgek alpha", briliant!)

A 500 game gauntlet against TSCP puts it at about 360 Elo weaker than TSCP, so the "Dorpsgek" name is very accurate.

It's also somewhat buggy, which is where the alpha part of it.

Daniel - no, ZirconiumX is my forum username, so it was referring to me.

[velmarin]

I'm glad to see you here, Matthew.

The youth is always welcome.
The name in Spanish sounds a "little weird"... too many consonants, difficult to pronounce...

[ZirconiumX]

I'm glad to see you here, Matthew.

The youth is always welcome.
The name in Spanish sounds a "little weird"... too many consonants, difficult to pronounce...

The name is Dutch, which explains why it's hard to say.

Literally, its name means "idiota del pueblo" in Spanish.

If it's any consolation, I can't wrap my tongue around the dutch 'g' sound either.

[flok]

Your name doesn't sound very Dutch, how did you came up with this name?

[ZirconiumX]

Your name doesn't sound very Dutch, how did you came up with this name?

I have good friends in Flanders, who suggested it.

[CRoberson]

Yeah, NoonianChess was below TSCP and its clones/derivatives until I implemented something that I was aesthetically opposed to - king safety.
Even the simple king safety algorithm I created off the top of my head was enough to move ahead of the TSCP clan.

All I did was to count the number of squares that were under attack around the king and bonus them. After watching several games, I saw a big flaw - the
queen attacks lots of squares when it gets near the enemy king. The queen would go on a king hunt without support from other pieces. So, I tossed
something together to count the number of pieces as well.

Simple, but effective enough to beat clan TSCP.

[ZirconiumX]

Yeah, NoonianChess was below TSCP and its clones/derivatives until I implemented something that I was aesthetically opposed to - king safety.
Even the simple king safety algorithm I created off the top of my head was enough to move ahead of the TSCP clan.

All I did was to count the number of squares that were under attack around the king and bonus them. After watching several games, I saw a big flaw - the
queen attacks lots of squares when it gets near the enemy king. The queen would go on a king hunt without support from other pieces. So, I tossed
something together to count the number of pieces as well.

Simple, but effective enough to beat clan TSCP.

I should be able to make a fairly advanced king safety rather cheaply, because Dorpsgek always has attack tables to hand, meaning I can do elaborate things like count the attackers and defenders of a piece near the king with masking and a handful of popcounts.

I'll probably start with the CPW example king tropism though.