Junior 2010 vs. Rybka 3, one of the best games I ever saw!

[Frank Quisinsky]

Hi there,

I found the game in my SWCR Ratinglist.
Clearly one of the best games I ever saw.
Junior 2010 gave Rybka 3 mate in 38 moves.

This should be a game for an analyses by a Grandmaster. I will asked here.

Conditions can be found here:
http://www.schach-welt.de

Code: Select all

[Event "Intel - 4600 (Junior 2010)"]
[Site "Trier"]
[Date "2009.12.19"]
[Round "3.9"]
[White "Junior 2010"]
[Black "Rybka 3"]
[Result "1-0"]
[ECO "B16"]
[PlyCount "75"]
[EventDate "2009.??.??"]
[EventCountry "GER"]
[Source "Quisinsky"]

1. e4 c6 2. Nc3 d5 3. Nf3 dxe4 4. Nxe4 Nf6 5. Nxf6+ gxf6 6. d4 Bg4 7. Be2 Qc7
8. h3 Bh5 9. O-O Nd7 10. a4 e6 11. a5 a6 12. c3 O-O-O 13. Nd2 Bg6 14. Nc4 Rg8
15. Bf3 Nc5 16. Nb6+ Kb8 17. dxc5 Rxd1 18. Rxd1 Qe7 19. Rd7 Qe8 20. b4 e5 21.
b5 Bf5 22. bxc6 Bxd7 23. cxd7 Qd8 24. Be3 f5 25. Rb1 Bg7 26. c6 Kc7 27. Na8+
Qxa8 28. Bb6+ Kd6 29. Rd1+ Ke6 30. Bd5+ Ke7 31. Bc5+ Kd8 32. Bxf7 Rf8 33. c7+
Kxc7 34. Bb6+ Kc6 35. Bb3 Kb5 36. Rd5+ Kc6 37. c4 Qd8 38. Ba4# 1-0

[swami]

Very aggressive game from Junior!

Too much costly sacrifices, too much threats, full control of initiative, possession and finally a strong attack.

[tomgdrums]

I am finding Deep Junior 2010 to be an exceptionally fun engine to play against and a really good engine for alternative analysis!!

Which is why I have engines to begin with.

[Frank Quisinsky]

Hello Tom,

yes, seems to be true. I will do it too

Any new knowledge to the Ktulu 9 settings?
I will follow with interest on it.

So far I am looking in all the fast games from the SWCR ratinglist I playing so far.

A lot of great games by Stockfish I can see.
Stockfish and Junior fabricated the most short games. Very nice fast games from Onno and Spark too.

But this game Junior 2010 vs. Rybka 3 is fantastic.

Best
Frank

[jdart]

17. dxc5! was nice, and appears best, according to my analysis so far.

--Jon

[Andre van Ark]

Very nice game, and thx for sharing, Frank

[M ANSARI]

I think the critical point of the game is here

[d]2kr1br1/1pq2p1p/p1p1ppb1/P1n5/2NP4/2P2B1P/1P3PP1/R1BQ1RK1 w - - 0 16

Where Rybka had just played Nc5? Rybka 3 needs depth 17 to avoid it. Obviously in search it had to see that the Queen exchange is possible and that would have taken a few seconds depending on the hardware. To be fair to Rybka given some time it will quickly look for alternative move and sees that Nc5? fails.

Here is a clip analysis of Rybka 3 after Nc5? on fast hardware.

Junior 2010 - Rybka 3, Intel - 4600 (Junior 2010) 2009
2kr1br1/1pq2p1p/p1p1ppb1/P1n5/2NP4/2P2B1P/1P3PP1/R1BQ1RK1 w - - 0 1

Analysis by Rybka 3:

16.Nb6+
+/= (0.54) Depth: 6 00:00:00 0kN
16.Nb6+
+/= (0.54) Depth: 7 00:00:00 0kN
16.Nb6+
+/= (0.54) Depth: 8 00:00:00 0kN
16.Nb6+
+/= (0.54) Depth: 9 00:00:00 0kN
16.Nb6+
+/= (0.54) Depth: 10 00:00:00 0kN
16.Nb6+
+/= (0.54) Depth: 11 00:00:00 0kN
16.Nb6+
+/= (0.54) Depth: 12 00:00:00 0kN
16.Nb6+
+/= (0.54) Depth: 13 00:00:00 0kN
16.Nb6+
+/= (0.54) Depth: 14 00:00:00 0kN
16.Nb6+
+/= (0.54) Depth: 15 00:00:00 0kN
16.Nb6+
+/- (0.74 !) Depth: 16 00:00:00 2kN
16.Nb6+
+/- (0.94 !) Depth: 16 00:00:00 72kN
16.Nb6+
+/- (1.34 !) Depth: 16 00:00:00 165kN
16.Nb6+
+- (2.14 !) Depth: 16 00:00:22 7284kN
16.Nb6+
+- (1.59) Depth: 16 00:00:42 15358kN
16.Nb6+
+- (1.79 !) Depth: 17 00:00:53 19384kN
16.Nb6+
+- (1.99 !) Depth: 17 00:00:59 21115kN
16.Nb6+ Kb8 17.dxc5[] Rxd1 18.Rxd1 Qe7 19.Rd7[] Qe8 20.b4[] e5 21.b5[] Bf5 22.bxc6[] Bxd7 23.Nxd7+[] Kc7 24.Nxf6[] Qe6 25.Nxg8[] bxc6 26.Bg5[] f5 27.Nf6 h6 28.Bh4 Qc4 29.Ne8+ Kc8 30.Bf6 Qxc3 31.Rd1
+- (2.17) Depth: 17 00:01:45 37499kN

(, Microsoft 04.01.2010)

[Matthias Gemuh]

Here is a clip analysis of Rybka 3 after Nc5? on fast hardware.

Junior 2010 - Rybka 3, Intel - 4600 (Junior 2010) 2009
2kr1br1/1pq2p1p/p1p1ppb1/P1n5/2NP4/2P2B1P/1P3PP1/R1BQ1RK1 w - - 0 1

Analysis by Rybka 3:

16.Nb6+
+- (1.99 !) Depth: 17 00:00:59 21115kN
16.Nb6+ Kb8 17.dxc5[] Rxd1 18.Rxd1 Qe7 19.Rd7[] Qe8 20.b4[] e5 21.b5[] Bf5 22.bxc6[] Bxd7 23.Nxd7+[] Kc7 24.Nxf6[] Qe6 25.Nxg8[] bxc6 26.Bg5[] f5 27.Nf6 h6 28.Bh4 Qc4 29.Ne8+ Kc8 30.Bf6 Qxc3 31.Rd1
+- (2.17) Depth: 17 00:01:45 37499kN

(, Microsoft 04.01.2010)

More interesting is the analysis before Nc5.

Matthias.

[M ANSARI]

Here is a 5min R3 clip analysis before Nc5? Unfortunately I don't have Junior 2010. It takes up to depth 17 to avoid Nc5? I guess this is just one of these positions that can quickly blow up with one wrong move ... similar to many poison pawn positions.

Junior 2010 - Rybka 3, Intel - 4600 (Junior 2010) 2009
[d]2kr1br1/1pqn1p1p/p1p1ppb1/P7/2NP4/2P2B1P/1P3PP1/R1BQ1RK1 b - - 0 1

Analysis by Rybka 3:

15...Bf5 16.Kh1 Nc5
= (0.20) Depth: 6 00:00:00 10kN
15...Nc5 16.b4 Nd3
= (0.07) Depth: 6 00:00:00 13kN
15...Nc5 16.b4 Nd3 17.Kh1
= (0.07) Depth: 7 00:00:00 18kN
15...Nc5 16.b4 Nd3 17.Qe2 Kb8
= (0.12) Depth: 8 00:00:00 28kN
15...Nc5 16.b4 Nd3 17.Be3 Kb8 18.Qe2 Bf5
= (0.17) Depth: 9 00:00:00 83kN
15...Nc5 16.b4 Nd3 17.Be3 Kb8 18.Qe2 Bf5 19.g3
= (0.16) Depth: 10 00:00:00 114kN
15...Nc5 16.b4 Nd3 17.Be3 Kb8 18.Qe2 Bf5 19.g3
= (0.16) Depth: 11 00:00:00 145kN
15...Nc5 16.b4 Nd3 17.Be3 Kb8 18.Qe2 Be7 19.Kh1 Nf4
= (0.23) Depth: 12 00:00:00 253kN
15...Nc5 16.b4 Nd3 17.Be3 Kb8 18.Qe2 Nf4 19.Qd2 Nd5 20.Bxd5 Rxd5 21.Kh1
= (0.19) Depth: 13 00:00:01 567kN
15...Nc5 16.b4 Nd3 17.Be3 Kb8 18.Qe2 Nf4 19.Qd2 Bh6 20.Kh1 e5 21.Nb6
= (0.22) Depth: 14 00:00:02 838kN
15...Nc5 16.b4 Nd3 17.Be3 Kb8 18.Kh1 e5 19.Qd2 Qd7 20.Nb6 Qe6 21.b5 cxb5
= (0.21) Depth: 15 00:00:04 1560kN
15...Nc5 16.b4 Nd3 17.Be3 Kb8 18.Kh1 e5 19.Qd2 Qd7 20.Nb6 Qe6 21.b5 cxb5
= (0.21) Depth: 16 00:00:06 2495kN
15...Bf5 16.Ne3 Bg6 17.Qe2 Kb8 18.Rd1 f5 19.Nc4 f4 20.b3 Bf5 21.Re1 Bxh3 22.Qe4
+/= (0.37) Depth: 17 00:00:55 21036kN
15...Bf5 16.Ne3 Bg6 17.Qe2 Kb8 18.Rd1 f5 19.Nc4 f4 20.b3 Bf5 21.Kh1 Bh6 22.Re1 Nf6 23.Qe5 Nd5
+/= (0.40) Depth: 18 00:01:20 30265kN
15...Bf5 16.Ne3 Bg6 17.Qe2 Kb8 18.Rd1 f5 19.Nc4 f4 20.b3 Bf5 21.Kh1 Bg7 22.Re1 Bh6 23.Ba3
+/= (0.41) Depth: 19 00:02:02 44633kN
15...Bf5 16.Ne3 Bg6 17.Qe2 Kb8 18.Rd1 f5 19.Nc4 f4 20.b3 Bf5 21.Kh1 Bg7 22.Re1 Bh6 23.Rg1
+/= (0.41) Depth: 20 00:04:04 89842kN
15...Bf5 16.Ne3 Bg6 17.Qe2 Kb8 18.Rd1 f5 19.Nc4 f4 20.b3 Bf5 21.Kh1 Bg7 22.Re1 Bh6 23.Rg1
+/= (0.41) Depth: 21 00:04:57 109mN

(, Microsoft 04.01.2010)

[Rolf]

Last month here in CCC I read this header:

Junior 2010 ... MISS THE POINT of chess ... NOTE = 6 !!
Author: Frank Quisinsky (note 6 is meant as the worst possible in German school system)
-------------------------------------------------------

In general it is unsound to pick up a singular position to then seemingly proving that some program is eithe God or Nothing. The whole approach is wrong. Among thousands of games one will always find this extreme or that exceptional item. For a tester it's very bad if he is basing his verdicts on such details. More it is absolutely wrong testing. Test hypotheses must have a prior claim and then the tests say this or that. And then the hypthesis is either true or not.

But this sort of aftermath peanuts seeking to once making this sensational headline then weeks later that sensation is absolute crap.

Just for the young readers:

If you seek a novelty or news about something in computerchess entities and the results are basing on games, you simply think about a theory and then construct a test design that could be appropriate. At best with _many_ games. Most of the time you wont find anything spectacular at all.

But if you seek sensations and direct knockouts, please go into boxing where you have fairy tales and fraud. Computerchess however is a bit like science and you never can force a sensational message just to be in the news. The two totally opposing and each very wrong headlines in our case here is doing no good for the status of an alleged tester. Sorry about that.