Game 4 of 10

Another win for TheKing!



According to DH15.1 (Deep HIARCS 15.1) CG(ChessGenius) made 1 blunder and 8 inaccuracies whilst TheKing made 1 mistake and 2 inaccuracies. These figures explain the result but where did it go wrong for CG? According to DH15.1 all was level after Black’s move 39...Rg5:

Black has a pawn poised for promotion on b2 tying up White’s Rooks but White’s pieces are also in strong attacking positions. Here, White moved 40.Nd6?? DH15.1 after about a minute on my desktop looked 24 ply ahead and gave an eval of -6.26. One better continuation would have been:
40.e6 Bxe6 41.Nd6 Kh8 42.Nxf5 Bxf5 43.Bxf5 Rc1 44.Qe8+ Rg8 45.Qe3 Rg5
Resulting in the following position with White to move and the position level:


Chess Genius, despite examining over a million positions per second, and often 2 or 3 ply extra ahead than TheKing’s displayed values, does not seem to comprehend the positions as well as TheKing.

TheKing is not able to make the same moves as DH15.1 but does take advantage and from here on dominates the rest of the game.

Current Score 3 - 1 to TheKing

On to the next…

Game 5 of 10

Another win for TheKing!

They played the Four Knights, Belgrade Gambit and it looks like TheKing’s Master book is better than the one in CG. I’ve noticed in this match that CG is out of book far sooner than TheKing, although as we know book opening size is not always indicative of playing strength, but in these games it has certainly helped with TheKing’s clock!



The first position where CG made a poor move is in the opening. After making a novelty with 10… Nc6 (theoretical move is 10…d5!) play followed 11.Qh4 h6? (See diagram)

This is not good as it weakens the King’s line of defense. Better would have been to block off White’s Bishop on c4 with 11…d5

Maybe if it had looked far enough ahead TheKing may have seen the following attacking sequence: 12.Bxh6 d5 13.Qg5 threatening mate on the next move, 13…Ne8 the only sensible option, 14.Bxd5 gaining a pawn, 14…Qf6 15.Bxc6 bxc6 16.Qxf6 Nxf6 17.Bg5 and White is in a good position.
However, this is academic because it made the move 12.Nd4.

An interesting position occurred after TheKing's move 22.Bb1 giving this position:

Here, CG replied 22…a4?! To which TheKing moved 23.b4 and Black’s Queen is now almost hemmed in with only one slow escape route! This is something I noticed that TheKing was able to often achieve against CG on the Millennium unit, so an indication perhaps that CG for Android is not so different which I suspected.

Play continued with TheKing outplaying CG. Here is another example of a bad choice of moves for CG:

In this position CG moved 36…Qxb4?! To which White replied 37.cxb4 Now White has a passed pawn. After this, TheKing’s play was strong enough to sail home to victory. Not always perfect, but of course it’s not as powerful as DH15 on my PC!

In this game TheKing made 2 mistakes and 3 inaccuracies whilst CG made 2 mistakes and 12 inaccuracies.

So far, TheKing has 3 wins and 2 draws, still time for CG to win.

Game 6 of 10

In this game the Old Benoni, Franco-Benoni Defence was played. TheKing wins another game, it’s on a winning streak making this 4 wins and 2 draws.

In the following position at move 23 both sides have equal value in material, Black has a Bishop and Knight whilst White has two Knights. Black (TheKing) may have a slight edge according to DH15:

Maybe there is some chance of White starting a King-side attack if he positions pieces on this side of the board. However, CG seems to want to maintain a presence on the Queen-side. The problem is it moved 24.Qc5? This leads to material loss which TheKing could see but alas CG could not. The game continued 24…Qf6 25.N2f3 Rc8 26.Qxd5 Be6 27.Qe4 Rd8 28.c3 Nxc3 29.Qe3 Rxd4 30.Nxd4 Qxd4 31.Qxd4 Ne2+ 32.Kf1 Nxd4 and now Black is up a pawn and has a very good winning advantage, which it achieves.

It seems that TheKing is able to ‘see’ more than CG despite the much lower speed of the machine, impressive!

CG made 1 mistake and 8 inaccuracies whilst TheKing made 3 inaccuracies and no mistakes.

Current score is 5 – 1 to TheKing. So now CG cannot win the match, at best make it a draw.

Game 7 of 10

I think this game is the most impressive and is also the shortest game in the match at 53 moves.
It’s another win for TheKing and it’s pretty much one-sided from the start!


They played the Caro-Kann, Advance variation with TheKing(White) this time making a novelty move in the opening with 10.a4 (the theoretical move is 10.Ng3!):

The game continued 10…Nb4 11.a5 Nc8?! Run away and hide! Surely, 11…N6d5 would be positionally stronger?
TheKing is not without making inferior moves. Take the following position after move 22:

Here, TheKing with the white pieces is a pawn up and has two Bishops compared to Black’s Bishop and Knight. However, Black’s Knight on c8 is not doing much, and also Black’s Rooks are tied up. Here, TheKing played 23.Bd2?! But 23.d5 would have been much better and more in the style of TheKing’s dynamic play I would have thought. I guess TheKing was planning an attack on the Knight on b4.

Here is the position after TheKing’s 33rd move:

Black is up by a pawn but White’s position is much stronger! CG’s King on b8 doesn’t look safe but to make things worse it moved 33… Kc7?! White only needs to advance the Pawn on d4 and Black’s King is exposed! Well, TheKing did not make this move straight away, but it was only a matter of time before it did and CG suffered as a consequence. Here is the final position:

I did not expect to see a win like this, Black’s Rooks still on their original squares and Black’s King in the center checkmated!

In this game TheKing made 4 inaccuracies whilst CG made 2 mistakes and 10 inaccuracies.

Very good play by TheKing, very poor for CG in my opinion. It does look like CG on Android is still not scaling up well on faster hardware, and possibly not benefiting much or any from being a 64-bit program? Am I being too harsh? After all, running on Pentium PC it beat Kasparov with normal time controls. It’s certainly way stronger than me, so I’m not saying it isn’t good, it’s just that TheKing is so impressive.

Score now 6 – 1 to TheKing. This is looking very one sided!

Game 8 of 10

In the following game the advantage swings from one side to the other for about half the game then one side takes over. You can probably guess which.


They played the Queen’s Gambit Declined with TheKing deviating at move 5 from the following position:

Here, the preferred move is 5…b5! But TheKing played 5…Nf6!? In keeping with it’s style, speculative and exciting.

There were two blunders in this game one from each side. The first blunder was made by TheKing in the following position:

CG had just made the questionable move 49.Kg3?! Possibly after the pawn on h5. Better would have been 49.Nc1 preventing the advance of Black’s passed pawn on b4. But to benefit from this move does require looking further ahead then CG could probably manage. However, TheKing responded with 49…e4?? Giving up a pawn bringing the position level again.

The second blunder was by CG in the following position:

Black had just played 62…Kf5 and CG responded with 63.Nc5?? Probably to support advance if the e5 pawn. The better move is 63.Nc1 to eventually reach and protect the pawn on g2, but it’s academic with Black in such a strong position. From the blunder onward TheKing pretty much played the best moves to victory.

So not perfect play with CG making 1 blunder and 12 inaccuracies and TheKing making 1 blunder and 9 inaccuracies. But a win is a win with the score now 7 – 1 in favour of TheKing.

Game 9 of 10

Another win for TheKing! After the first 25 moves, TheKing’s advantage started growing rapidly and CG didn’t stand a chance!

They played the Centre game with TheKing playing the white pieces.


There was a time in this game where CG had a better position. For example, in the following position it is Black to move:

CG is playing with the black pieces and its Queen and Bishops are in aggressive positions. White’s King is stuck in the middle of rank one and Black only needs to castle queen-side to bring a Rook into action pinning White’s Knight on d2. However, rather than a Queen-side castle, CG castles King-side 16…O-O?? The first and only blunder of the game.

In total White(TheKing) made 1 mistake and 4 inaccuracies and Black 1 blunder, 1 mistake and 9 inaccuracies.

Score now 8 – 1 in favour of TheKing.

Game 10 of 10

Here is the final game in this very one-sided match. I double checked the settings to be sure I’d not put CG under any handicap.

In this game they played the Sicilian, and another win for TheKing!

Up until move 26 CG playing with the white pieces had a very slight advantage but this soon disappeared.

In the following position TheKing had just moved 27…b5 and all is even:


Here CG replied 28.Qh3?! Is he trying to tempt Black’s Knight to move to g5? More likely to pin Black’s Rook on e6. The better move here would have been 28.a4 to block the advance of Black’s pawn. In response, Black does indeed move his Pawn here 28…a4 forcing pawn exchanges and weakening White’s Queen side defenses.

From move 35 onward TheKing sailed home to victory making the final score an impressive 9 – 1.

We know TheKing is good but with the hardware being heavily in favour of CG, even more so. I’m just wondering if Chess Genius for Android has lost something from earlier versions running on dedicated machines, or even from the version that beat Kasparov? Maybe the time control was not so good for CG? With this result that puts Chess Genius around 375 ELO lower than TheKing at this time control!

I have Shredder Chess for Android, maybe I should try a match between that and TheKing? I suspect in this case Shredder would win.

Hope you enjoyed that match even though it was one-sided!

Ray

Millenium King is a very strong Chess programm, but I think against Shredder there is no chanche….

I agree, Shredder is a very good engine. Would be nice to have that in the Millennium Element.

Wow Ray, that' fast!
You ask for predictions but less than one hour later, the outcome of 8/10 games is already revealed (and now 10/10!!)
But an interesting comparison. I would have guessed a more balanced result, as I have Chess Genius Android in my private list with 2490 CElo; and TKP with 2529, so about "only" 40 points above.
But in addition, my Android is an old Galaxy Tab 2, featuring 2 ARM Cortex A9 cores@1.01Ghz.
I can confirm Chess Genius only uses one core (so the performance gap is not so much to your Tab A8, apart from twice the Mhz) and I do agree it does not scale that well with more speed. Your results confirm!
Thanks for the games
Eric