"Daeth Trap" Engine Tournament

This thread is not to discuss virtuous or dubious nature of an early unprovoked h4 by White. Its only purpose is to let a bunch of top chess engines take a chew on it for a bit and see what comes out.

Here's what I'm doing. I have just started up a multi-engine tournament running inside Fritz 13. It will be a double round-robin tournament with five entrants, for a total of 20 games, game in 5|2. I've decided against a longer time control for now because I have no idea how long this is going to tie up my computer. I will post the score card when it is done, whatever it may be.

Entrants: Fritz 13, Rybka 2.3.2a, Critter 1.6a, Crafty 23.01, Spark 1.0

Now starts the pool. Out of a possible 20 points, how many do you predict White will win? Remember, we are scoring for White, not any particular engine.

I am marking myself down for 7.5 out of 20.

What about "Spike"?  How com nobody gives any props to the chess engine "The Spike" yo?

Maybe he'll make the cut next time. Sorry, but this isn't a test of engines, it's a test of an opening!

Miami is nice, so I'll say it twice.  Miami is nice...Miami is nice...Miami is nice...

Good to know if I end up bothering with a standard time tournament. I think it's still of interest, though. Where the weaker engine prevails will tell us something.

Not that any of this is definitive, of course. Well, unless one side crushes the other time after time. Longer games will have more value, but this is my first engine tournament, and I'm just trying to make sure it all works first.

Let's try this again:

http://youtu.be/MA-BYgduGzY?t=31s

Ok, I promised results, and here are some preliminaries. 

In the above mentioned tournament, White scored 52% (10.5 out of 20) from the position given. This shows that 5.Be2 is playable for White. That suggests that if (and only if) the first four moves can't be busted, then the opening might be playable. So I moved on to earlier moves.

Since many of the engines actually preferred 3... c5, I decided to take a look at that. It's all fine and dandy if you can make it to move 5, but if you're busted on move 3, then the opening as a whole is a dud. 

So I fired up another tournament, this one a single Round Robin at 25 minutes/game and let it run overnight. White scored 1/1/8, for 50%. Not exactly a challenging opening for Black, with an easy draw in almost every game. In fact, in several of the draws, White barely held on by the skin of his teeth. One was an 80-mover where White managed to draw by trading down to K+B+P vs. K where the Bishop was on the wrong color to help Black's pawn promote. White was lucky to have Fritz for this side because the engine playing Black was still evaluating the position as -3.5 up to and even after the exchanges. With table bases, this likely would have ended as a win for Black.

Conclusion, after 1. d4 d5 2. h4 Nf6 3. Bf4, White is playing for a draw that he will have to work hard for.

Now we trace back to the original problem with the opening , 2. h4 -- the point of controversy. If this move is busted, the whole line goes down in flames.

So this morning I fired up another tournament. This time, I'm going for a 6 engine double RR blitz format. I have left it running at home and will get the final results this evening. When I peeked at the progress before leaving the house, White had 5.5 points out of a possible 13, putting him at 42%. Still there are 17 games left, so maybe he'll pull out a miracle? I'm guessing no.

Actual games to come, if requested.

2. h4 BUSTED

Short version:

2... c5 is better for Black. A lot better.

Slightly longer version:

The last above mentioned tournament turned out to score evenly for White and Black - 50%. But I noticed something interesting. For those games where the computer chose 2... c5 the wins for Black were crushing.

So I did what any good researcher would do... start another tournament! This time I went for a 7 engine, double RR for a total of 42 games played at 5|2. And what did I find?

Black scores 27.5 out of a possible 42 points -- better than 65%. 

That's right, black scores better than 65% with 2... c5. That busts the entire opening and all following variations, since move two is simply wrong.

Running another batch at g/3 and will post the final scores and report either tonight or tomorrow.

By the way, the most surprising find was just how strong Critter is. 

Thanks, pfren. I can't believe we're actually analyzing this, but it's something to do, I guess.

I'll post continuations later. Most of the games had Black smashing the queenside wide open. The h pawn did cause some minor trouble here and there, but wasn't typically a big factor. 

I didn't look at the game, but gotta love the venue!

Final Report on Opening

After running the two double round robins, I have tabulated the results. One of the engines (Crafty) scored so poorly that I removed it from the results. Crafty did not win a single game in either tournament, drew four or five, and lost the rest. I didn't realize it was such a poor engine, but the rest seem to be ok. What we have left is 60 games, a reasonable sampling.

Final engine score for those interested:

Critter 1.6: 17; Fritz 13: 9; Fritz 11SE: 9; Rybka 2.3.2: 9; Spike 1.4: 8.5; Spark 1.0: 7.5

And now for White vs. Black:

White: 21.5 (8 wins, 27 draws, 25 losses) - 35.8%

Black: 38.5 (25 wins, 27 draws, 8 losses) - 64.2%

It's not looking good for 2. h4

Now we get to White's actual responses to 2... c5, of which there were 3 played. Here the starts of the lines and the score for White in each:

3. e3 (43.5%, 23 games)

...to which Black responds Nc6 (50%, 13 games); Nf6 (16.7%, 6 games); e6 (66.7%, 3 games); cxd4 (50%, 1 game)

3. Nf3 (30%, 20 games)

... to which Black responds cxd4 (37.5%, 8 games); e6 (37.5%, 4 games); Nc6 (12.5%, 4 games); Nf6 (25%, 4 games)

3. dxc5 (32.4%, 17 games)

... to which Black responds e6 (36.4%, 11 games); Nf6 (20%, 5 games); Nc6 (50%, 1 game)

There is barely a hint of a hope for White here.

Please let me know if you want to see games for any particular line and I will post them.

What if we make black want to play the KID with 1... Nf6?  So 1. d4 Nf6 2. h4!?/?!/?/??/!/!! (I'm not sure what symbol to use...).

1.d4 Nf6 2. h4 may be a completely different can of worms, I can't say. It may be awhile before I feel like expending that kind of effort again, so anyone who wants, please feel free to do some testing.

Daeth hasn't said anything on this forum.  Does he know about it?

I posted a link to it on his thread 2 or 3 times.

So probably not.

Joseo, it's d4, not e4.

JoseO is right, though.  And you did say 1. e4 :-p  Obviously a typo since you're refering to my line of 1. d4 Nf6 2. h4.

Typo expunged. Thank you.

Now no one knows where to reference, making JoseO look like an idiot!  YOU RUINED IT!!!!  :-p