I have seen an Evans Gambit, and a Two Knights (the one where White plays Ng5 and Black plays ...b5), but those are quite standard openings...

Here's a Frankenstein-Dracula (apparently called the Blanel Gambit in more refined circles).

AlphaZero vs Stockfish, 2018 (Draw, 135 moves)

Blumenfeld:

Stockfish vs AlphaZero, 2018 (Draw, 194 moves)

King's Bishop's Gambit:

Stockfish vs AlphaZero, 2018 (Drawn in a brisk 56 moves)

Finally, a Fried Liver:

Stockfish vs AlphaZero, 2018 (Draw, 198 moves)

Alphazero little open source sister leela aka lco breaksthrough at TCEC. Qualifies for TCEC superfinal against stockfish ahead of Komodo and Houdini 6, by winning division 3, division 2 and finishing 2nd to stockfish in div premier.

https://tcec.chessdom.com/live.html

It is also currently at 2nd place in cccc

If all the games were available, I'd divide the match into a series of matches, one for each opening (which I think was 20 games/color(?)). In other words, each game for an opening would have its own Event tag. That would make getting the actual stats easy.

(Aside - does anybody know a table of the numbers, as compared to the colored bars in the supplemental. I'd like to know the numbers, instead of eyeballing 'em).

I'm also thinking of removing the misleading "book" comments, and instead use the FEN tag for the PGN, putting the opening moves into a comment.

That would have the advantage of clarity (ie. what moves the engines actually made). It would also eliminate having to play over the same opening moves in the 40 (?) match games.

Hopefully this time round we'll be able to get all the games (now that DeepMind has gotten its paper published).

Anyone else have thoughts on this, particularly on using FEN to specific the starting position?

A PGN or even a list of FEN's would be nice.

And can someone tell me when the openings were used? Was it in the superfinal stage only?

1) ECO code?

2) By S4 diagram?

http://www.mediafire.com/file/i311b...

Has < [ 2006, 2008, 2012, 2012 (Topical), 2014, 2016 ]> Noomen Testsuites.

(Following the "canonical" link from chessprogramming/etc yields a stale mediafire link:

http://blogchess2016.blogspot.de/20...

from here:

https://www.chessprogramming.org/TC...)

I like this one opening example PGN, which provides a fuller description than Noomen's.

http://www.chess2u.com/t10903-tcec-...

I'm trying to fill out the PGN info similarly.

It's not a question of being skeptical, it's just that whenever one engine is running on much more powerful <hardware> than the other engine then the results are inconclusive as to which engine is superior. Remember that we are trying to evaluate the relative superiority of two <software> engines, not two combinations of <hardware> + <software>. At least that's what I thought.

Now, if the time control for the engine running on more powerful hardware was reduced to try to compensate for that engine's superior hardware capability over the other engine's hardware, then the contest would be more even. So, as a first order approximation if one engine's hardware was capable of executing 100X more instructions than the other engine's hardware, then a time control reduction of 1/100 for that engine would allow both engines to execute about the same number of instructions in the same amount of time.

And that's what Figure 2 in "A general reinforcement learning algorithm that masters chess, shogi and Go through self-play" (https://deepmind.com/documents/260/...) shows. When AlphaZero was given the same amount of time to play as Stockfish, AlphaZero won convincingly. When AlphaZero was given 1/30 of the time to play as Stockfish, Stockfish outscored AlphaZero with both the White and Black pieces. And when AlphaZero was given only 1/100 of the time to play as Stockfish, Stockfish outperformed AlphaZero as convincingly as AlphaZero outperformed Stockfish when both engines were given the same amount of time.

So, given that AlphaZero was running on hardware that was able to execute approximately more than 100 times the number of instructions in a given amount of time as the hardware that Stockfish was running on (see AlphaZero (Computer) (kibitz #348) above), what word would you use to describe the relative <software> performance of the two engines as a result of their 2 matches? I think that calling it "inconclusive" is being very generous to AlphaZero.

Care to discuss 'evolution'???>

@worstplayerever, If you *really* think Google broadcasting live matches in 2017 is as hard as filming and televising the apollo projects in 1969-72 then I am not up to the challenge of elaborating and explaining the differences between the two events. But I can give you a small hint: Chess.com, which is remarkably a smaller company than google, regularly broadcasts two engines playing each other live.

But anyway, there may be some possibilities like a human co-operating with Houdini or other engine (or maybe Stockfish itself) and trying several strategies. I do not know but there are good reasons to doubt the event:

The authors of LC Zero have much fewer ressources and compete on equal terms with a newer Stockfish than the one AZ played against...

<For google it is so easy to broadcast the matches but they did not. Why? What reason could have made them to play the matches behind the closed doors without prior announcement, without post-game announcement. Nothing, nothing. They just popped up and 'revealed' they have conquered the chess world.>