TCEC Superfinal Update: Komodo Leads 20-18; 62 Games Remain

There have been lots of draws so far in the Season 8 Superfinal of the Top Engines Chess Competition (TCEC), but Komodo has managed to win three games while Stockfish has won but one.

The last game, game 38, was won by Komodo, and it was an impressive victory. That said, it also revealed one of the things engines still struggle with this, or - perhaps more precisely put - still cannot do: recognize in some conceptual way that a position is drawn. Failing to recognize a fortress as a fortress is the most familiar manifestation of this shortcoming, but not the only one. Pull up the game from the file structure (Season 8-Superfinal-game 38) on the archive page, and look at the position before White's 87th move. The plan with 87.Rf3 followed by 89.h4 is horrible, and practically speaking, it would be horrible even if it somehow turned out to be a draw.

Before striking on this "opportunity", Stockfish (and Komodo too) seemed to focus on lines where White kept his rook active on the queenside, being sure to be able to defend the d-pawn. In those scenarios Black would almost surely have to play ...R~3 followed by ...Nxh3(+) Bxh3 Rxh3 at some point, and the resulting endgame would be drawn. (If I'm not mistaken, it wouldn't even be a difficult draw.) A human seeing this would just stick to the program, steadfastly ignoring rabbit trails that might in some utterly meaningless sense be - or in this case, merely seem, briefly - one tenth of a pawn better. And the human would be right: there isn't anything to calculate or figure out; the work has been done and the half point is there for the taking.

This is not some sort of quirk that only pertains to Stockfish; Komodo lost to Jonny in the so-called world computer championship in the same way: it had a positional draw but got attracted to some utterly non-simple option that was microscopically better in its evaluation. The refutation took some time (and was beautiful), but it was there and Jonny went on to win. The important thing to realize is that this wasn't "bad luck". Komodo, like Stockfish in the ongoing TCEC, went from a simple draw - a draw requiring only maintenance - to one (not really, but if it had worked) where the game remained a game and thus more work would be needed to save the game.

I say this not to criticize any programmers. This sort of problem has been around forever, and perhaps it's simply impossible or too costly (from an efficiency standpoint) to fix, at least at this point in time. (Maybe those of you more familiar with the nuts and bolts of programming chess-playing programs can weigh in on how, whether and when such problems can be fixed.) It's interesting that while so many of the things that chess computers supposedly can't do are now routine, this problem continues without any signs of abating.

TCEC Superfinal Starts Friday

I had initially thought the TCEC Superfinal was to start today, but in fact it starts on Friday at 1 p.m. ET (= 7 p.m. CET). Here (from the event website) are the rules for the final stage, which will be between Komodo and Stockfish:

Superfinal
This match is played with 50 different openings so that each engine plays both black and white of the same position. The match will be presented with opening 1 used in games 1 and 2, then opening 2 used in games 3 and 4 etc. If the match is theoretically won for one side before game 100, the match will still continue until all 100 games have been played. In the case of a drawn match there will be a rapid match of 16 games with a time control of 25' + 10" with random openings selected from earlier in the same Season. In case it is still tied there will be a Blitz match of 8 games with a time control of 3' + 2". When the Superfinal is over, the current Season ends.

Komodo and Stockfish Qualify For the TCEC Superfinal

For the fifth straight season (sort of - see below), the TCEC (formerly the "Thoresen Chess Engines Competition", now renamed the "Top Chess Engines Competition") will have a Komodo-Stockfish final. Komodo won seasons five and seven, while Stockfish won season six and a special sub-season six for Fischer-random chess. Who will the superfinal for season 8? The monster match will begin November 3, here.

A Ruling on the Stockfish Time Forfeitures

I mentioned a few days ago that Stockfish had lost a couple of games on time in the third stage of the ongoing 8th season of the Thoresen Chess Engines Competition (TCEC). The forfeitures clearly represented a bug and not some sort of fundamental problem with the program, and there was nothing about the positions where it forfeited that were problematic - Stockfish wasn't worse in either case. On the other hand, the rules are the rules, and in past TCECs they haven't allowed programmers to change the programs in the middle of a stage to correct bugs. So what was to be done?

As it turns out, a smart and equitable solution was found. Good for everyone involved!

An Impressive Stockfish Game

The Stockfish chess engine won the last season of the TCEC, but hasn't been dominant thus far in season 7. This, however, is a very impressive game, with a long-term rook sacrifice and some other spectacular moves like 37.Kh2 and 41.Rf2. Although the win was Stockfish's, I think the triumph is more generally one for computer chess. I ran the game to see if Komodo and Houdini would find and favorably evaluate the rook sac, and they did, and then I went for a real test: Fritz 11 running on one core. Even it was fine with giving up the rook, and while it wasn't as excited about the sac as the stronger engines were - running on more cores - it certainly evaluated White's compensation as at least sufficient.

HT: Jeffrey Hall

Which Engine Should You Choose?

As I recall, I linked to an article on this topic some months or maybe around a year ago, and now with a new edition of Komodo on the market a new article has been written. The author addresses the big three engines: Stockfish 5, Komodo 8, and the aging but still worthwhile Houdini 4. His conclusion, in a nutshell, is that if one is engaging in serious, deep analysis and not just basic tactical checking then one should of course download Stockfish, as it's free, and if one must choose between Komodo 8 and Houdini 4 the former is the better choice due its positional strength.

Stockfish Beats Nakamura

And in other news: ten is greater than five and Usain Bolt is faster than I am. Still, the story is more interesting and the match was closer than one might expect.

In fact, the match between the latest version of Stockfish and Hikaru Nakamura (currently #7 in the world; more precisely the #7 human chess player in the world*) was quite close, the 3-1 score notwithstanding. It wasn't an even fight, as Nakamura had help in the first two games while Stockfish had handicaps in all four, but it was a close battle all the same.

In the first two games Nakamura was helped by an older version of Rybka (rated approximately 200 points lower than the version of Stockfish he was facing), while Stockfish wasn't given access to either an opening book or tablebases.  Nakamura drew game one with White, and in game two it seemed that he was headed for a draw by the 50-move rule before playing 83...Bh6. Stockfish ground that one out in 147 moves.

In the next two games Nakamura was on his own - no Rybka - but was given White and an extra pawn. In the first game Stockfish played without an h-pawn and the game was drawn, and in the second game the computer started without its b-pawn. Nakamura wasn't in any danger when they reached the endgame, but he'd need to win to tie the match, and once he opened the board he was gradually outplayed, losing in 97 moves.

Overall, Nakamura acquitted himself well. It's hard to remain vigilant and not miss anything when a game lasts 97 or especially 147 moves, and all four games were on the same day -  there were more 10 hours of play in all. So while the computer played with a handicap, the human did too: computers don't suffer from fatigue (or at least not in the relevant sense)!

* Then again, as computers don't play chess (in part because they don't exist**, in part because there's nobody home "upstairs"), we can dispense with the qualifying adjective "human".

** In saying that they don't exist, I don't mean that they are illusions but that they aren't unities in their own right. Think of a cup sitting on a table. Is there a further entity we could call a tablecup, comprised of the table and the cup put together? Most people would say no, and I agree. The table and cup are not subsumed into a further whole. By contrast, an atom of hydrogen or a molecule of water is a genuine whole in its own right. Its parts are genuinely subsumed into the whole and are defined as parts of that whole. Those parts do not act independently, but in a behavior that's determined by their function within the hydrogen atom or water molecule.

So is a computer more like a tablecup or a water molecule? Both, really - but it depends on one's perspective. As a purely physical object, it's like the tablecup. Its parts interact with each other, but there is no intrinsic principle of unity to the computer's constituent parts. There is a unity to the computer, however, but it comes from us, from our purposes. What the computer does it does only if there are intelligent outside interpreters to understand its results. Otherwise, it is just an unnatural collection of heterogenous parts. It is not a thing in its own right, and thus doesn't exist per se.

(HT: Jason Childress, by way of Allen Becker.)

Stockfish Wins TCEC Season 6

This should have been reported a few days ago, but I was waiting (in vain!) for Squarespace to get back in touch with me about the comments problem. Time to stop waiting and start reporting!

The sixth season of the TCEC (Thorsen Chess Engines Competition) has come to an end, and in the final match Stockfish (using a build published May 17, 2014) rather handily defeated Komodo 7x by a score of 35.5-28.5, going +13 -6 =45. Of Komodo's six wins, only two weren't matched by Stockfish winning the reversed color game (openings were repeated so each engine would get a shot with White), while Stockfish won nine such victories.

It's good news for chess fans that Stockfish is open source and free to download, but one also hopes that its dominant performance doesn't chase out the for-profit Houdini and Komodo engines. With no competitors the Stockfish programmers may lose their motivation as well, and the project might stagnate.

TCEC Season 6: Another Stockfish-Komodo Final

Last year Komodo beat Stockfish in the TCEC final, but so far this year Stockfish has been the engine to beat. In the semi-final stage Stockfish won by a decent margin, with Komodo coming in clear second. Critter was the whipping boy, finishing with only 14 points out of 48, but the surprise was Houdini's distant third place, struggling to reach 50%.

The final will be a 64-game monster; for now Stockfish has an early 2-1 lead, having won game 1 and drawn the next two. In the previous stage they went even against each other and had similar plus scores against Houdini; the difference maker was Stockfish's brutalization of Critter: 10-0 with six draws.

You can watch the ever-live action here, and in a week or two we'll know which 3100+ rated beast will be the champion (until the next build, when the previous results may be meaningless).