Tuesday, May 26, 2015

A new (computer) chess champion is crowned

A new (computer) chess champion is crowned, and the continued demise of human Grandmasters

Garry Kasparov vs. IBM's Deep Blue in the 1997 rematch

Share This Article

952923
It’s almost 18 years since IBM’s Deep Blue famously beat Garry Kasparov at chess, becoming the first computer to defeat a human world champion. Since then, as you can probably imagine, computers have firmly cemented their lead over puny, fallible meatbags — Garry Kasparov is still considered by many to be the greatest chess player ever, while computers are only getting more and more powerful. Today, following the completion of TCEC Season 7, we have a new computer chess world champion. Called Komodo, the software can reach an Elo rating as high as 3304 — about 450 points higher than Kasparov, or indeed any human brain currently playing chess.
In 1996, IBM’s Deep Blue chess computer lost to Garry Kasparov — then the top-rated chess player in the world. In the 1997 rematch, following some software tweaks (and ironically, perhaps thanks to a very fateful software bug), Deep Blue won. Over the next few years, humans and computers traded blows — but eventually, by 2005-2006, computer chess programs were solidly in the lead. Today’s best chess programs can easily beat out the world’s best human chess players, even when they’re run on fairly conventional hardware (a modern multi-core CPU).
The supremacy of machine over man is mostly down to two factors: Moore’s law (i.e. computer chips doubling in complexity every two-ish years), and improvements to the underlying software. In computer chess circles, Moore’s law is thought to add around 50 Elo rating points every two years — or about 450 points in the 18 years since Kasparov was beaten. Iterative versions of computer chess software can also boost the Elo rating somewhat: The new world champion, Komodo 8, has an Elo rating that’s around 60 points higher than Komodo 7a using the same hardware. It’s also worth noting that most of these chess programs are being run on fairly small computers, usually on 4 CPUs or less — while Deep Blue was a bone fide supercomputer (the 259th fastest computer in 1997, in fact).
Current top computer chess programs (December 2014) via CCRL
Current top computer chess programs, according to CCRL (December 2014). These ranks were confirmed by season 7 of TCEC.
Anyway, as computers began to clearly outstrip human chess players, there was little point in continuing to pitch them against each other. As a result, there are now computer-only chess leagues, where the top chess programs play against each other, for all eternity — or at least until the guy running the league turns his computer off, anyway. The CCRL is probably the most detailed/involved of such leagues, but there’s also the IPON and CEGTtoo. As far crowning some kind of winner, however, the Thoresen Chess Engines Competition (TCEC) is regarded by some as the de facto computer chess championship.
Our glorious leader, Sebastian Anthony, violating Watson at IBM Research
IBM Watson, shown here with a hairy British person captured within, is a spiritual successor of Deep Blue.
Season 7 of the TCEC concluded a couple of days ago, with Komodo 8 just managing to beat the reigning champion, Stockfish 5. You can actually watch the whole season via the TCEC web UI if you like — or any of the previous seasons, for that matter. Komodo’s rise to the top of the charts is most likely due to chess Grandmaster Larry Kaufman joining the development team. Kaufman is very good at evaluation — the value of a particular position of chess pieces — rather than depth (thinking dozens of moves ahead). Likewise, Komodo relies more on evaluation than depth, which results in it playing an interesting, highly positional style. Seemingly, given Komodo’s universal ranking as the top chess program, this evaluative technique seems to be working out quite well.
I’ll leave you with a fun, human-computer chess-related anecdote. In the first game of the 1997 rematch between Garry Kasparov and Deep Blue, the computer (reportedly) encountered a bug. This bug resulted in Deep Blue performing a fail-safe move — but Kasparov didn’t know about the bug, and he couldn’t work out what Deep Blue was trying to do with the move, and so he mistakenly concluded that the computer was better than him. He won the first game, but was on tilt for the second game due to the bug, resulting in him accusing IBM of cheating and eventually resigning the game — and later, losing the match. If you have 15 minutes to spare, I strongly suggest watching FiveThirtyEight’s short documentary about the Kasparov-Deep Blue rematch and the software bug that ultimately defeated the world’s greatest chess player

No comments:

Post a Comment