Neil Hester

All poems © Neil Hester unless otherwritten

Name:
Location: North Carolina, United States

Sunday, December 22, 2013

Game of Go

In the last month or so, I’ve adopted an interest in the board game Go (Igo in Japanese, Weiqi in Chinese). The game is often compared to chess, since it is a game that is culturally revered and seen as a pinnacle of strategic competition. The first written reference to the game dates back to around 400 BC, and the game has persisted ever since. Essentially, if The Seventh Seal were made by some Japanese or Chinese director, rather than Ingmar Bergman, Antonius Block would have played Go with death, rather than chess.

In Go, you start with an empty 19 x 19 grid. Each player has a set of stones (either black or white), and they alternate placing stones on the board to try to claim territory and attack the other player’s territory by creating groups of stones. The black player goes first, but the white player receives a handful of points as compensation for going second (komi; number of points depends on which country’s rules you use). If one player’s group of stones is completely surrounded by enemy stones (or cannot escape), then it is “dead”. However, by creating two or more “eyes” of stones, players can create groups that are “alive” and cannot be captured.

Part of the beauty of the game originates from the development of a few smaller battles into a larger conflict: smaller battles around the edges of the board run into each other, and the form of these battles dictates the nature of the conflict over the center of the board, as well as later conflicts around the edges. You have to see very, very far ahead when you play your stones, and even though there is a strong technical element to the game, there’s also some element of creativity and “feeling” the board.

I’m in no position to comment on the technical aspects of Go, since I’m a beginner (and probably won’t progress too far; it’s tough to learn and I’m not motivated enough to regularly study the game). However, the game is fun to play, and has a certain elegance. Here are a few links if anyone is interested in trying it out.

An excellent starting tutorial

A wiki/forum about Go with lots of interesting material

A solid online Go program

~~~

In the few decades prior to the 2000s, interest in Go among Japan's youth was in steady decline. A manga and anime series, Hikaru no Go, actually changed that substantially and caused a resurgence in the game. The anime series is quite good overall; it's a coming-of-age story with a wide range of characters that experience many of the ills associated with professional game-playing, such as jealousy, rivalry, falling in to slumps, having to compete with your friends, etc.

Another interesting bit about the game: right now, the strongest Go program in the world can’t beat higher-level professional Go players. In the 1990s, IBM started working on creating a computer than could beat professional chess players. Kasparov won the first showdown with Deep Blue in 1996, but lost the rematch the following year, in 1997. Now, computers don’t play chess in the same way that humans do; at least in part, they use brute force by running thousands of different simulations, then choosing the move that yielded the best outcomes. Not exactly an elegant approach (Mikhail Tal, it ain't). Some programmers in the early 2000s estimated that using the same kind of algorithms to conquer Go will require around 1000 times more computing power, because of the longer game length and larger board. There are other problems, too, as described in this article:

~~~

[A]ccording to David Stern of the Cavendish Laboratory in Cambridge, who is working on a doctorate on computer Go with the team at Microsoft, it is very hard to determine, for each move, what its effects will be. Although the stones do not move, their presence affects the value and "strength" of the others; adjacent stones of the same colour form "groups" which are harder to capture. That's unlike chess, where it is comparatively easy to determine the "static value" of all the pieces at any time, because there are only 32 at most, where a Go board constantly fills with new pieces. "It is very difficult to produce an efficient 'static evaluation' function to compute the value of board positions in Go for a particular player," notes Stern. "The reason is that the stones on the Go board influence each other's value in complex ways. The value of a particular stone to a player is derived from its relationships with the surrounding stones, not from itself."

The effect is that in Go there are many non-ideal moves at any point, but because games last longer - typically about 200 moves (100 stones placed by each side) rather than 70 (35 by both sides) in chess - it's harder to look far ahead enough to see a non-ideal move's defects show up. David Fotland - author of the Go-playing program Many Faces of Go, still ranked one of the strongest available - reckons that for humans, reading ahead is actually easier in Go than in chess. "People are visual, and the board configuration and relationships change less from move to move than they do in chess," he told the Intelligent Go website (intelligentgo.org).

It's the visual element of the game that nobody can quite put into code. Go has a visual element; a high-good level player will reject a potential move because its "shape" - that is, the position of a stone move being considered in relation to the stones already there - "looks bad". They're not intuitively obvious. Equally, good players also talk of stones and groups having "influence" on other parts of the board, or being "heavy" or "light" or "overextended". More simply, "urgent" moves are those that will bolster the player's position; good players consistently choose the most urgent moves.

~~~~~~~

However, this article was written in 2006. Things have moved quickly since then; last year, a top professional lost against a computer program, Zen, that received only a 4-stone handicap.  Now, that’s a fairly large handicap, but we might see a Go program beat a top-level Go player in the next decade or two, far less time than many estimated about 10 years ago (I believe that new, more efficient algorithms are the reason, not just computing power; see this article for more details, plus some interesting statistics). The last bastion of human dominance in board games may soon fall. Maybe it’ll be something like this:

~~~

A Game of Igo: Honinbo and the Machine
    ~Thirty years after Kasparov vs. Deep Blue

Stones on a kaya sky—Honinbo guards the ego
of a game, whose disciples only years ago
asserted their soulful dominance of Igo,
which, as the elegant sky-shape Virgo,
is pure, and alive, and beyond. But now, Deep Indigo
solidly sits at the board, to undergo
a thousand million calculations, and forgo
the grace of constellations. It is strong, ergo,
but only brutishly skilled, like an imposter Van Gogh
who knows where—but not why—all the stars go.

~~~~~~~

Take care,
~Neil

0 Comments:

Post a Comment

Subscribe to Post Comments [Atom]

Links to this post:

Create a Link

<< Home