Go vs Chess: What Makes Go Harder for Computers

Chess has six kinds of piece, each moving differently, plus castling, en passant and promotion. Go has one kind of stone and a rulebook you can state in four sentences. By any reasonable measure chess is the more complicated game to explain. Yet chess fell to computers in 1997 and Go held out until 2016. That gap is the most interesting thing about comparing them.

The rules, side by side

Go. Players alternately place a black or white stone on the intersections of a 19 by 19 grid. Stones never move once placed. A group with no adjacent empty intersections is captured and removed. You may not repeat a previous board position. The player surrounding more territory wins.

Chess. Sixteen pieces per side on an 8 by 8 board, six types with distinct movement, captures by displacement, and a win condition of checkmating one specific piece.

So Go is placement, chess is movement. Go accumulates, chess transforms. In chess the board empties as the game goes on; in Go it fills.

Why the size difference matters so much

• Branching factor. A chess position offers roughly 35 legal moves worth considering. A Go position offers around 250 early on. Looking three moves ahead in chess means examining tens of thousands of positions. In Go it means tens of millions.
• Evaluation. A chess position can be scored crudely but usefully by counting material: a queen is worth about nine pawns. In Go every stone is identical, so there is nothing to count. Whether a position is good depends on shape, influence and potential, which resists a simple formula.
• Horizon. A chess tactic resolves in a few moves. A Go stone placed in the opening can decide a fight forty moves later, in a different corner of the board.

That combination is why search alone was never enough for Go. Deep Blue beat Kasparov largely through enormous search plus handcrafted evaluation. Go needed something closer to trained intuition, which is what AlphaGo brought in 2016 by learning a sense of which positions look promising rather than enumerating them.

Chess punishes miscalculation. Go punishes misjudgement. One is arithmetic under pressure, the other is aesthetics with consequences.

What each game rewards in a human player

Chess rewards concrete calculation, pattern recognition of known structures, opening preparation and endgame technique. Strong players are often very good at holding branching sequences in mind accurately.

Go rewards judgement about the whole board: whether a group is alive, whether a wall is worth more than the territory it faces, when to leave a local fight for a bigger point elsewhere. Professionals describe it in terms of thickness, influence and balance, language that sounds vague until you have felt it decide a game.

Neither is a general brain upgrade, and claims that a board game makes you broadly smarter should be treated carefully. What both reliably develop is depth in their own domain, plus the habit of sitting with a difficult position instead of reaching for the first plausible move.

The handicap system, and why it is remarkable

Go has a feature chess lacks: a clean, granular handicap. A weaker player begins with two to nine stones already placed, one stone per rank of difference. The result is a real game that either side can win, played at full effort. That is why Go clubs can pair a beginner with a veteran and both enjoy it.

Chess handicaps exist, such as removing a knight or giving time odds, but they distort the game more crudely. If you want to learn from someone much stronger, Go makes that easier than almost any other game.

Which to choose

• Start with chess if you like concrete goals, immediate tactical feedback and a deep library of study material in every language.
• Start with Go if you like open ended positions, whole board thinking and a game where the rules stay out of the way. Begin on a 9 by 9 board: the ideas transfer and a game takes fifteen minutes.
• Play both if you want the contrast. Many strong Go players find chess refreshingly concrete, and many chess players find Go refreshingly free of memorised theory.

The rules of Go, with the capture and territory concepts explained from scratch, are in how to play Go. For the Japanese chess variant where captured pieces change sides, which sits interestingly between the two, see Shogi vs chess and Shogi rules for beginners.

The free ebook The Mind Games of Japan, included with MyRin on iOS and Android, covers Go, Shogi, Mahjong, Hanafuda and Karuta with complete rules: see what is inside the ebook.

FAQ

Is Go harder than chess?

They are hard in different ways. Go has far simpler rules and a far larger board, giving roughly 250 plausible moves per turn against about 35 in chess. That makes calculation less useful and judgement more important, which many players experience as harder to learn well.

Why did computers beat chess before Go?

Chess yields to search: evaluate positions, look many moves ahead, prune the rest. Deep Blue beat Kasparov in 1997 largely that way. Go positions resist simple evaluation and the branching factor is enormous, so strong play waited for approaches based on learned intuition, which arrived with AlphaGo in 2016.

Which game should a beginner learn first?

Chess gives faster early feedback because captures and checkmate are concrete. Go gives a gentler handicap system, so a beginner and an expert can play a genuinely balanced game, which is unusual and makes learning with a stronger partner easier.

How long does a game of Go take?

A full 19 by 19 game between club players typically runs one to two hours. Beginners usually start on a 9 by 9 board, where a game takes ten to fifteen minutes and the fundamental ideas still apply.