On the value of sente

Published 24 Jun 2020 by antti (last edited 24 Jun 2020)
tags: tldr, terms, sente

Sente, or initiative

Sente is a Japanese go term that roughly translates to ‘initiative’. A ‘sente move’ is a move that your opponent has to respond to, so you can then switch elsewhere on the board if you want to. ‘Having sente’ means that it is your move turn and there is nothing urgent on the board; in other words, that you are free to choose your next plan from many similarly good alternatives.

Gote, or losing the initiative

Gote is the opposite of sente. A ‘gote move’ finishes the play in a local area of the board, letting the opponent move elsewhere. ‘Taking gote’ means choosing a plan or line of play that optimises your result locally but allows the opponent to freely choose their next plan.

Go as a game of exchanges

A game of go is a series of exchanges between the two players. This can mean strategical trades during a game – ‘you take territory while I take influence’ – but also the actual process of Black–White–Black-White alternating their moves.

Every move in a game contributes some number of points to its player’s final score. Traditionally it was difficult to estimate the value of opening moves, but nowadays we have strong ais that can actually fairly accurately judge the points values of different moves. For instance, the first few moves in a game all are worth about 13 points; then during a game the values might gradually drop to 10 points, 8 points, etc.; and at the very end, the smallest moves are worth ⅓ points.

Dia. 1

What determines the winner and the loser in a game of go is simply the question of which player got to make better exchanges. Consider the (highly artificial) position in Dia. 1. There are four moves remaining: a 4-point move a ‘a’, a 3-point move at ‘b’, a 2-point move at ‘c’, and a 1-point move at ‘d’. Whoever goes first plays the 4-point move; the second player plays the 3-point move; the first player plays the 2-point move; and the second player takes the last, 1-point move. If we look at the four moves as two-move exchanges, each exchange netted one point for the first player, who is two points better off at the end of the sequence.

In Dia. 1, we would say that the player who goes first has an advantage of two points – or alternatively, that the value of ‘having sente’ is two points. This value depends completely on the board position at hand; for example, if there had instead been six moves of values of 6 points, 5 points, 4 points, 3 points, 2 points, and 1 point available, then the value of sente would be three points. At the start of the game on a 19×19 board, the value of having sente has been estimated at 6.5 points, and that is why the komi is 6.5 points to make for an even game.

While the value of the first move in a game is roughly 13 points (double the value of the komi) and the smallest move imaginable is worth ⅓ points, during a game the values of moves can easily rise to the 20s and 30s. This does not have to mean that the players have erred; for example, consider the next diagram.

Dia. 2
Black 1 creates a big follow-up move.
Dia. 3
If White doesn’t answer, White’s corner is captured and turns into black territory.
Dia. 4
White 2 prevents black 2, and is therefore worth a lot of points.
Dia. 5
White 1 only secures one point at a.

Black 1 in Dia. 2. is an innocuous forcing move that reduces a point of territory while threatening to kill White’s corner. If White doesn’t answer, Black next kills the group for a large profit, as shown in Dia. 3. White 2 in Dia. 4 saves the group, and is therefore also worth a large number of points.

Before Black gets to play 1 in Dia. 2, White might consider to play 1 in Dia. 5 himself. However, this White move only secures one point more at ‘a’, while Black 1 in Dia. 2. will almost at any moment prompt a white response because of its large follow-up move. We call 1 Black’s ‘privilege’.

The above actually gives us the strict definition of a ‘sente move’: a move that creates a follow-up move that is bigger than the initial move (white 2 in Dia. 4 is far more valuable than white 1 in Dia. 5). The player who is getting forced can of course always follow Jean-Paul Sartre’s philosophy, exert their radical freedom and not respond (or else flip the board over); but as long as the forcing move is not terribly timed, the responding move is always the largest move on the board. And, on the other hand, because the responding player is not actually taking a loss in the two-move exchange, they should be happy to respond.

While weaker players like to play out their sente moves early, strong players are actually very careful of their timing – how a player uses their sente moves is actually an excellent indicator of their playing strength. When you play a sente move, you always at least waste a ko threat, but you may possibly limit your future options as well. The best timing to play a sente move is usually just before it stops being sente, or just before the opponent would be inclined to play the same move themselves.

Dia. 6
Sente does not equal good.
Dia. 7
Black can aim at a placement at a.

Dia. 6 shows a classic amateur mistake. Black 1 is sente, forcing white 2; but black 1 is not a good move because it did not net Black anything he didn’t already have. The 1–2 exchange was always Black’s privilege, because white 1 is an extremely small move through most of the game.

Dia. 7 gives an example of what black 1 in Dia. 6 lost. Black 1 can be a good move for building a black position on the upper side; and, unless White defends, Black can next aim at the placement of ‘a’, which gouges out a significant portion of White’s corner territory.

Sente moves, in other words, should be something that you sit on and hoard until the very last possible moment to use them, because that way their value is maximised – that doesn’t quite sound like the common practice, does it? What about the value of having sente, then – is that misunderstood, too?

While the ai in general likes having sente to push the game into its preferred direction, ‘getting sente’ is not necessarily good in and of itself. Sente is always worth only what you can get with it; recall that in Dia. 1, for example, the value of having sente was only two points.

It may come to you as a surprise that many modern professional fighting games turn crazy because the two players are fighting for gote – in other words, the reverse of sente! Below is a sample from a recent ngd online league game where such a fight came about.

Dia. 8
It is White’s turn.

Dia. 8. This is the initial position.

Dia. 9
White 1 claims sente, as black 2 is a necessary defence.

Dia. 9. In the game, White captured two stones in the centre with 1, strengthening his group. In response, Black was forced to catch the triangle-marked stones with 2, and White got sente.

The problem for White is that black 2 is an extremely big move, and White cannot find anything of similar size on the board; regular opening moves in the upper-left and right corners are not simply of the same magnitude. Black will end up having played a big move with 2, and white 3 will be smaller, meaning that White loses in the two-move exchange.

Dia. 10
Because White is not in a hurry to play elsewhere, he should take his time...

Dia. 10. Instead of 1 in Dia. 9, White should have fought harder like shown here. After white 3, Black again has to capture the triangle-marked stones with 4, but this time...

Dia. 11
...optimising his local result.

Dia. 11. White does not have to play elsewhere, but he can instead continue the local exchange with a valuable move at 5. (Note that the ladder favours him.) Following, the sequence up to 16 gets played, after which White has no further way to pressure Black. Now, however, White has built a strong wall on the left side, and he can use it to take gote with a valuable corner enclosure with 17, securing most of the upper-left quadrant as his territory. Although this whole sequence constitutes a ‘gote’ for White, it is a good way of playing, because black 18 next will be of smaller magnitude than white 17.

As we have seen, ‘sente’ is not so much a positive, rather than a neutral term. In many whole-board positions where there are no particularly urgent areas, a solid gote play can easily be worth as much, or even more than a flashy sente play. Sente moves are not something that you should happily play out of the way, either, but that should be saved until the last moment for maximum effect.


This post has no comments so far. Please log in to comment!

Events

All times are in Helsinki time (eet with summer time).


New pairing for the online league made every 1st and 3rd Saturday of the month.


Public lectures on Twitch every 2nd and 4th Saturday of the month at 1 pm.


Jeff and Mikko stream on Twitch on Fridays at 6 pm.