19x19 FOR BEGINNERS SERIES
THINGS I WISH I KNEW WHEN I WAS 25 KYU
Part 1: Sente and Gote
Welcome back! In my introduction to the series, I tried to give the beginning player a broad overview of the 19x19 Go game. In this section, I am going to drill down to the nitty gritty of one element of the game, and I could think of nothing more fundamental than the paired concepts of sente and gote. Loosely translated, the terms mean initiative and response, and once you understand how these concepts permeate and inform every aspect the game, you’ll be able to confidently answer questions like
- Is it time to attack?
- Is it time to defend?
- Is it time to play somewhere else (tenuki)?
If the Go board is the map displaying the broad story of the whole campaign, sente and gote are the urgent dispatches from the front, informing the strategic Emperor about the most important decisions to make next. If the beginning player cannot hear the information that sente and gote are shouting from the board, they will play the right moves at the wrong times, and find themselves in trouble very quickly.
As with everything else in Go, sente and gote are situational - meaning they are dependent upon the risk implicit in the shape of the stones in the present, as well as the best and worst possibilities that each player can imagine in the future. For instance - a move that might be viewed as sente in a game between two beginning players may be interpreted as a gote move in a game between a beginning player and an advanced one - because the advanced player might ignore what they perceive as a minor threat and play a riskier attack somewhere else that can make more profit. Likewise, an enclosure that a beginning player might assume is safe, might look completely invade-able to a more advanced player - so the beginning player omits a gote move to secure that territory, and the advanced player jumps in and kills the group. So, with that in mind, I propose the following working definitions for those terms:
Sente: an attacking move which is so threatening that it demands an immediate, local response because the cost of the worst case scenario of ignoring it is bigger than the potential profit of any best-case scenario of playing elsewhere
Gote: an immediate, local, responding move, which is necessary because the cost of the worst case scenario of ignoring the previous move is bigger than the potential profit of any best-case scenario of playing elsewhere
Remember how in the introduction I told you that part of learning Go involves absorbing a ton of life experience for what is possible and what is impossible within the limitations of the game? Sente and gote get right to the heart of that because at each point you need to look forward in time and imagine the best and worst case scenarios.
STARTING WITH THE FUNDAMENTALS
Now that I’ve made it overly complex, let’s back up start off with some simple examples so you can get a sense of the fundamentals. In the most basic sense, you can think of sente and gote as a way of keeping track of whose turn it is to attack. If an exchange of stones has an odd number of moves, then sente will pass from one player to another. If a player attacks in such a way that the exchange has an even number of moves, then that player gets to keep sente and is able to attack somewhere else. And here I’m going to introduce another metaphor - I find it useful to think of sente/gote sequences as the exchange of thrusts, parries, and ripostes in a sword-fight.
So let’s say Player A lunges out with their sword and attacks, and Player B raises their sword and blocks. The problem here is by reaching forward, and putting themselves off-balance to attack, Player A has actually left themselves open to counter attack - let’s say their arm is stretched forward, and all their weight is on their forward foot, exposing their side. So now Player B, seeing that weakness, counter-attacks towards the side, and Player A responds by blocking and regaining their balance. Now that this cut-block/counterattack-block sequence is over, both fighters return to a neutral state. At that moment, they are both even and ready for the next sequence.
ODD NUMBERED SEQUENCES - SENTE PASSES FROM ONE PLAYER TO ANOTHER
I’m going to use two of the simplest corner joseki variations because they nicely illustrate the ways in which sente and gote play out in real life.
So, Black has a stone at Q16, and then White approaches the corner in sente at O17. White is implicitly attacking Black - meaning - if black doesn’t respond, then white will play at R17 or R14 and have a good chance of taking over that corner. Black needs to respond with a gote move to protect its territory.
So, black plays calmly at R14. Even if White can do something on the top, Black can still get the right side of that corner, and maybe part of the top edge. But now, White is outnumbered! In the sword-fight metaphor, White has lunged forward and left itself open to counter-attack. If Black plays at M17, White will be in trouble, so now white has two choices - play a gote move and fall back, or play another sente move and press the attack. Let’s start with the gote variation:
White plays L16 in gote and falls back to secure its portion of the top. The little sword-fight exchange is over, the number of stones is balanced, and both have returned to a ready state. Also - because this exchange had an odd number of moves - sente has passed from White to Black, and now Black’s can make its next attack in sente.
Now, let’s go back and take a look at the sente variation. What if - instead of falling back - White decides the best defense is a good offense?
Since corner and edge real estate is the most valuable/efficient territory in the game, Black wants to guard their part of the corner as much as possible. Now that White has made a second attack in sente, Black must play another gote move or White will play at R17 and grab the whole corner. If Black tries to counter-attack and pincer at M16 or N16, White will still invade the corner, so this risk is too big to take.
Black plays R17 in gote to secure as much of that corner as it can, and - once again White ends up outnumbered. If White doesn’t protect its two stones and tries to attack somewhere else (like R18 or O15), then Black can play at M16 or M17 and fence White in. White can still make living shapes or run to the middle, but these will result in giving Black the upper hand and making do with smaller enclosures. White has run out of sente moves in that area for the moment, so the simplest option is to consolidate its gains and play a securing move in gote.
White completes the joseki with a gote move at L16. Once again, both sides have returned to a steady state. The number of moves was still odd, so sente has passed from White to Black, and now Black can attack somewhere else.
EVEN NUMBERED SEQUENCES - ONE PLAYER RETAINS SENTE
Now let’s look at an exchange where one player exits in sente and can attack somewhere else.
Black has a small enclosure in the bottom right and has played M1 in sente, threatening to reduce or invade White’s enclosure. If White ignores that threat, Black will play at L1 and White will have to work a lot harder to close up the gap - suffering a major reduction.
So, White plays L1 - which is pretty threatening for a defensive move as it puts Black’s M1 stone into atari. I will spare you the look-ahead at this point, but let’s just say that all of the options for Black besides protecting that stone get really complicated.
Being sensible, Black plays N1 and secures. Now white is in a similar position, in that there are many bad options possible if Black plays at L2 and puts the L1 stone in atari.
The simplest option for White is to play a gote move at L2 and secure its own territory. Because this exchange had an even number of moves Black still has sente, and can now initiate another attack somewhere else. You will see these types of exchanges all the time if you watch games on OGS. Think of them as a free action - the person initiating such an exchange gets one set of actions in one area, and then is able to retain sente to take another set of actions somewhere else.
YOUR OPTIONS WHEN MAKING A GOTE MOVE - READING THE DANGER LEVEL
When you’re playing through the Opening portion of the game, and there are very few stones on the board, sente and gote are much more stark and clear cut because the groups of stones are so isolated and far away from other groups. Each sente move has to do its own threatening, and each gote move must work to secure or protect in relative isolation.
However, as you move into the Midgame, and the board fills up with stones, each player making a gote move now has more options, because one’s existing stones and territory can be utilized to connect or secure more effectively. Whenever possible, see if you can make your gote moves in a way that
- Expands or secures more territory
- Connects your pieces in such a way that make shape or threatens to cut or disconnect your opponent
- Perhaps even turns the tables on them so that they are forced into a gote response.
A good exercise for beginners is to open up a game that has been completed and replay it move-by-move. At each point, identify whether the move is a sente move or a gote move, and where the relative urgency of each gote response falls on the following DEFCON-style priority chart:
DEFCON 1: (100% defense / 0% offense) the attack demands an immediate, local response, even if that it doesn’t threaten the other player or make more territory
DEFCON 2: (75% defense / 25% offense) the attack demands an immediate, local response, but you can angle that response in away that connects your stones or secures more territory
DEFCON 3: (50% defense / 50% offense) you are wrapping up a sente-gote sequence where you need to make one last move to settle the stones, but you’re in no immediate danger so can you grab the most territory or make the best shape you can
DEFCON 4: (25% defense / 75% offense) in attacking, your opponent has exposed a local weakness or cut-point that provides an opportunity to counter-attack in sente, and force them into a gote response
DEFCON 5: (0% defense / 100% offense) your opponent just made a move that is not threatening enough locally to stop you from playing a bigger move somewhere else - you are effectively stealing sente away. You will still need to come back and respond to that attack somewhere down the road, but you can deal with that after you make your own attack.
Practicing this type of risk-analysis on the games of other players will help you begin to understand the rhythms of your own games. With trial and error, you will begin to hear what the stones are trying to tell you and be able to respond appropriately.
Here - why don’t you try it now! Open up the game below, and see if you can follow along, calling out which moves are sente / gote, and the relative DEFCON level of each.
WHEN TO BEND THE RULES - UNDERSTANDING HIGH LEVEL PLAY
If Go were a simpler game, sente would routinely pass back and forth between players like a game of tag. In real life, it’s much more complicated, and as you watch the games of more advanced players, you’ll see them routinely ignore the usual order of sente and gote and initiate multiple attacks in different parts of the board rather than respond locally. What gives?
In making those decisions, players use the same type of risk/reward analysis as the DEFCON 5 scenario above. If the last move made by Player A wasn’t threatening enough, Player B weighs the risk of ignoring the immediate response against the potential reward of a bigger attack somewhere else. It’s important to remember that - even if their other attack succeeds - that does not make the risk from the first move go away. Player B must deal with it eventually, or it can snowball to an even bigger risk if Player A returns to that spot and plays another attacking move. To be successful, Player B must calculate the pros and cons in such a way that they risk losing a small bet in one place to risk winning a bigger bet somewhere else. Let me try to break it down with an example and give you a sense of the look-ahead and risk analysis required.
We start with Move 3 in a cross-cut game. Rather than trying to develop a framework on the left side, Black has played in the lower right. It would be fairly standard for White to now play on the lower left corner at D4, but this has some disadvantages - which are amplified in a cross-cut game.
By approaching White’s lower left corner in sente at C6, black would get a distinct advantage in having a better chance to grab a large framework at the beginning, while White would be penned in right out of the gate. White decides to turn the tables and play a sente move at O3 approaching Black’s corner instead:
Black responds calmly at R6, and now White usual response would be to play its own gote move at L3 or L4. However, looking ahead, White isn’t satisfied with Black playing in the other corner in that scenario:
So White decides to make a calculated risk - it will break the usual order of sente & gote again, grab the other corner, and sacrifice a small loss on the left by using its final gote move to secure on the lower right. Here’s how it went:
1: White plays at C4 - trying to grab as much territory as it can between C4 and its O3 stone
2: Black isn’t having it, it approaches the corner at E4 because it doesn’t want to give White that big of a framework.
3: White responds in sente at E3, threatening black’s stone.
4: Black responds at gote at F3 to protect the stone and begin making shape.
5: White pulls back at D3 in gote and connects its 3 stones, making a tentative enclosure on the corner.
6: Black has to play a gote move at F4 to close the lethal cut point.
7: If White was following the usual pattern, it would now make a gote move at C6 to complete the joseki, settle its stones, secure its corner, and extend up the left side. However if it did that, Black would complete the joseki at J4 or K4 - severely threatening White’s O3 stone. Instead, White breaks the order and plays its gote move at L3 - settling the stones on the right side.
Now it’s time to pay the piper - because White didn’t settle its stones on the left, Black can now counter-attack in sente like this:
1: Black threatens White’s C4 stone by playing at C5 in sente.
2: White blocks at B5 in gote - trying to get as much of the left corner as possible.
3: Black plays a sente forcing move at D4 - threatening to atari White at C3 or B4.
4: White protects both cut points in gote with B3 and secures the corner.
5: Black extends in gote at C5 and settles the stones.
By stealing sente not once but twice, White has now established a fairly secure hold on the lower left corner, and still got the beginnings of a framework on the right - a two for one deal! It paid a small price for omitting its last gote move on the left (giving Black more influence up the left side) but still came out (slightly) ahead in the end. So - in essence - this is why you bend the rules of sente and gote - you’re gambling that you’re going to lose the smaller bet in one place to win a bigger bet somewhere else.
Please keep in mind, this was still a relatively simple exchange - in that it wrapped up fairly quickly, both sides settled all their stones, and there are no loose groups facing immediate threat. In many high level games, advanced players will consciously make the choice to leave multiple groups in danger, and initiate multiple sente attacks in a row. By generating more unsettled groups, they are deliberately increasing the analytical complexity of the game - gambling that they can make better choices about trading small losses for big profits than their opponent.
Take a look at this game over here:
If you try to follow along, you will notice many times when there might be 7 or 8 sente moves in a row before someone plays a gote move.
By move 85 - both Black and White have multiple groups that are hanging on by their fingernails - just one or two stones away from being completely surrounded - and capturing other stones as fast as they can to survive. So, where beginning or intermediate players will follow the common wisdom of sente and gote to settle their stones and minimize complexity - advanced players go the opposite direction. They will deliberately maximize the complexity of the game so as to create more opportunities for their opponent to become overwhelmed, which increases the chance of them making small tactical errors in analysis that might snowball into a huge advantage.
I know that I am being a bit unfair by throwing you - the beginning player - into the deep end of high-level Go analysis here. My goal isn’t to confuse or overwhelm you - it is to give you some sense of the depth and scope of what is possible within the game. If anything - the main nugget I want you to take away from all this is - DON’T CREATE MORE COMPLEXITY THAN YOU CAN HANDLE.
When you’re just starting to learn this game, it’s important to play opponents who are close to your skill level, so that their ability to track sente/gote can be somewhat close to yours. While you’re playing, try to keep track of as much information as you can, and don’t create more cut points, loose groups, and un-finished gambles than you can keep track of. If you do play opponents who are more skilled than you, see if you can do it within the context of a teaching game, so that the two of you can have a discussion about why each move is being played, and hopefully you can learn new ways to attack, connect, and secure more effectively. If you play someone stronger and lose, review the game, and look at the ways in which they attacked or defended more efficiently than you. Try to find those points where you created more complexity than you could handle, and remember those things so you can avoid them in future games. Over time, you will find that the things that took you a lot of effort at the beginning soon become second nature through practice and repetition. You will slowly absorb a body of experience for what is possible vs impossible, and your abilities to attack and defend will grow organically out of that.
On a final note, learning to read the exchanges of sente and gote had a huge side-benefit for me, because it transformed the process of viewing high level games into a fascinating spectator sport. Think about watching James Bond playing Poker against Le Chiffre in CASINO ROYALE. Even if you know very little about poker, you’re not just watching two men pushing piles of money at each other. What you’re paying attention to is how each person manages their own risk by creating more risk than the other one can handle - and then that player turns the tables on them by creating even more risk!
Once I learned to read the language of Go, watching high level games became an edge-of-my-seat suspense story full of fearless attacks, sudden reversals, and by-the-skin-of-one’s-teeth escapes from certain death. Not only did I find myself intellectually and emotionally engaged, but my brain was slowly absorbing information about what was possible and what was impossible within Go, and I started picking up strategies from much more advanced players than I would have encountered playing other beginners on my own.
Good luck, and see you next time!
NEXT ARTICLE: SETTLING YOUR STONES