History and overview of time control formats

Published 9 Jul 2020 by antti (last edited 10 Jul 2020)
tags: tldr

Two recent occasions got me thinking about the time control systems used in go, and their relation to the game.

First was our teachers’ exhibition game that was covered in the second June lecture, which you can view here. In the lecture, Jeff remarks that while the game might be objectively even during the middle-game fighting, in reality it favours White because of the impending start of Black’s overtime, and because White has the more comfortable fighting position. If there were plenty of time available, Black could probably look after his group while fighting for his territorial lead; but as the amount of time at hand was limited, the risk of making a game-losing mistake was rising sharply.

Second was a discussion on our online league’s Discord server. One of our students recalled a go camp from several years ago where I was playing a game with Jeff to entertain the other participants. When the game started nearing the end, both me and Jeff fell deep into thought, reading the correct sequence of moves until the end of the game. After the reading was done, we then quickly played out the rest of the game in a matter of minute or two. While this may have been good for the quality of the game, it apparently wasn’t very entertaining to watch.

The second occasion made me realise that time controls can also exist for the benefit of the spectators, and not just as a way to ensure fair time usage (and that the game will eventually end). This realisation in turn got me interested in researching the game timer’s history.

Time control in chess

The history of time control in board games probably starts with chess. According to an article on chess.com, chess games in the early 1800s could last for 8–10 hours – and chess players would actually try to stall and exhaust each other in order to win the game. This motivated organisers to come up with time limits; apparently for example hourglasses were employed. Interestingly, at first players didn’t actually lose their games if they ran out of time, but they would instead have to pay a fine to the organisers.

The first mechanical chess clocks appeared in the 1880s, and they were subsequently improved upon, eventually leading to the development of the kind of iconic analog chess timer that many go players are familiar with. This kind of an analog timer is essentially just a set of two clocks that can be paused, and it can only be used for two time control formats: absolute time (where you just have a set amount of time, say, 30 minutes, to play your moves) and Canadian overtime (where you have to play a predetermined number of moves in a set amount of time, after which your time is manually reset). Chess, however, doesn’t have a history with Canadian overtime and analog clocks because of the infeasibility of counting the number of moves made during a game. In go, by comparison, it is easy to pick out for example 20 stones from your bowl, and you don’t need to count to know when you have used them all.

The chess world used analog clocks for a long time, until, in the 1970s, digital chess timers were invented. A patent filed in 1975 describes a chess timer that can be set to perform Canadian overtime, absolute time or Japanese overtime – although these were not the names used. A bit later, in 1988, Bobby Fischer patented a timer that could add a small amount of bonus time to a player after each move. This gradually became probably the most commonly used time control format in chess.

Time control in go

Not surprisingly, the history of time control in go starts in Japan. According to Japanese wikipedia, in 1922, the Hiseikai go organisation set a time limit of eight hours for its players, following a Japanese labour movement that called for a maximum daily working time of eight hours. In 1924, the Japan Go Association was founded and the Hiseikai disbanded, and with it the eight-hour time limit; the Japan Go Association instead used varying longer time limits such as 16 hours per player, which could make a game last even for three days (presumably including breaks). For decades, these kinds of extremely long time limits were used – although at least in 1954, there was apparently a professional ‘blitz championship’ where the time limit was ‘only’ four hours apiece.

In 1953, the Japanese national broadcasting organisation nhk started a special tournament that was broadcast on radio. The time setting in this tournament was 50 minutes and 30 seconds of Japanese overtime for each player (i.e., after the main time has been used up, every move has to be played within 30 seconds). As this was still long before digital game timers, presumably the games had dedicated time keepers – as is actually the Japan Go Association custom even now for higher-profile games. A timekeeper’s job is to look at a clock and to jot down the time used for each move, and then to read out the seconds when a player uses up their main time. In the 30-second Japanese overtime, the timekeeper would read: ‘10 seconds … 20 seconds … one, two, three, four, five, six, seven, eight, nine, ten’, and the game is lost of time if the player hasn’t moved before ’ten’ is said. This is the origin of the Japanese word for overtime, byō-yomi, which literally means ‘second-reading’.

The general tendency in go, as well as in chess, has been that time limits get shorter as we move forward in time. For example Ōhashi 6p, five years my senior, recounted that when he’d just become a professional, tournament games would sometimes last until 3 or 4 in the morning (by comparison, my longest game ended at around 6 pm). Nowadays, slower professional tournament games have a time limit of three, four, five, or eight hours along with a one-minute Japanese overtime; but an increasing number of newer tournaments have even shorter settings. The most prominent of these is maybe the nhk television go tournament, where the players start in Japanese overtime of 30 seconds, but have ten instances of one minute of extra time to use during the game. Although the tournament games are not broadcast live, their commentary videos are made during the game, so presumably the time limits exist to balance the time usage in the commentary.

Choose the time format for the occasion

As far as I know, there are six main time control formats used in go: absolute time, Canadian overtime, Japanese overtime, Fischer increment, the nhk format, and the seldom-used Ing time format. Each of these has their merits and demerits, which is probably why none of them has completely disappeared from the go scene.

The absolute time format has fallen out of favour among Western go players, but is still standard in Asian amateur tournaments. This is partly because both analog and digital game timers can work with absolute time, but mainly it is so that the tournament can start the next round on time. If overtime is used, there is a risk of games running late, which may in turn cause hundreds of people to have to wait.

Western players generally dislike the absolute time setting because time management becomes a game-deciding factor; use too much time in the early game, and you will have a hard time finishing the endgame and filling the dame before your time runs out.

The Canadian overtime is a good low-cost alternative to the absolute time setting. It, too, can be used with analog clocks, but it also mostly avoids the time management issue that is prevalent in the absolute time setting – there is no longer a risk of running out of time towards the end of the game. Recently, the Canadian overtime is used less often mainly because digital game timers have become more common, allowing for better choices listed below.

An article in the Ranka Yearbook (1999, pp. 88–90, by Steven j c Mays) suggests that the Canadian overtime was invented for use in go in around 1980.

The Japanese overtime is by far the most common time control format used in go. This is at least partly because of its long history, but possibly also because Japanese overtime requires even less management from the player than the Canadian overtime – in the Canadian overtime, it is possible to wake up to having 15 stones left to play in just one minute unless one plans ahead.

The main downside with the Japanese overtime is that it can cause considerable delays in a tournament’s schedule. For example, if the time setting is 30 minutes and 30 seconds, a game that lasts for 300 moves can take more than three hours to finish; however, a majority of the players would probably be done in 90 minutes. In general, the Japanese overtime doesn’t make a good fit for tournaments that have more than one round in a single day.

I found it interesting that the Ranka Yearbook article, mentioned above, suggests that Japanese overtime may have been considered as a time format for a Quebec tournament in 1979. While the 1975 patent for a chess timer was indeed capable of this, there was probably no way that the timer was so commonplace already in 1979 – especially as the game timer patented by Fischer in 1988, which prevented a player’s time from running out, was still generally considered an ‘innovation’.

The Fischer increment is the new kid on the block for go; most go servers, aside from the Online Go Server, don’t even implement it. In many senses, however, the Fischer increment is like the Japanese overtime format done right. In the latter, to make the most use of your available time, it is optimal to always wait until the last second in the overtime before making your move. In the former, you can play your obvious next move immediately, and the Fischer increment is then added to your total remaining time, which you can save for when you actually need it. Historically, the main reason not to use the Fischer increment in go was that there were not enough digital timers available; recently, this is no longer the case, and more and more go tournaments are making the change.

The one argument I can think of in favour of the Japanese overtime over the Fischer increment is that it produces a more constant inflow of moves for the spectators; the Fischer increment can cause a quick exchange of moves followed by a long period of thinking, while the Japanese overtime will ensure that a new move is always played within x seconds.

The nhk format is perhaps the ultimate format with spectators in mind. There is no main time, so moves will get played quickly from the start as per the Japanese overtime format. Because the players have the ten extra minutes available for use, they can however still pause for a moment in the important parts of the game to read more deeply, which will help avoid a premature end of the game. Unfortunately, no go server has so far implemented this format.

Lastly, the Ing time format is a strange relic of the 19th century chess time formats where you had to pay money for exceeding your time limit. In the Ing format, there is a main time and, if you exceed it, you then have to pay an amount of points (usually two) to get an extra amount of thinking time – usually 10 or 30 minutes, depending on the tournament. You can do this three times, and, if you also exceed the last extra period, you then lose the game on time.

I have a fond memory of beating Ilya Shikshin in the European Youth Goe Championship in 2005 thanks to the help of Ilya’s time payments, but I think go players are in general happy that this format isn’t used more often.


Comments (3)


Mikho wrote 4 months, 4 weeks ago:

Great article! One question though: Why is it called "Canadian"?


antti wrote 4 months, 4 weeks ago:

The Ranka yearbook article suggests that the name stuck after Canadian players first came up with the format, and an American player then took it from Canada to the United States, describing it as ‘Canadian system of overtime’ to distinguish it from the Japanese one.

I’m still interested in what it means if Japanese overtime was also an option during this time. If game timers at that time could effortlessly run Japanese overtime, there wouldn’t be much of a reason to use the Canadian version (unless the preferences of players during that time were grossly different from the present day). My guess is that solutions involving Japanese overtime must have relied on a third player to read the clock.


KA0S wrote 4 months, 3 weeks ago:

Nice article! I have noticed that Canadian time system hasn't completely dropped out of use, at least on KGS. One reason for this might be that while KGS hasn't implemented Fischer, Canadian offers more flexible overtime compared to Japanese overtime. You can spend say 2 minutes to read out a position and then use less than 30 seconds to actually play the moves and repeat this during your next period. While in Japanese system that way could be used only perhaps once and then you'd have lost the time periods indefinitely.


Events

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


2nd Corona Cup on November 2 – December 15! Tournament details are here.


Gothenburg Online Open on December 5–6! Tournament details and registration are here.


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.