I have now played in several tournament with Fisher timing. If I get a shortage of time, it usually happens in the middle game. Then in yose I usually accumulate more time. I came up with an idea to have accelerated Fisher timing. The increment decreases when the game progresses. Here is an example, but other schemes are also possible.

45 minutes of basic time
20 seconds increment at start
after each 10 moves the increment is one second smaller
after 170 moves the increment is 3 seconds and does not decrease

This way the players could use their time efficiently, but it would be easy for the organisers to predict when the games are finished, and therefore to schedule the tournament. What do you think?

I like the idea. (For reference, for your numbers, given a game of about 130 moves each, which is about average, both players get an additional 30 minutes on top of their 45 minutes basic time)

Personally, I also think there is value in adding a small Bronstein factor, i.e. a delay before the clock starts running which does not accumulate. If you have such a delay of e.g. 3 seconds, it means there is no value in trying to slam down your stone as quick as possible after your opponent's move, which can be distracting and could be perceived as rude.

Try playing a game with 3 seconds per move and see what happens. I think that you will very quickly adjust that part of your idea.

Why do you think it is a problem that you use less than the full time that you are given? Keep in mind that is the 'problem' you are trying to fix.

As far as scheduling goes I assume that you do not use either Canadian or byo-yomi since both are worse than Fischer in their effect on the predictability of the game time. I say that because due to the spilling of unused time they normally require longer periods than Fischer in order to provide a positive game experience for players.

_________________
Dave Sigaty
"Short-lived are both the praiser and the praised, and rememberer and the remembered..."
- Marcus Aurelius; Meditations, VIII 21

That's how it works on GoQuest, 13x13 is 5 mins main time plus 3 secs Fischer[sic]. I have got down to little time and played fast to build up my time again, though of course that's easier online than in real life with a clock to press. But even 3 seconds (increment not byo-yomi, you can still have some accumulated time left) in real life is better than sudden death for dealing with those jokers who keep on playing when 100 points behind to win on time (and it's probably pretty easy yose by move 340 unless there was some really long ko). For the record I have played 50 stones in 5 minutes Canadian in real life tournament games which is an average of 6 seconds a move but actually played quite a lot faster than I needed to.

I'm jealous of stronger players' ability and apparent confidence to play endgame so quickly. My experience is that it is not just oh, hane connect here, monkey jump there, but more often: "wait, is this even sente? Does this kill?" At 3 seconds per move I'd just be filling in random intersections near the edge of the board. The outcome of the game would have little correlation with anything that happened in the first 100 moves. (I understand Fischer time and believe it's theoretically good. However, I have no experience playing it in real-time games. So while I am aware that a 3 second increment is not the same as a 3-second byoyomi because the chance to accumulate, and that should make for a difference in feel, it still sounds scary.)

The saving grace in Matti's suggestion is the 45 minutes basic time. Since I can play 45 minutes absolute I would probably play as if the game were like that and just save the overtime in case of emergencies like multiple kos or games with a lot of dame. (BTW, Matti in your suggestion, does the 20-second increment start after the 45 minutes is elasped or right after the 1st move? Also, I'm curious what the time limits were in the tournaments were you felt pressure in the middle game, just for comparison.)

Don't think I can play the endgame well at six seconds a move. I might play like a 10 kyu rather than 4 dan. But that is usually enough to win a game at move 300 that I'm winning by 100 points against a 2 dan. On the other hand with sudden death I have zero chance to win.

For some other examples my so far only rated tournament win against a 6d was at the Brussels tournament which has quite generous overtime, 15 stones in 5 minutes iirc. I went from about 30 points ahead to 15 in endgame. Against a 7d in the online PGETC with 25 stones in 10 minutes from maybe a generous 20 to a very fortuitous 0.5. Even against Dinerstein 3p/7d I probably lost about five points in the endgame and that was with near infinite time.

People seem to think that a 3 second increment is 3 seconds per move. It is not: only if you have used up all your previous time. If you reach the 3 sec/move point with 20 minutes left then you will hardly notice: you use 10 seconds per move and get 3 second added on, for a net of 7 seconds. All it means is your time is decreasing more slowly thank you are actually playing.

_________________
Still officially AGA 5d but I play so irregularly these days that I am probably only 3d or 4d over the board (but hopefully still 5d in terms of knowledge, theory and the ability to contribute).

This kind of delay is good, IMHO. Although there may be rules that allow on to stop the clock when removing captured stones, doing so is somewhat disruptive so having time to at least manually perform that operation for one or two stones is beneficial.

Interestingly, in the USCF rules it says this:

Matti, for comparison can you tell us what time control was used in the tournaments you played?

I realized the text above might be slightly ambiguous. I meant that after a player has åplayed 10 moves and the opponent also 10 moves, the increment is reduced.

Between 45 minutes with 20 seconds increment and 20 minutes with 10 seconds increment.

A player gets down to 3 second increment after 340 moves have been played. Unless there has been long or numerous ko fights there would be only dame left. If 3 you still consider 3 seconds too short,one might adjust the scheme and finish the game with 5 seconds increment.

I think this isn't really the problem he is trying to fix. Rather, as with most timing systems, you are trying to make sure that people have as much time available for their games as possible while also allowing tournament schedules to stay within certain bounds (and while protecting them from timeouts).

To keep your schedule, you sort of have to plan your tournament around the worst case of your timing system. If the way people use their time *on average* is far away from the worst case, your solution is not optimal.

So Matti is noticing a gap between the way people *normally* use their time (i.e. using very little of it in the late endgame) and the worst case (using all your time even in the late endgame) and is trying to make more of that time available to more players.

Is fischer time even used in tournaments? Are their clocks which can do that?

_________________
Still officially AGA 5d but I play so irregularly these days that I am probably only 3d or 4d over the board (but hopefully still 5d in terms of knowledge, theory and the ability to contribute).

Often enough that the European Go Federation added rules to their rating system to determine what weight should be given to games played with Fischer time (see tournament classes at: http://www.europeangodatabase.eu/EGD/EG ... ubmissions).

In the Netherlands, I see a lot of DGT2000+ and DGT2010 clocks being used, which definitely support Fischer time. My own Excalibur clock at home also supports it.

When I tried out Fisher-timing, I experienced the same problem: with time settings that are correctly adjusted for opening and middle game, too much time will accumulate during endgame.

My conclusion was that however superior Fisher might seem in theory, Japanese byo-yomi is simple and gets the job done best.

_________________
Stay out of my territory! (W. White, aka Heisenberg)

I'm also a fan of byo-yomi. Not because I think it's theoretically better, but because it's simple: X seconds a move. Period.

I prefer to spend my time thinking about the game than thinking about the time period. From this perspective, systems like Fisher and Canadian are too distracting for my taste.

_________________
be immersed

Check Matti's OP again. There are no average people there only 'I'. As we all do, Matti has noticed something about the way he plays and extrapolated it to everyone else.

This I have to disagree with. Obviously absolute timing is the only timing system that sets out to make sure that people have as much time available for their games. In three weeks I will take part in the second Tokyo qualifier for the Takurashuzohai, Japan's largest amateur tournament. Some 1400 people will sit down in the main auditorium in Tokyo's Kokusai Forum building and contest 5 rounds between 9:30 and 18:00. We use 40 minutes absolute timing.

All other timing systems set out to do something else. Most of them set out to mitigate the problem of time crunches at the end of the game (your timeouts). Fischer, Bronstein, byo-yomi, and Canadian are all examples of this. They all take away part of the time available earlier in the game and deliberately deliver it later to prevent players from from using it too soon. Realize that this deliberately sacrifices the goal of giving people as much time as possible for their game. It then becomes a matter of how much to penalize people in the early going and how much (and how) to deliver them time in the late going.

As to whether people don't need time in the end game, I bow to your knowledge of how you yourself play. Certainly I do need it. I would refer everyone to all the videos of pro TV games now to be found on YouTube for overwhelming proof that pros need it. Perhaps one of the themes that should be included in all of our 'think like a pro' threads should be - 'use your time like a pro'. Matti would no longer see his time accumulating if he did so.

_________________
Dave Sigaty
"Short-lived are both the praiser and the praised, and rememberer and the remembered..."
- Marcus Aurelius; Meditations, VIII 21

I don't think we disagree actually. I specifically added the "to protect them from timeouts" clause because absolute time delivers the most time.

So I'm saying timing systems try to achieve a balance between three things. 1. Maximize player time, 2. Keep tournament schedules 3. Prevent game results by timeout. Note that point 3 is not just about protecting players against themselves, but also about protecting them from unscrupulous players who would play on after the endgame just to run an opponent out of time.

The Maidenhead tournament uses accelerated Canadian overtime: the main time is 1 hour, with then 10 stones in 5, then 20 in 5, then 30 in 5, then 40 in 5 etc. so works with simple analogue chess clocks. I had to clarify with the TD if the etc meant 40 for ever or 50 then 60 then 70 and so on and it was the latter. (I probably should have played more than 60 moves in the main time!) Those time settings allow a European class A rating and the schedule allows maybe 2h30m between rounds which is not the theoretical maximum but usually enough. I like having overtime as a prompt to play faster, though perhaps if we played with 75 minutes sudden death and I managed to self-discipline myself to treat it as 60 minutes main time then the effect would be similar. But I just dislike sudden death for the fact people have an incentive to play on hopelessly lost games to win.