I like Fisher time for long multi month games on https://online-go.com/ Basically it means that I always have one spare week to think or do something else and never run out of time. For 2 days live tournaments, one minute Fisher time means that a game could last plus thatn 5 hours (300 moves -> 300 minutes) and that would make the tournament fail to complete. With a short base time plus 30 seconds Fisher time we are back to a duration comparable to 1 hour each plus byoyomi: one turn every 3 hours.

The main time period in classical byoyomi has always seemed strange to me: if it's expected that you'd reach byoyomi at all in a game of a typical length, then the main time isn't making much difference. It lets you allocate extra time to a certain critical move, but only if it's early enough in the game. If say you have 10 minutes of main time with a single 30 second byoyomi, and you're disciplined enough to use 30 seconds per move generally, then the main time is irrelevant past the opening - you'll have used it by your 20th move or so even if nothing gave you a reason to take extra time.

Yes, I'm aware that traditionally instead of main time you'd have a large (for important slow games) number of one-minute periods. Using multiple byoyomi periods is even stranger - more complex for questionable effect on time management strategy.

One idea I've had that I'd like to see implemented is "reverse single byoyomi": you have a certain amount of time per move (which resets for each move), and when exceeded on a given move you start using your "main time" (and lose when that runs out). I think that fixes the above problems: you can spend a little time on each move (but you and your opponent can expect reasonably brisk play from each other most of the time), and if you prefer you can use your reserve in late game to count the score, double-check for teire and semedori, etc. And, of course, you don't encounter extra time pressure just because the game went to an unusually high number of moves (and your opponent doesn't get to waste huge amounts of time during dame-filling just because the game went to an unusually low number of moves).

Fischer is close to accomplishing this, but it makes you "bank" and manage your thinking time for the longer moves.

Apart from the reasons you mentioned, the advantage of byo-yomi or Fisher time over the "add time after move 40" as often used in Chess, is that it doesn't create an artificial barrier.

What's so special about move 40? If you analyze chess games, you'll probably find that moves 40-45 are on average more accurate than moves 35-39 even it feels this number shouldn't be part of the game.

I also prefer Fischer, but I think that like most rules discussion, the passion for arguing about it mostly comes from a drive towards simpler systems as an aesthetic preference, not major problems with the status quo.

Does anyone use delay with Go?

In the chess world there are two delay systems used as an alternative to Fischer timing: Bronstein timing and simple delay timing. I don't know how common Bronstein delay is. In the US simple delay is the recommended timing method for USCF events.

Fischer and Bronstein (which is several years older than Fischer) are very similar. The difference is that if the increment (Fischer) or delay (Bronstein) is N seconds and the move took T seconds, Fischer unconditionally adds N seconds whereas Bronstein adds min(N,T) seconds.

For example if you were playing with a 5 second increment or delay and your move takes 12 seconds than you clock ends up at 7 seconds less than where it started.

If your move took 3 seconds your clock would end up 2 seconds above where it started in Fischer and back to where it started in Bronstein.

Simple delay simply delays the start of the clock by the delay amount. For example if playing with a 5 second simply delay when it becomes your turn your clock simply waits 5 seconds before it starts counting down. If you move before that 5 seconds up there is no change to the clock. If you take long then you will use some time off the clock.

A nice thing about simple delay is that the clock is monotonic. In Fischer and Bronstein the clock has count up after each move. A monotonic clock should make it fairly easy to add simple delay to an analog clock, for those who occasionally want to recapture some of that classic feel.

Get one of those $20-30 range analog quartz clocks that you can find at many chess and games retailers, e.g. [1]. I've never actually had a chance to look inside one but it is very likely to simply be two battery powered analog clock modules, with the switches simply toggling which clock is powered.

It should not be too hard to put an MCU in there like an ATTiny85 and have it monitor for clock presses. Put a transistor in series with each clock and controlled by the MCU so the MCU can delay the start of the clock.

You will want some way to change the delay time. Maybe add some buttons on back, but if you want to keep the hardware simple and you only support a small number of fixed delays you couple probably do something involving a pattern of button presses within a short time of powering on the clock.

[1] https://www.houseofstaunton.com/diamond-quartz-analog-chess-...

> If I had one euro per each time I heard a clock saying "30 seconds one time" I'd be rich.

Fun diversion. I figure 10 clocks in the room, a generous estimate is that they alert an average of 50 times per player per game, 4 games per day. A very solid day's work, but most weeks you can only attend a tournament on Saturday/Sunday.

I don't think you'd end up rich, but you could probably make a living.

Fischer is "simpler" if you have a Fischer clock (and "simplicity" is basically irrelevant if you have a programmable software clock). If you have to manually add time each move it's a whole other story. A traditional analog clock (as used for chess) only knows how to count a fixed amount of time for each player, and anything else is a manual adjustment. If you have to do it mechanically, surely a single byoyomi period (reset to X if the button is pressed while the needle is between zero and X) is at least as easy to implement as Fischer (move the needle back by Y every time, limited to some Z maximum).