I suppose @torusle has real games in mind where this has been an issue so would be a good starting point?
Torusle, what settings prompted you to make this thread?
3 Likes
Hi teapowerdrobot. This is the time setting that killed me twice on 13x13:
Canadian Byo-Yomi: Clock starts with 3 minutes main time, followed by 30 seconds per 5 stones.
5 Likes
Short main time, so most of the game is in byo-yomi. So most of the game will be 6 seconds per move that you have to manage yourself over 5 stones. Doable, maybe, but difficult for sure - a borderline case I would say. I wonder if @shinuito would be willing to do some unranked experiment games starting from those settings and then reducing gradually to see when a warning appears.
Yes I think for the purpose of the calculation I linked above, which I’m relatively sure is the actual formula controlling the warnings, it would be counted as 6s per move.
case "canadian":
return !(
time_control.main_time > QUESTIONABLE_ABSOLUTE_TIME ||
time_control.period_time / time_control.stones_per_period > QUESTIONABLE_SECONDS_PER_MOVE
);
So there’s a warning when main time is <=15mins AND the number of seconds per stone is <=4s. In this case that quantity is 30s/5stones=6s/stone so it’s not giving a warning.
So then the question is, what’s the new reasonable threshold for a warning? Is this just a warning that needs to be changed for Canadian byo-yomi specifically or more generally for the other time controls.
For reference, online-go.com/src/components/TimeControl/util.ts at b80d7811e3bd785a8654dd82a4a00958f0ab2f2e · online-go/online-go.com · GitHub
// either there has to be enough time for the whole game or
// a sensible ongoing per-move allocation
switch (time_control.system || time_control.time_control) {
case "simple":
return time_control.per_move < QUESTIONABLE_SECONDS_PER_MOVE;
case "absolute":
return time_control.total_time <= QUESTIONABLE_ABSOLUTE_TIME;
case "canadian":
return !(
time_control.main_time > QUESTIONABLE_ABSOLUTE_TIME ||
time_control.period_time / time_control.stones_per_period > QUESTIONABLE_SECONDS_PER_MOVE
);
case "byoyomi":
return !(
time_control.main_time > QUESTIONABLE_ABSOLUTE_TIME ||
time_control.period_time > QUESTIONABLE_SECONDS_PER_MOVE
);
case "fischer":
return !(
time_control.initial_time > QUESTIONABLE_ABSOLUTE_TIME ||
time_control.time_increment > QUESTIONABLE_SECONDS_PER_MOVE
);
case "none":
default:
return false;
}
I believe the current warnings are:
- Simple it’s just a warning if the move time is <4s.
- Absolute it’s if the time is <15mins
- Canadian if main time <=15 mins AND period/stones <=4s
- Byo-yomi if main time <=15 mins AND period <=4s
- Fischer if initial time <=15 mins AND increment <=4s
We don’t even have to do unranked games, we can just restrict rank in the challenges to min and max 9d+ and expect no-one will accept 
But from the above it’s very like the main time won’t matter if it’s low enough like 3mins.
So it’s just whether 30s/N <=4s. So there’ll probably be a warning at N=8 stones, but not N=7 stones.
Anyway though, just because it’s quick enough to test:
No warning
Warning
4,5 even 6s: all too fast for me. But ok if they can let them do.
1 Like
I suggest:
const QUESTIONABLE_SECONDS_PER_MOVE = 5;
const ASSUMED_MAIN_TIME_MOVES_CANADIAN = 60;
const ASSUMED_MAIN_TIME_MOVES_JAPANESE = 75;
const QUESTIONABLE_INITIAL_TIME_FISCHER = 60;
-
Simple it’s just a warning if the move time is
<4s< 5 s -
Absolute
it’s if the time is <15minsalways warn absolute -
Canadian if
main time <=15 minsmain time / 60 < 5 sANDOR period/stones<=4s< 5 s -
Byo-yomi if
main time <=15 minsmain time / 75 < 5 sANDOR period<=4s< 5 s -
Fischer if
initial time <=15 minsinitial time < 1 minANDOR increment<=4s< 5 s
2 Likes
I think the current thresholds are better than these which would give too many warnings.
It’s fairly normal to have sudden death timing in live tourneys. I don’t think absolute needs a warning all the time.
On the other hand simple could probably do with a warning all the time since it’s buggy anyway!
And for the others I think 15 mins AND is a good qualification. It’s common I think to have relatively longer main time with a Fischer increment of only a few seconds and not unreasonable.
And specifically on Canadian which started all this, the problem was not the main time exactly (although if it had been long enough to finish there may not have been an issue) but rather the 6s per move for a good amount of the game.
In short I think a general rule of warning at 15m AND 6s would be fine. You could maybe go to 7 or 8s (or more for simple)
Look at my game #44684930. According to the rules provided above, The average time of each step in this game is 9.98 seconds, but it is classified as live, and I think this may need further research.
1 Like
It seem that games are classified as live as long as the average time of each move is longer than 9.95 seconds due to rounding .
Personally, I love Canadian time because it allows me to modulate the slowest pace for a game. I was going to write more, but I think it’s better in a separate post.
I don’t play blitz, but in general I think a live game that plays below 10-seconds-per-move is too fast for me. 
3 Likes
Canadian byo-yomi is the most-used time setting in tournaments in Europe. It’s a great time setting that makes it possible to make better use of your time, spending less time on obvious moves and more time on moves that require calculating.
If anything should be marked with the warning sign of “unusual time setting”, it should be:
- Japanese byo-yomi without extra periods (making it sudden death);
- Fisher with a low maximum time (making it not Fisher at all).
3 Likes
Is there a way to check how often European tournaments use certain time control methods?
I don’t know whether the European Go Database has an API that would allow to compile that information.
What I do know is that I’ve played 142 tournaments in 5 European countries, with the following statistics:
- 1 tournament used Fisher time;
- 2 tournaments used Japanese byo-yomi;
- 139 tournaments used Canadian byo-yomi.
(I’m not counting lightning tournaments, 9x9 and 13x13 tournaments, etc.)
3 Likes
Wow 
That does seem overwhelmingly Canadian.
I think I’ve seen a bunch of French tournaments used Fischer, while the Irish tournaments tend to use Japanese byo-yomi.
Could be country dependent, but still that’s at least 5 countries as you said using Canadian anyway.
1 Like
Canadian byo-yomi is ubiquitous in France because it works well with a mechanical clock and doesn’t require a digital clock. There has been some push for Fisher in recent years, but:
- Changing habits is hard;
- Tournament organizers are used to Canadian but not to Fisher;
- Players are used to Canadian but not to Fisher.
Changing from Canadian to Fisher can be especially stress-inducing for tournament organisers, because they don’t know how long a round will last with Fisher timing (there have been attempts to calculate and “equivalence” between different time settings, but what do you base the equivalence on? “average” games? or “the longest game”, since it’s that game which will slow down the tournament?
It’s also annoying for long-time players, who are used to managing their time well with Canadian byo-yomi, but have no experience with Fisher time in a serious tournament.
3 Likes
Of the three, canadian definitely takes the most conscious thought to stay on top of, taking away mental resources from Go playing.
I can tolerate playing with japanese byo-yomi, though I prefer Fischer. I think canadian time would be enough to stop me from entering a tournament.
1 Like
In the UK all BGA supported tournaments now use fisher timing. This was prompted by the BGA replacing their old mechanical clocks, which were becoming less and less reliable, with digital clocks. It’s not really possible to buy new mechanical clocks at a reasonable price compared to digital ones it seems! And with digital clocks Fischer is more natural than Canadian I suppose. People are used to counting out stones and looking at the clock for time only. Is it too odd to have to look at the clock to see time and remaining stones? I dunno.
Anyway it took a few tournaments to get used to the new clocks and Fischer but not too much trouble I’d say.
2 Likes
I don’t understand your argument about counting stones. I always count stones with physical stones when I’m playing with Canadian byo-yomi, no matter whether the clock is digital or mechanical.
I also disagree that Fischer is more intuitive than Canadian. With Canadian byo-yomi, if the Clock says “you have 17 minutes remaining before byo-yomi”, it means you have 17 minutes remaining before byo-yomi. I’m used to existing in the real world so I know what 17 minutes means and feels like. But with Fischer time, if the clock says “you have 17 minutes remaining before byo-yomi”, it actually doesn’t mean 17 minutes. It means 17 minutes plus some increments, depending on how many moves I play during this time, and depending on the value of the increments, which might differ from one tournament to the next. I have no intuition for that. And if I look again at the clock and it tells me I know have 11 minutes remaining, with Canadian byo-yomi it would mean I’ve spent 6 minutes of thinking time; but with Fischer timing it would not really mean 6 minutes, it would mean mean 6 minutes plus some increments. With Fischer time settings, the clock is literally lying to you.
Finally, in tournaments with Canadian byo-yomi, we typically play with a byo-yomi of 15 stones per 5 minutes. That’s a safety net. No matter what happens during the game, even if you spend a unusually long time, you’ll always have these 5 minutes for 15 stones to play with. I player in a tournament with Fischer time where the organiser had calculated that Fischer 10 seconds increment was “equivalent” to Canadian 15 stones / 5 minutes. When the main time ran out… Suddenly I had no safety net, and had to play my moves in 10 seconds per move. That means I had to play the remainder of the game as if it was a lightning game, something I’ve never had to do with Canadian byo-yomi, even though it was supposed to be “equivalent”.