I am looking for a timing system that would yield games on a server which closely resemble casual play at my club. The closest I found so far is the timing used on IGS where one gets 10 minutes to play every 25 moves (Canadian timing). I've been trying Capped Fischer on OGS recently but that didn't work as well (see http://www.lifein19x19.com/forum/viewto ... =10&t=9736 for details). Yet, Fischer timing has the advantage that one always has at least the increment to play a move, whereas with Canadian timing one occasionally gets in a serious trouble with several moves to play and almost no time left. This got me thinking that perhaps a combination of Fischer and Canadian might be better than both. What I'm going to describe may have been proposed before but I'm not aware of it.

The idea is to use Fischer timing (uncapped) during a set number of moves (I will call it the period, which could be, say, 25 moves) and then reset the clock to some predefined initial time at the start of the next period. You need to specify the increment and the period, then the initial time is computed as the product of the increment and the period. For instance, if the increment is 12 seconds and the period is 25 moves, then the initial time will be 5 minutes (12 seconds times 25 equals 300 seconds). The game starts with 5 minutes on the clock and the timing works like regular Fischer timing (uncapped) during the following 25 moves. Once 25 moves have been played, the clock resets to 5 minutes for the next period, regardless of the current value on the clock. In this example, the game is guaranteed to last no longer than the longest game possible with the 10min/25moves Canadian timing.

This design takes care of two things I don't like about Fischer timing: 1) the possibility to lose all your saved time and end up playing the rest of the game by the increment, 2) the flip side of (1), which is when one of the players racks up a huge amount of time and then slows down the pace of the game too much. Constrained Fischer timing retains the advantages of Fischer timing and maintains a good pace throughout the game. It is also quite flexible. For instance, one can choose the period to be one move. Then it is equivalent to a single-period byo-yomi. On the other hand, one can choose a very long period, say, 300 moves. Then the whole game will consists of a single period with long initial time, so it will be equivalent to regular Fischer timing.

A vaguely similar improvement on Canadian has been proposed: http://senseis.xmp.net/?SteadyAverageTiming. I suppose I like your idea better, however, and don't think I'd particularly mind it as a way of playing.

_________________
Occupy Babel!

Thanks for the link. I'm glad you like my proposal.

I don't think Steady Average Timing addresses the drawback of Canadian timing. You could still end up with several moves to play and very little time left to play them, whereas what I'm proposing guarantees that you will have at least the increment to play a move. Moreover, you won't get stuck with the increment forever, since your clock will reset after the period is over.

Regarding Canadian timing, one of the problems it addresses is analog clocks that are not easily incremented or reset. Unless a proposal to fix it takes into account that the clocks may not have easy increments or resetting, it's never going to replace Canadian timing.

I guess that you mean the same way that increment timing (in its various competing forms) has not gained any ground in chess!

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

I can't really speak to the chess world, as I don't play competitively or really much at all. At least as regards go, for our club tournaments we're not going to get away from Canadian overtime until either there is a better method that is easily workable with analog game clocks, or somebody ponies up the thousands of dollars it would take to replace all of our clocks. It's really not feasible to increment by, say, 30 seconds every move when the clock face is not precise enough to do it and incrementing requires that you pick up the clock to turn the dial in the back, generally while it's paused.

My concern is only with casual games on a game server like OGS. I wasn't really talking about tournaments (least of all tournaments that use analog clocks). I think it's a bit harsh to discard alternative timing systems on game servers just because you won't be able to use them in tournaments that rely on analog clocks.

Ah, I think the bit about games 'similar to those in clubs' in an earlier post threw me off. I don't see much reason to use Canadian timing online, since more complicated systems are easily implemented.

In some sense, playing on a game server is much like "training" for real-life games. So, the usual time settings used in real life are mimicked in servers. It could be improved? Sure. Should be improved? Maybe, improvement is good. For the record, I don't like Canadian timing either, because it can make for long-drawn games that seem to never get to an ending (I think I started disliking it back when I played chess, and it's rained a lot since then.)

_________________
Geek of all trades, master of none: the motto for my blog mostlymaths.net

Almost every tournament I've been to has been at least 45 minutes + byo a side. Online games are much, much faster than this, even the slow ones (automatch that is).

What I meant is that servers usually don't offer time settings unavailable in real life, but more like offer what you expect from a normal club clock. I have also played only in 45 (or 60) main time tournaments, and my KGS games are 30 min, usually. This does not detract from being a training: in basketball game-training you don't train a full game, but either half-court, or use shorter time periods.

_________________
Geek of all trades, master of none: the motto for my blog mostlymaths.net

I only disagree that the time lengths are similar, they're not. I'd fully agree that "training" on shorter time limits is very useful for someone. Actually, a variety of limits is probably a good idea to focus on different aspects of your game.