Forum for Timing Systems - Redux
Time plus [#816]
Interestingly, all three games show a common feature. The mode is around 3 sec/move. This suggests to me a simple form of timing. It has two parameters, sec/move and extra time.
The idea is that you always have so many seconds in which to make a move. If you take longer than that, the extra time you take is deducted from a reservoir of time that you start with. Once your reservoir runs out, if you exceed your time for a play you either lose the game or lose a play (forced pass).
In the example games, the slower players might be satisfied with 10 sec/move with a 40 min. reservoir, or even 5 sec/move with a 45 min. reservoir. The faster players might be happy with 5 sec/move with a 20 min. reservoir. (Since the mode is around 3 sec/move, regardless of overall speed of play, 5 sec/move might be generally acceptable.)
Since you get no credit for taking less than the time alloted per move, there is a kind of spillage. However, it is quite small.
The main difference I see between this and Fischer timing is that you do not build up your reservoir, so that you have more flexibility about when to use it. You can use it all in the opening, if you want.
(A little later.) Hmmmm. This is basically Bronstein timing, isn't it?
In terms of how the time works it is Bronstein timing. However, in terms of the way you imagine managing your time something is missing from pure Bronstein. Usually Bronstein operates visually the same as Fischer, an increment is added to your main time (up to but not beyond where you started). This way does not distinguish between the increment and the initial main time. Yet I think it is reasonable to say that there are two distinct types of time here: unconditional main time and conditional increment. The event that got me started on all this was the purchase of a chronos game clock, which I have used to test different systems in a Saturday study group that I play in at the Nihon Ki-in. It really is great (not perfect, just great :-). It has 3 different ways of displaying what Bill describes. The best (IMHO) is what the chronos manual calls "andante" modes. These display the conditional (incremental) time on the left side of each player's display and the unconditional time on the right side.
Here you see the clock ticking down in the early stages from original settings of 15 minutes initial time and 9 seconds increment (delay, whatever) per play. People really enjoy this display, we have played with it frequently at the Ki-in. An advantage over Fischer timing is that people grasp what is happening more quickly and accept the idea more readily. Unfortunately, after all my work on this, I conclude that Fischer is better. However, the Fischer display is boring :-)
I have considered using an "andante" style display with Fischer - increment on the left, initial time on the right and left over increment being added on the right. This would make it easier for newcomers to understand what happens in Fischer. However, it may give a false impression that Fischer is more complex than it really is. Above all Fischer is simple. All time received is equivalent, all unconditionally available to the player.
Dave: An advantage over Fischer timing is that people grasp what is happening more quickly and accept the idea more readily. Unfortunately, after all my work on this, I conclude that Fischer is better. However, the Fischer display is boring :-)
When you get to longer games, where you want the minimum time per move (= delay) to be substantial, like 30 sec. or more, I agree that Fischer timing is plainly superior. But for online play I think that a delay of 5 sec. is appropriate for most games, even with an initial allowance of 1 hour. In that case the Bronstein spillage is trivial.
A little web searching revealed that among chess players a Bronstein delay of 5 sec. is popular.
The spilling might be marginal under some circumstances. On the other hand, I don't see how it could constitute an advantage in the first place. Bronstein feels to me like "Fischer light" - following the same principal idea, but stopping halfway.
As I understand the situation in the chess world, the U. S. Chess Federation decided on a spilling delay (Bronstein or similar variants) some years ago. As these things usually go, the FIDE then standardized on Fischer! :-) You can see quite a few discussions (and flamewars) on this issue in rec.games.chess.misc.
Although some of the other systems can be fun to use, ultimately I think that Fischer is the best choice for serious play. Additionally, I think it is more so for Go than for chess! This is due to the difference in typical game length. A typical game of Go is 120 to 150 plays per player versus 40 to 60 (?) for chess.
- If you use Bronstein with a small increment and you run down your main time, you are out of luck. Every X seconds you must play. With Fischer it is different. If you play quickly (and the analysis shows that people do naturally a significant portion of the time) you will accumulate a few spare seconds to use later under even the shortest increment.
- On the other hand if you use Bronstein with an increment large enough to ensure reasonable thinking time per play when your main time is gone, the reward for playing "unnaturally" becomes larger while the longer game length in Go means the impact of the choice on total game time will be much larger than in chess.
I argued pro Fischer timing above, but IF the players really wish to restrict the freedom which part of the game to spend most time on, I'd prefer Bronstein over pruned Fischer. Bronstein time delivery is conditional, but at least, that condition is simple, evident and consistent over all moves.
@blubb
Sorry I didn't follow this. What is the context of your comment here? Is "pruned Fischer" related to the new material I added at the end of the main page (in which case a new topic perhaps) or to the material above? In any case, Bronstein will not regulate the flow of the game differently than Fischer until the main time runs out at which point Bronstein reduces to byo-yomi (the big negative side of Bronstein in my opinion).
Yes, sorry, I better should have opened a new thread. My comment indeed refers to your addition to the main page. I attached it to your last loosely related post at that time, since you prefer discussing the main page in the forum.
In any case, Bronstein will not regulate the flow of the game differently than Fischer until the main time runs out
There is a strategical difference that applies from the start:
Under Fischer timing, you can gain something from playing fast, no matter at what stage of the game. That also means that, in terms of "time tally" between you and the opponent, playing not so fast necessarily costs you something. Under Bronstein however, even when there is main time left, there is a protected minimum thinking time per move (that is, the delay). Playing faster than that doesn't give you any more, so exhausting that free time doesn't hurt you, either. In some sense, Bronstein timing resembles a protectionist and Fischer timing a libertarian concept. ;)
Taking this a bit further, mixing both could make sense, too. We could ask "Which activities do we want to be subjected to time competition, and which shall be sheltered?" This leads to a hybrid we could call "Delayed Fischer". For instance, the first five (or even ten?) seconds after each opponent's move could be a Bronstein delay, taking account of varying perceptional and physical abilities too much influence of which might be considered unfair. After that delay, Fischer would start counting as usual.
Of course, since the distinct sorts of times can't really be assigned to certain activities, there would still be a difference. A more agile player might use more of the delay for real thinking already, such that his/her main time would be reduced by less. Nevertheless, when things came down to zero maintime, a languid player would still have e. g. 10 seconds per move at no cost.
I am not sure whether the added bit of fairness is worth the complication, though.
blubb ...Under Fischer timing, you can gain something from playing fast, no matter at what stage of the game. That also means that, in terms of "time tally" between you and the opponent, playing not so fast necessarily costs you something. Under Bronstein however, even when there is main time left, there is a protected minimum thinking time per move (that is, the delay). Playing faster than that doesn't give you any more, so exhausting that free time doesn't hurt you, either. In some sense, Bronstein timing resembles a protectionist and Fischer timing a libertarian concept. ;)
You're post is very interesting to me because it shows how perceptions can differ. My real playing experience with Fischer is still quite limited. So far, I would think that the advantage you can gain (or lose) over your opponent by playing fast under Fischer is less than it is under the Bronstein, byo-yomi, or absolute timing. In Fischer your opponent has much less opportunity to dominate you via the clock since you can rebuild your supply of time as long as you are basically comfortable with the rate at which time is being added on each play. For example if you are playing with a 20 second Fischer increment, having 2 minutes left on your clock versus 5 minutes for you opponent is a difference equal to the amount of new time you are given every 9 plays (20 seconds times 9 equals 3 minutes). This seems a lot better than facing the prospect of being held to 20 seconds per play for the rest of the game while your opponent still has several minutes left. This is not to mention the dreaded 2 minutes to finish the rest of the game versus your opponent's 5 minutes (been there, done that :-)
So far, I would think that the advantage you can gain (or lose) over your opponent by playing fast under Fischer is less than it is under the Bronstein, byo-yomi, or absolute timing.
Please take a look at the following graphs (unfortunately, I can't host them currently by myself to get inlineable pictures):
![[ext]](http://wm26.inbox.com/thumbs/1_8a9d0_6586393f_tn.jpg.thumb)
http://wm26.inbox.com/thumbs/1_8a9d0_6586393f_tn.jpg.thumb
The slope of the Fischer time advantage curve is the same as with absolute time, and steeper than Bronstein, as long as they differ. The 10 s bonus / delay / byoyomi length is arbitrarily chosen and might be replaced by any suitable value.
![[ext]](http://wm26.inbox.com/thumbs/2_8a9cf_144cc4d_tn.jpg.thumb)
http://wm26.inbox.com/thumbs/2_8a9cf_144cc4d_tn.jpg.thumb
This graph shows the incentive for immediate play. Both Fischer and absolute timing are constantly at 1.0. Bronstein and delayed Fischer begin with a period of 0.0 momentary time pressure and then continue at 1.0, as well. Japanese byo-yomi is as jumpy as expected.
In Fischer your opponent has much less opportunity to dominate you via the clock since you can rebuild your supply
Personally, I clearly favor Fischer timing. Anyway, let's be fair about the pros and cons: your opponent may likewise try to increase his/her supply better than you, by playing fast and even aggressively. I think, when you're involved in complicated fights with not too much main time left, time tally matters more than potential one-sided recovery alone. Fischer rewards faster play, in the very first second after a move as much as in the 23rd or 37th. One may like that time competition or not, and of course, its significance depends on the actual board and time situation, but with Fischer, it is always there.
Another aspect is that Fischer time control allows very efficient time management - not at least because nothing spills -, so the bonus can be (and to my knowledge, usually is) set tighter than e. g. a Bronstein delay or even byoyomi would be.
blubb wrote:
Taking this a bit further, mixing both could make sense, too. We could ask "Which activities do we want to be subjected to time competition, and which shall be sheltered?" This leads to a hybrid we could call "Delayed Fischer". For instance, the first five (or even ten?) seconds after each opponent's move could be a Bronstein delay, taking account of varying perceptional and physical abilities too much influence of which might be considered unfair. After that delay, Fischer would start counting as usual.
After viewing your graphs I see what you were getting at here. This is complicated. We might set out to help balance differences in physical abilities (my impression from reading about the chess world is that this was one significant consideration in the original introduction of a delay). However, there is no "protection" that does not cut both ways. The spilling increment before the Fischer increment means that in a game between two physically comparable players the clearer thinking, quicker player will be penalized by losing the unused time.
Like you, I think Fischer is to be preferred in the end - simplest is best.
Dave: As I understand the situation in the chess world, the U. S. Chess Federation decided on a spilling delay (Bronstein or similar variants) some years ago. As these things usually go, the FIDE then standardized on Fischer! :-) You can see quite a few discussions (and flamewars) on this issue in rec.games.chess.misc.
Hmmm. I wonder how much Fischer's U. S. patent on clocks using his timing scheme has to do with that. The USCF may not have wanted to guarantee royalties or license fees to Fischer, while the FIDE may not have had that concern.
I think that the psychological aspect is worth exploring. As you say, Dave, with a fixed delay (Bronstein) system, the players will tend to use up the entire delay time on each move, even when they would naturally play faster. (However, the behavior of Japanese pros in byo-yomi does not always bear that out.) That being the case, a short delay seems appropriate with Bronstein time.
Consider the NYC tournament which used Fischer timing. The players got 10 min. plus 20 sec/move. If they each make 150 moves, that comes to 1 hour. Perhaps the tournament had 2 1/2 hour rounds.
I think that that schedule is problematic. Why? Because the original thinking time is too short. Suppose that a player takes 1 min/move in the opening. After 30 moves (15 per player) he has used up his 10 min. and is effectively in byo-yomi.
How about 35 min. plus 10 sec/move? After 30 moves at the same pace the player still has 22.5 min left. That seems more reasonable to me. It gives the players more flexibility in their use of time.
The psychological point? With Bronstein timing, I don't think you would even consider 10 min. plus 20 sec/move.
Let's take this even further: 50 min. plus 5 sec/move. 5 sec/move is not long enough for any real thinking. If you don't know what to play, you are out of luck. However, playing at that speed in the last stages of the endgame is not unrealistic, as a rule. (I can usually finish off a game at 10 times that speed. ;-)) And 5 sec/move is enough to protect against unreasonable plays that simply prolong the game.
OC, we are approaching fixed time limits. But I would not go below 5 sec/move, because you do need protection against unreasonable prolongation of the game.
With 5 sec/move, the basic idea is that you have a certain amount of thinking time, and you always have time for a semi-automatic play. Overtime is not really for thinking. If that is your model, the ability to extend those 5 sec/move in overtime does not fit. You do that when you want to use the overtime period for thinking.
I've played in the NY Go Center tournaments using Fischer timing, and this doesn't seem to be a problem. First, players don't use 1 minute per move for 15 moves; instead, the first couple of moves are quick; then a couple of moves taking a minute or so each; then a joseki begins and several moves get played quickly, and so on. So you might have, say, 9 moves that averaged one minute per move, and 6 moves played in a minute total; at move 30 you'd have 5 minutes left.
The increment seems to work well later in the game, as well: you can calculate until you're low on time, and then play several moves quickly (that you calculated) and build up a minute or two to think again. I'm only 4 kyu; maybe it's different for the strong players--but fast moves with occasional pauses for thought does seem to be the natural playing rhythm implied by the graphs on this page, and Fischer timing with a longish increment seems to support that rtythm well.
I like the twenty-second increment a lot. A ten-second increment wouldn't be too bad, but I'd find it hard to physically play moves at 5 seconds a move, never mind trying to (for example) refute an unreasonable invasion, even by instinct.
Hi, edgy!
I'm curious. At about what move do you tend to run out of the extra time?
BTW, I based the 1 min/move on myself. That's how I played openings when I used to play in tournaments. And that was with 2 hr. rounds. I would not like to enter the middle game with only 5 min., either. One move could easily take that much time.
I was well aware that my opponents did not play the opening so deliberately. But with regular byo-yomi, I could use my reservoir of time as I saw fit. With 10 min./20 sec. Fischer timing, I would not have had that flexibility.
Another BTW. I think that playing joseki quickly is a bad habit. (But then, I would think that, wouldn't I? ;-)) Aren't openings much more crucial in chess than joseki in go? Here's what E. A. Znosko-Borovsky said in How Not to Play Chess:
Do not make the opening moves automatically and without reflection.
Hi, Bill,
I try to keep a couple of minutes on my clock, so if something complicated comes up, I can at least try to read it. (That's not a possible way of time-management for a byo-yomi game, because you can't build a time reserve back up).
There's been only one game (out of eight) where I played a lot of moves with less than that. My opponent was a very young 2 kyu who responded to almost all my moves instantly. We got into a complicated fight right in the opening, and around move 40 I had to play several moves woth no extra time left. He had 20 minutes on the clock at that point. In the other games, I never spent more than a move or two with under 2 minutes, so I think the answer for those games is that I never ran out of the extra time.
I don't see how one could use byo-yomi in a four-round one-day event (like the NYGC tournaments), without having either a very short main time or a very short byo-yomi period. If 20-second byo yomi periods were used, wouldn't the main time still have to be 10 minutes, just as for the Fischer timing? Would one really use 50 minutes main time and 5-second byo-yomi periods as a tournament time limit (or even 35 minutes and 10-second byo-yomi)? Or, what am I missing?
edgy: In the other games, I never spent more than a move or two with under 2 minutes, so I think the answer for those games is that I never ran out of the extra time.
So when do you tend to get down to the 2 min. level?
edgy: I don't see how one could use byo-yomi in a four-round one-day event (like the NYGC tournaments), without having either a very short main time or a very short byo-yomi period.
When I was playing tournaments people avoided byo-yomi. So for a 2 hr. round it was feasible to give each player 45 min., with 30 sec. byo-yomi. Most games finished without going into byo-yomi, and most of those that did so had byo-yomi for only 10-15 min. You can see that the worst case scenario is horrible, however. These days people are more used to byo-yomi and many would probably take advantage of such time limits.
Well, maybe they would not. Take a look at Dave's graphs. It looks like the players would not want to go into 30 sec. Japanese byo-yomi until after move 200. Canadian byo-yomi is another matter, however, because you can spend a good bit of your alloted time on one or two moves in a block.
BTW, there were also tournaments where we played 5 or 6 rounds per day with no clocks at all. :-)
First, I need to correct myself: since normal Fischer timing adds time starting from move 1, a Fischer n minutes + m seconds increment per move is obviously longer than n minutes base time + m seconds byo-yomi. What would be equivalent is a system with n minutesbase time, and then m seconds increment _after the base time is used up_. That would be the same schedule as byo-yomi and be more flexible for the players. I wonder if the clocks the NYGC has (Excalibur) can be coaxed to do that.
So when do you tend to get down to the 2 min. level?
Usually in the middle-game, maybe move 100 or so on average. Before move 80 means I've started an unreasonable fight and am trying to dig myself out of it.
When I was playing tournaments people avoided byo-yomi. So for a 2 hr. round it was feasible to give each player 45 min., with 30 sec. byo-yomi. Most games finished without going into byo-yomi, and most of those that did so had byo-yomi for only 10-15 min. You can see that the worst case scenario is horrible, however. These days people are more used to byo-yomi and many would probably take advantage of such time limits.
So what would be a reasonable time-limit for contemporary tournaments with byo-yomi or the Fischer variation I described? 30 minutes and 20 seconds would work if we could count on players getting to move 100 moves before entering byo-yomi. Is that reasonable? It's about my pace, and I'm usually among the last to finish but not the very last.
There were also tournaments where we played 5 or 6 rounds per day with no clocks at all. :-)
Well, as long as you know in advance that everyone in the tournament is a reasonable sportsman, anything can work. 8)
Bill: So when do you tend to get down to the 2 min. level?
edgy: Usually in the middle-game, maybe move 100 or so on average. Before move 80 means I've started an unreasonable fight and am trying to dig myself out of it.
So your average to that point is about 30 sec/move, and now you can maintain at 20 sec/move.
edgy: So what would be a reasonable time-limit for contemporary tournaments with byo-yomi or the Fischer variation I described? 30 minutes and 20 seconds would work if we could count on players getting to move 100 moves before entering byo-yomi. Is that reasonable? It's about my pace, and I'm usually among the last to finish but not the very last.
I don't know. With Fischer timing, 10 min./20 sec., 30 min./12 sec., and 45 min./6 sec. are roughly the same in regard to the expected length of a round. Maybe people would like the middle one.
For instance, if you, edgy, play at an average of 30 sec/move, after 90 moves (move #180) you will have 3:00 on your clock. To maintain that you will have to make a play every 12 sec. on average, but you are entering the endgame. If you play at your usual pace, at move 100 you will have 15:00 on the clock, and then at move 180 you will have 9:40 on your clock. At move 240 you will have 5:40 on your clock. You might like 30 min/12 sec. :-)
Well, 30 minutes / 12 seconds would probably be fine. I'm sure my time usage would become slower in the fuseki and I'd wind up in byo-yomi sooner, but even 12 seconds is enough time to actually see and react to the opppnent's move, at least face-to-face. It occurs to me that one of the big advantages of real-life play is that you see your opponent reaching toward a part of the board to play his stone, and you can start recalling what your ideas in that area were even before your clock gets started. On the 'net, the move just appears--boom! and your clock is ticking already--so a longer increment or byo-yomi period is desirable.
By the way, Znosko-Borovsky was writing in a time when Chess time limits were much slower than they are today, and nobody had come up with the bright idea of sudden-death time controls. These days, if you're in an opening you know--and serious players usually are--you better just play the next book move you remember and save your thinking time for later. (Bent Larsen wrote about time-trouble addict Pal Benko, and the opening he invented: "I think this is why he likes the Benko Gambit so much: for a Pawn, he gets to play 10 fast moves.")
The NY Go Center changed the time limit for their tournamnet this weekend to 18 minutes main time and a 15 second increment (it had been 10 minutes/20 seconds). This both allowed a little more time for reflection in the opening and sped up the games somewhat (I think all the games in each round finished in well under 2 hours).
I talked with Roy Laird (NYGC Vice-President) about the timing; he says they've been experimenting with various settings at the Princeton Go Club (if I remember correctly), and that increments under about 15 seconds seem to cause trouble when the players run short of time.
For me, the new time limit was fine (though unfortunately I did win one game on time, in a very complicated middlegame fight when we were both got very short on time). Possibly they could add a few minutes to the main time (at 15 seconds of delay) and still keep the round schedule brisk.
This is an interesting change. I have not used Fischer so much yet but I have been concentrating on timings like 10/10, 15/15, and 20/20. Not only are the names cool :-) but it is easy to remember as a rule of thumb that they deliver about 3 times the figure in minutes per person for a typical game (120 plays because it's easy to multiply). Most important though is that the balance between initial and increment seemed about right with 1/3 of the expected total up front and 2/3 to come through the rest of the game. When I read about the NY tournament originally (quoted in the background section) I wondered if the initial 10/20 used could have given people who are not used to Fischer the impression that they had less time than they really did because they only see 10 minutes on the clock at the beginning of the game. This could have contributed to the number who finished with more time on their clock than they started with.
The choice of 18/15 may be a little greedy on the TD's part. Note that 10/20 uses 60 minutes for 151 plays. The real equivalent with 18 initial would be 18/17 (60.5 minutes for 151 plays). The 2 second difference is means 5 minutes less over 151 plays (55.5 minutes versus 60.5). It also is a signficant difference to the player that ends up in time trouble, trying to survive a difficult fight on just the increment.
People used to playing on KGS might think the 15 second increment is not so bad. However, with the physical need to pick up and play a stone plus picking up captured stones, it is probably more like 10 seconds on KGS.
The choice of 18/15 may be a little greedy on the TD's part.
If by "a little greedy", you mean they were trying to cut down the total time for the game, then yes--that was an explicit aim of changing the time limit. My understanding is that this was in response to player feedback from the first two tournaments, and also some experiments at other clubs.
Previously, they had trouble getting the rounds started at 2 1/2 hour intervals; this time they managed a round about every two hours. The awards-giving ended before the last round would have in the previous events. The problem is that with 40 or 50 players, _somebody_ will probably play a 350-plus move monster almost every round.
I did suggest that they could probably add a few minutes to the main time.
People used to playing on KGS might think the 15 second increment is not so bad. However, with the physical need to pick up and play a stone plus picking up captured stones, it is probably more like 10 seconds on KGS.
I believe that if you're capturing more stones than you can scoop up one-handed, you're allowed to stop the clock while you're clearing them off. Also, it's an advantage to see where the opponent is reaching (you can start thinking about that part of the board again), plus you see the move before your clock starts running; on-line the move suddenly appears out of nowhere and your clock is already running as you recognize it. So I think it's six of one, half-dozen of the other.
![[Welcome to Sensei's Library!]](images/stone-hello.png "Sensei's Library")