EGF and Fischer

[Harleqin]

If we ask in what part of the game (fuseki, chuban, yose) a player can spend the largest block of thinking time on his next move, Fischer time emphasizes the later game stages more than the opening.

No, it does not. You are just presenting a scenario where the bonus has been chosen too long in comparison to the basic time. Your proposed setting 20/20 is just a bad choice; it would be better to use something like 32/15. Since you mention it, yes, wms has expressed the same misconception.

You afterwards went to the other extreme, setting an extremely short bonus in 45/4. That is also a bad choice, in my view, because 4 seconds per move is an inadequate pressure in comparison to the overall game length.

Then, you chose these words:

The dilemma is that the smaller the bonus, the fairer the players are treated in terms of when exactly they would like to invest most of their time, while on the other hand a smaller bonus will move the time system closer towards the problem faced in absolute time where unreasonable moves can be played to attempt and make the opponent lose on time (or make a really big mistake).

This is not a dilemma, it is a simple optimization. Make the maximum forced pace (determined by the bonus time) adequate in comparison to the overall thinking time. In my experience, 20/10, 30/15, 40/20 are good ratios.

I do not see a need to make this system more complicated.

[C. Blue]

If we ask in what part of the game (fuseki, chuban, yose) a player can spend the largest block of thinking time on his next move, Fischer time emphasizes the later game stages more than the opening.

No, it does not. You are just presenting a scenario where the bonus has been chosen too long in comparison to the basic time. Your proposed setting 20/20 is just a bad choice; it would be better to use something like 32/15. Since you mention it, yes, wms has expressed the same misconception.

Well, the way I see it is that the misconception is actually on your side, because all the time settings you suggested and dubbed optimization do not really seem to work around system-immanent issues by providing a sufficient general improvement over all aspects. (On a side note, in this regard I'd also like to point again to the quote about Fischer time experimenting given in the Redux text.) Of course if you are fine with a certain bias in either direction, because it fits your playing style, the settings will certainly work well for you.
Although complicating systems is certainly a disadvantage, a Fischer-byoyomi does actually seem easier and not more complicated in comparison to currently established time systems aka japanese and especially canadian byoyomi.

You afterwards went to the other extreme, setting an extremely short bonus in 45/4. That is also a bad choice, in my view, because 4 seconds per move is an inadequate pressure in comparison to the overall game length.

This value was picked for a very specific reason, as was explained in the text.

I'd actually discard "Bronstein Carry" in favour of Fischer byoyomi if it's just for the sake of an easy system. Maybe at least this one could be added to KGS, since it should widely follow wms' demands to a time system.

[Harleqin]

[...]because all the time settings you suggested and dubbed optimization do not really seem to work around system-immanent issues by providing a sufficient general improvement over all aspects.

Can you give an argument for this claim? Can you show why the balance between time management flexibility and maximum forced pace might not be achievable by simply balancing out basic and bonus time? Somewhere on the line (60/3 - 55/5 - 50/7 - 45/9 - 40/11 - 35/13 - 30/15 - 25/17 - 20/19 - 15/21 - 10/23 - 5/25) is an optimum. Why is it not good enough?

Can you show how your proposals could improve on that balance? Start with Fischer time 30/15, for example. How do you switch to Bronstein time and still keep the same overall game length?

(The report from the New York Go Centre is certainly an interesting anecdote, but you may note that they chose a rather large bonus again, compared to the game length. I do not see what kind of conclusion you want to draw from that.)

"Fischer-byoyomi" would certainly be an improvement over Canadian byoyomi (low hanging fruit), but it would still suffer from the same problem as all the other "main-time plus byoyomi" systems: you never know how many stones the players play during main time (they do not need to play a single stone), so the overall game length is very hard to guarantee. Such uncertainties always mean that a tournament director has to err preferably on the short side of time settings. In other words: the more uncertain the maximum game length is, the shorter the average game length must be to keep the tournament schedule. You may note that spilling also introduces such an uncertainty.

[C. Blue]

[...]because all the time settings you suggested and dubbed optimization do not really seem to work around system-immanent issues by providing a sufficient general improvement over all aspects.

Can you give an argument for this claim? Can you show why the balance between time management flexibility and maximum forced pace might not be achievable by simply balancing out basic and bonus time? Somewhere on the line (60/3 - 55/5 - 50/7 - 45/9 - 40/11 - 35/13 - 30/15 - 25/17 - 20/19 - 15/21 - 10/23 - 5/25) is an optimum. Why is it not good enough?

Can you show how your proposals could improve on that balance? Start with Fischer time 30/15, for example. How do you switch to Bronstein time and still keep the same overall game length?

The supposed imbalances of Fischer time system have already been laid out in detail and it'd actually been nice if you could back up the claim of how to "optimize" it with some concrete examples, in which is visible how either drawback of the time system does not surface very much. Anyway, some more concrete stuff here:
-It is pretty clear that Fischer time is actually by far the system that most easily allows game total duration (and thereby tournament round duration) planning.
-If you have at any stage of the game less than about 15 to 20 seconds for your next move, you might be very hardpressed to respond to nonsense invasions.
-If you gain more than about 10 seconds on each move, it becomes possible that you amass time rather quickly (ie surpassing your starting time pool) in a way that might deemphasize the opening a lot.

Now, if we want to go for perfect tournament round planning and make everything else low priority, then of course we need Fischer time.

With the other points in mind, being asked about what other time settings I could imagine, for example I wouldn't mind 30min/20s/40% "Bronstein Carry". I'd expect to take at least 3 seconds per move, so assumed the opening is really played very quickly, the time bonus per move would be around 7 seconds. Still, if I find myself in a situation where the opponent start playing odd invasion moves, I have at a guaranteed decent minimum of 20s to think about my move. If I play a move very quickly, I won't have to be too concerned about finding and thinking about another issue on the board, since I know I could at least get 40% of my remaining time added to my pool, and spend it after the opponent made his move and it's my turn again.
I will certainly admit of course, that this system will not allow for as smooth tournament planning as Fischer, which is and will be system-immanently the leader in this department. For the above example, each move would take 10s..20s. The median would probably be around 13-15s, but I don't have data to back this up. That we have a reduced interval of 10s..20s per move is however already a nice improvement over the usual 0s..20 when going for a maximum round duration without presenting the players too "edgy" a time system such as raw Bronstein.
An advantage of "Bronstein Carry" might be that the base time size and time period setting are rather independant, whereas compared for Fischer time, depending on the kind of optimization as was mentioned, those might need more attuning to each other.

About Fischer Byo-yomi, mainly I just wanted to emphasize that it should be fitting into KGS design (lots of main time to spend during fuseki etc.), since it'd be pretty neat to at least have one smooth type of overtime available.

(The report from the New York Go Centre is certainly an interesting anecdote, but you may note that they chose a rather large bonus again, compared to the game length. I do not see what kind of conclusion you want to draw from that.)

Well, if the text appears to you to not provide relevant information, I don't really want to stretch it, since I already said that it is just a side note anway.

[Harleqin]

[...]because all the time settings you suggested and dubbed optimization do not really seem to work around system-immanent issues by providing a sufficient general improvement over all aspects.

Can you give an argument for this claim? Can you show why the balance between time management flexibility and maximum forced pace might not be achievable by simply balancing out basic and bonus time? Somewhere on the line (60/3 - 55/5 - 50/7 - 45/9 - 40/11 - 35/13 - 30/15 - 25/17 - 20/19 - 15/21 - 10/23 - 5/25) is an optimum. Why is it not good enough?

Can you show how your proposals could improve on that balance? Start with Fischer time 30/15, for example. How do you switch to Bronstein time and still keep the same overall game length?

The supposed imbalances of Fischer time system have already been laid out in detail and it'd actually been nice if you could back up the claim of how to "optimize" it with some concrete examples, in which is visible how either drawback of the time system does not surface very much. Anyway, some more concrete stuff here:
-It is pretty clear that Fischer time is actually by far the system that most easily allows game total duration (and thereby tournament round duration) planning.
-If you have at any stage of the game less than about 15 to 20 seconds for your next move, you might be very hardpressed to respond to nonsense invasions.
-If you gain more than about 10 seconds on each move, it becomes possible that you amass time rather quickly (ie surpassing your starting time pool) in a way that might deemphasize the opening a lot.

First of all, you should not work with absolute numbers in the criteria, because the adequacy of a specific pressure is dependent on the overall game length. For example, in a blitz game that is meant to last for about half an hour, it is not sensible to call a maximum pace of 15 seconds per move "too fast", because that pace would imply an overall game length of at least an hour. On the other hand, in a five hour game, 20 seconds is quite fast, and one might find 30 seconds more adequate. You always have to compare to the average pace of the game.

So, let us take a 30/15 game of 240 moves. The average time per move is 30 seconds. The maximum pace, i.e. the miniumum forced time per move, is 15 seconds. That seems sensible to me: you cannot be forced to play more than twice as fast at any point as required on average. I think that this shows that inadequate time pressure is not possible.

Another way to see it is to divide the overall time into two parts: one that you have from the start, and one that will be unblocked bit by bit as you move. Of the same game, each player gets 30 minutes from the beginning, and another 30 minutes are unblocked at 15 seconds per move. That also seems sensible to me: half the time is free, half is rationed. In other words, you can spend at least half of your overall time on the first few moves. I think that this shows that the opening is not "deemphasized".

You claim that keeping a specific schedule is not the most important part of having a clock. However, in order to compare two timing systems at a respective setting, you need them to keep the same schedule. Otherwise, it is no question that the longer one likely has less potential pressure and more time for the opening.

[C. Blue]

-If you have at any stage of the game less than about 15 to 20 seconds for your next move, you might be very hardpressed to respond to nonsense invasions.
-If you gain more than about 10 seconds on each move, it becomes possible that you amass time rather quickly (ie surpassing your starting time pool) in a way that might deemphasize the opening a lot.

First of all, you should not work with absolute numbers in the criteria, because the adequacy of a specific pressure is dependent on the overall game length. For example, in a blitz game that is meant to last for about half an hour, it is not sensible to call a maximum pace of 15 seconds per move "too fast", because that pace would imply an overall game length of at least an hour. On the other hand, in a five hour game, 20 seconds is quite fast, and one might find 30 seconds more adequate. You always have to compare to the average pace of the game.

In the two cases with hard numbers specified we were looking at effects that are mostly independant of the base time (in the second case, they are not really independant per se, but forcing a certain relation, see below). If your <base time> is used up, the bonus gain per move dictates how well you can handle nonsense moves, no matter how large <base time> was. If you gain an amount of bonus time each move that is much greater than the time it takes you to make a "fast" move, then you will store up time quickly and easily end up with more time than you started with, depending on the bonus gain. In this second case, the actual relation of <base time> and <base time + piled up time> might be compensatable with high <base time> (so there is a relation), but here we see how the bonus time actually sort of forces you to increase the base time accordingly if you'd want to steer against the effect of piling up. Although regarding this second case we have settings that we can adjust and locally optimize, we realize that this in turn would reduce again our freedom to choose less base time for thinking (in the fuseki) while still providing a nonsense-invasion-proof game.

So, let us take a 30/15 game of 240 moves. The average time per move is 30 seconds. The maximum pace, i.e. the miniumum forced time per move, is 15 seconds. That seems sensible to me: you cannot be forced to play more than twice as fast at any point as required on average. I think that this shows that inadequate time pressure is not possible.

Regarding stockpiling time vs getting into trouble responding to nonsense moves, +15s bonus time is probably the best compromise of all settings. It's pretty fixed though, and will accordingly also strongly limit the minimum reasonable base time, in order to keep the relation between base time and bonus time fine. In my previous post I described the problems that arise if these restrictions are violated (by decreasing or increasing the bonus time). For example if we assume a 40/15 setting to make piling up time as little noticable as possible without getting too unreasonable, we couldn't downscale it anymore (for example to 20/8) without hitting the nonsense-move-issue. We could, however, upscale it. Or for example we might actually like 20 or 25 seconds much more to refute nonsense invasions, without being forced to grant a base time of around 60 minutes or similar (to lock out piling-up effects).

I'm actually still convinced that a mix of Bronstein and Fischer features is a solution that is most flexible and keeping the disadvantages of both at a minimum while utilizing more of their respective strengths.

You claim that keeping a specific schedule is not the most important part of having a clock. However, in order to compare two timing systems at a respective setting, you need them to keep the same schedule. Otherwise, it is no question that the longer one likely has less potential pressure and more time for the opening.

Certainly not. What I claimed was rather that Fischer time is the time system that allows to lay out a schedule most precisely (if the players play the maximum move number and use up the complete thinking time, that the Fischer calculations were based on), while other time systems might give somewhat more fluctuations around the expected average, and that the extent of those have to be weighed against possible game-play related advantages.

[Harleqin]

Hard limits do not make any sense here. How can you allow 15 seconds per move in a game of 240 moves that is scheduled for 45 minutes?

[C. Blue]

Indeed for lightning games we'd need lightning settings. Narrower time restrictions would naturally have the aspect of being safe against nonsense moves decline (same as the overall game quality, as is typical for blitz games). That aspect dropped, Fischer time works well, since it can operate here with low enough bonus times that avoid piling up and even provide some base time to spend at the beginning. (An alternative where we'd try to slightly improve the peak quality aspect might be 2:30/0:15/+35% for example.) Apart from theorizing it'd be nice if tournament holders would be open to some experimenting, so the go scene gets more practical experience. The sheer testing potential if one of our favourite Go servers *hint* provided alternative time systems is even more stunning.

[PeterHB]

Just to add my penny's worth.

I'm pleased that the EGF has made a positive move on Fischer timing.
Personally, I'm convinced that Fischer timing is a good thing ( the arguments have been made on SL & godiscussions.com in the past ).
Which exact intial & bonus time values to use is not obvious to me, but I see that as easily solvable by trusting the intial values suggested and then observing the results of trial and error.
I see the issue as transitional hurdles. Change is tough, particularly when people don't see the need for it, so decide it is unnecessary. ( A rational reaction, showing confidence in their own judgement.)

The 3 big hurdles are:
1) Having tournaments where the majority ( maybe even all ) of the clocks are electronic ones supporting Fischer timing.
2) The majority of people perceiving Fischer timing as preferable.
3) Using the change of perception achieved in (2) to persuade your national association to spend (invest/waste?) money on electronic clocks.

Oddly enough, all the online servers have electronic clocks for all players. Strange, but true. So if peoples' favourite server happened to support Fischer timing, they could try it. There's nothing like being allowed to try something for yourself to be able to make your own decision. What's the worst that could happen? They might like it. Scary thought, but ultimately harmless.

[Javaness]

Bronstein time will allow you never to lose because you were physically unable to place a stone on the board in the remaining time allocated to you.
This is what I like about Bronstein overtime. Fischer doesn't necessarily allow this to be the case.

Tournaments have time limits. I do not give a damn about players who are unable to manage their time properly. If they get into time trouble that is their own problem. I do not like seeing somebody lose because of Go's nature - that there are hundreds of legal moves that can still be played at the end of the game.

[Harleqin]

Bronstein time will allow you never to lose because you were physically unable to place a stone on the board in the remaining time allocated to you.
This is what I like about Bronstein overtime. Fischer doesn't necessarily allow this to be the case.

With Bronstein time, you always have at least the delay time for each single move. With Fischer time, you always have at least the bonus time for each single move. What is your point?

[Javaness]

Bronstein time will allow you never to lose because you were physically unable to place a stone on the board in the remaining time allocated to you.
This is what I like about Bronstein overtime. Fischer doesn't necessarily allow this to be the case.

With Bronstein time, you always have at least the delay time for each single move. With Fischer time, you always have at least the bonus time for each single move. What is your point?

That the bonus time need not be enough to make a move.

[Harleqin]

Bronstein time will allow you never to lose because you were physically unable to place a stone on the board in the remaining time allocated to you.
This is what I like about Bronstein overtime. Fischer doesn't necessarily allow this to be the case.

With Bronstein time, you always have at least the delay time for each single move. With Fischer time, you always have at least the bonus time for each single move. What is your point?

That the bonus time need not be enough to make a move.

Nor needs the time delay.

[Javaness]

Then you would not understand what Bronstein time was for.
I also do not believe that increments to time allowance are necessary. There is no reason to cushion a player's time. Time to think and time to play. That is simple and beautiful.