Forum for DragonTourney2006

Rules: Time [#118]

Back to forum     Back to page

New reply

Walin: Rules: Time (2005-10-31 22:45) [#299]

Moved here from main page.

trafalmadorian : In order to complete the tournament in 1 year, if we do 2 rounds as in 2005 and start in january, I suggest 3 months per player without byo-yomi as time settings (that would result in max 12 months duration).

Walin: I agree. Or use Canadian byo-yomi of say 25 moves / 5 days after 2 months.

X Re: Rules: Time (2005-11-03 14:55) [#302]

IanDavis: sounds good to me Re: Rules: Time (2005-11-12 23:29) [#458]

What about vacation though? Some people are bound to take some vacation. I suppose you could go by "it's unlikely both players use all their time", but maybe you should make it 80 days instead of 90 days, to compensate.

Just a thought. Doesn't really matter, I probably won't be playing, unfortunately.

trafalmadorian: ((no subject)) (2005-11-03 14:31) [#328]

why is the "clock runs on WE" flag set differently in the 2 leagues? (I think should be yes in both cases)

Walin: ((no subject)) (2005-11-03 19:49) [#335]

someone changed the other to no, and I didn't bother changing it back yet :)

reply Cheating through time-abuse (2005-11-14 10:22) [#474]

90 days a side means you should probably make 3 moves every 2 days if you want to be guaranteed of not timing out. (That's 135 moves per person for a game of length 270 moves.) (If your opponent played as fast/slow as you do, 3 moves every 4 days should be enough, but some opponents, e.g. me, who won't be playing though, may tend to move *very* often.)

Note also that no byo-yomi could mean if you're in time-trouble, your opponent may waste time playing dame or even filling in his own territory instead of passing, thereby forcing you over your time limit, and winning the game on time. In fact, any game can be won like this, by your opponent moving faster than you and simply refusing to pass. Of course we assume Go players aren't as badly-mannered as this, but maybe it makes sense to just mention "refereeing" in the tournament rules? Just officially allow a player to raise protest and claim a win from a referee (possibly a certain number of other players agreeing?)

axd: Re: Cheating through time-abuse (2005-11-14 14:31) [#476]

Typical DGS problem. Someone glued to his terminal will have an advantage over other people that use DGS just because the time scale is different. That's exactly the reason why they visit DGS, so why would you impose "traditional" time schemes?

This is why Fischer time should be used (e.g. 1 day - yes, this should be reasonable on DGS): at least one will get a decent piece of time whenever a move has been submitted, independent of the other player's possibilities.

uxs: ((no subject)) (2005-11-14 17:23) [#477]

I have played some games where I noticed as it was finishing that my opponent was almost out of time. Any time this happened, this was on a time setting other than fischer.

To illustrate just how good Fischer time is: I exclusively play Fischer time, with 21 days main time and 2 days extra per move. As of now, I have slightly less than 30 games running. Of those 30, there is one game where my opponent hasn't played since november 2nd and as a result only has a few days left. Everyone else who plays even only every few days has almost the 21 days of main time left. (There are 2 or 3 who are in vacation as well, they haven't moved for a while but also still have about 20 days left).

This wouldn't necessarily work with someone who, for example, only moves once every week, but then again I think that's not enough to participate in the tournament anyway.

When I propose a 21/+2 game like this, I sometimes get people telling me they're not sure they will be able to play enough to not run out of time. But I can comfortably say that you're not going to time out if you don't do something like just completely stop playing for several weeks. And even with a 2-week hiatus, it just takes 7 moves to get back up to 21 days in the bank.

So to summarize, I think Fischer time would be perfect: - Anyone who plays reasonably fast (= at least a few moves per week) is never going to time out, and will in fact have a cushion of a few weeks at all times - Shennanigans like playing on to run your opponent out of time are impossible - If someone stops playing (without turning on vacation time), they are going to time out in a reasonable amount of time. (3 months isn't reasonable, a few weeks is.)

That's why I'm advocating Fischer time with 21 days main time and 2 extra days per move. This is just how I play practically all my games, and it serves me very well. I don't have much experience with other settings, but I would imagine that anything from 1 to 3 weeks main time and 1 to 3 days extra time/move would work equally well. It suppose it all depends on how fast these games should actually be played.

I could try calculating some of this stuff, with an average amount of moves and so on. I'd just like to know that a time system like this won't be dismissed out of hand, just because it may be unfamiliar.

trafalmadorian: ((no subject)) (2005-11-14 23:28) [#478]

Well, personally, I like Fisher time a lot too and use it often. But my concern was to have the 2006 tournament start and end in 2006 in the spirit of having a yearly event. Look at the 2005 session (we played it with Fisher time): one player has had his 1st round finished for some time while another has finished only 3 games out of 14. And we have still to start the 2d round (when?). I think that if the tournament drags on for years, it will get boring. In a live tournament, you have also strong time constraints. And I'm sure that DGS players will be "fairplay" enough not to play dame when the opponent is out of time! Also, think of the new players who have to wait for a new tournament to start: 1 year is quite long already. But maybe you will say "man, this is DGS! If you are in such a hurry, go and play on KGS" ;-)

uxs: ((no subject)) (2005-11-16 00:23) [#484]

Ok, I get why you'd prefer a fixed time schedule then, not a flexible one. In that case, I think you should change it from the proposed 3 months to 2 months, because of vacation time. I now have 30 days (which is the maximum), so in the worst case scenario you will have 2 times 3 months, plus two times 30 days, which is too much.

Simulation: suppose 2 players who can only play a few hours in the evening (they work during the day), and who are in time zones 12 hours apart. Both players necessarily lose half a day for each move they make. (I'm not 100% sure, but I think saying "half a day" eliminates any confusion with sleep time, opposed to saying "12 hours".) A quick look at my finished games list on Dragon lets me know that the longest games I have are in the vicinity of 250 moves, so each player would use 125 days in this scenario. 3 months is 90 days, which is not enough for this scenario.

Now in my case, I'm able to move both at noon and at pretty much any time in the evening, so I'm pretty confident I wouldn't time out in any game I'm in. I don't know about anybody else though.

trafalmadorian: ((no subject)) (2005-11-16 09:49) [#490]

You have a point.

Maybe my proposal would simply not work on DGS, i.e. maybe it's impossible to play a 2-rounds tournament in 1 year ...

Walin: ((no subject)) (2005-11-16 11:49) [#492]

Wouldn't they use only half of 125 days, if they use around 12 hours per move. That would mean 125 days for a whole 250-move game with 180 days available (90 days for both players).

uxs: time = complicated + more calculations (2005-11-16 17:55) [#494]

Ugh. I had a reply almost complete typed in when I got distracted for a few hours and then I shut down Opera. :-(

Well, I think it was interesting so here it is again:

Imagine a scenario with users A and B. They play a game with 10 days each. A and B are in different time zones, and more importantly their sleeping schedules are different: A has 0 to 9, B has 12 to 21. (Both are for our convenience in GMT.)

The game starts at 0 hours exactly, with A to start. He plays his first move at 12. His clock will have ran from 9 to 12, which equals 3 hours. His clock is now at 9 days, 12 hours. (NOT 9 days, 21 hours, because DGS days are 15 hours plus 9 hours sleeping time. This is a fact: if you don't believe it go check all your games on DGS, none of them will have X days plus Y hours with Y greater than or equal to 15.)

B also waits until the last moment to play, at midnight. Her clock will have also ran for 3 hours (21 to 0), and will also be at 9 days and 12 hours.

Next day, same scenario: both clocks run for 3 hours, both players make a move, A at 12 and B at 0. Clocks are then at 9 days, 9 hours.

Day 3: 9 days, 6 hours. Day 4, 9 days, 3 hours. Finally, at the end of day 5, both players' clocks are at 9 days exactly. Both have moved 5 times. Both players have now used 1 day of their allotted time. However, in real time 5 days have passed.

If we go on like this, after 50 days of actual time, both players will have moved 50 times and their clocks will be at 0.

Conclusion: in this example, 20 days of clock time equals 50 days of real time for 100 moves. So for a 250-move game, we would need 50 days of clock time, or 25 for each player.

(Edit: In actual time, this would take 125 days for a 250-move game. Without vacation time. So we have some time to spare for 2 rounds per year.)

This wasn't what I expected for a scenario like this. I wonder how other scenarios would turn out.

Edit 2: Let's find out.

Everything the same, except both players have their sleeping time scheduled at 0 to 9. To make it fair, we say both players use an equal amount of time to make their move, which is 7+1/2 hours.

After day 1, both players have their clocks as 9 days, 7+1/2 hours.

After the second day, both are at 9 days exactly.

So this means that in 2 days of actual time, both players have moved twice (= 4 in total), and have used one day of clock time each (= 2 in total.)

In other words, for 250 days, they logically need 125 days of real time (both moving once per day), but this time they would need 125 days of clock time as well. This is 62.5 days for each player, or more than double the time for the first scenario.

Edit 2: =============

Some more comments about scenarios 1 and 2. I'm assuming that both players behave like this:

Their sleep schedule is indeed when they sleep. When they wake up, they go to work/school/whatever, so they can't play. In the evening, they have some time to play.

Taking my own example, I would sleep from 23 to 8, then go to work and return home at about 18 hours. Then I can play until 23.

In scenario 1, A would be able to play from 19 to 0, B from 7 to 12. Their schedules don't overlap, so the scenario is essentially realistic.

For scenario 2 however, their schedules completely overlap. So one can assume that they each move, say, about 3 times per day. We also assume that another player has the last move every day. So let's make it 5 moves per day in total. After 2 days, they would have moved a total of 10 times, using 2 days of total clock time. For a 250-move game, they'd need 50 days of actual time, and more importantly, 50 days of total clock time, so 25 each. These are the same time settings as for scenario 1.

Variation: the same player has the last move every day. This means that the next day, the other player's clock is running for 10 hours when both players are unable to play. The remaining 5 hours are split between both players. A uses 2.5 hours per day, B uses 12.5. In 2 days, A uses 5, B 25. In 6 days, A has used 1 day, B has used 5. For 250 moves, they would still need 50 days of actual time, but those 50 would be split in 8.33 for player A, and 41.66 for player B. So using 25 days for each player, B would time out. In a scenario where A would usually go to bed only 1 or 2 hours later than B (for example, only slightly differing timezones), this is entirely plausible.

trafalmadorian: ((no subject)) (2005-11-16 17:53) [#498]

So, the conclusion of your nice scientific study would be that in those realistic scenarios, 3 months per player would do.

The very worst case is when you play against a robot (or someone behaving like a bot ;-) i.e. he always plays just after you (or let's say 1 hour after you - when you have left) and has always the last word : his clock nearly doesn't move.

Then with you 90 days, you need sometimes to play several moves a day. Those who log only once a day or even less may have some games timed-out.

Degan: Re: sudden-death time (2005-12-23 03:41) [#780]

I hope your experiment in so many sudden-death games goes well! Personally, I cannot be certain to play every day without fail, and my times to play moves is usually limitted to the evening. So, I cannot play with your proposed time formats.

I think you will find many more of the games being won on time by percentage than the "2005" Tourney. Of course, anyone in the game(s) approaches it with awareness of the time constraints, so it is a completely valid win to take advantage of greater access to playing and tracking your opponents playing behavior such to put them at a time disadvantage.

I am uncertain of the underlying reason behind wanting to set time constraints to complete both rounds in one year, could you elaborate?

I intentionally named the other event "2005" to distinguish it from those that would come after. It is quite possible that DragonTourney2005 will not be complete in 2006. Such is the nature of the tails of normal behavior.

With all sincerity, I hope that DragonTourney2006 goes great!

trafalmadorian: Re: sudden-death time (2005-12-24 00:08) [#784]

Well, I was the one who initially proposed tighter time settings for 2006. My point is partly made in my post from Nov 14th above, but to sum-up: if the tournament drags on for too long, I simply will get bored and lose interest. In fact I'm beginning to get this feeling for the 2005 tournament now. I admit this is totally subjective, and didn't mean to impose my view. I just wanted to get some feedback from other players, to see if they thought the same, before a decision is taken on the rules.

GoJaC: Re: sudden-death time (2005-12-24 09:55) [#785]

How about using fischer time, but not +1 day per move, rather +12 hours or even +10 hours (bear in mind a day is 16 hours, so +8 hours would be half a day)? Just another option to consider, I think the suggested 90 days absolute time might be worth trying out as well, to see how it works.

uxs: Re: finishing on time ? (2005-12-23 15:49) [#782]

Quick reply after a cursory glance over my own conclusions:

If you set the time limits to 90 days each, no extra time:

  • For players that have completely seperate sleeping schedules, the maximum actual time that can elapse is 2.5 times the clock time. This would be 540 (!) days.

This seems a lot, and it is, but it's correct. After all, during most of the day the clock is stopped because one of the players is sleeping. Still, this is an absolutely worst-case scenario.

  • For players with completely equal schedules, actual time is equal to clock time: 180 days in total.

Both of these ignore vacation time. Quick calculation: 2.5 days per month extra means 15 days in six months. When both players start with 30 days (the maximum), and take it right away, they can both take 45 days of vacation. (This is for the second scenario: in the first one, they can take 75. Both of them.) (Plus the time they spend in vacation also probably earns them extra vacation time, so it would be even more.)

Even shorter conclusion: I think that if you want to finish 2 rounds with 90 days per player, you're going to be rather disappointed. I'd cut it to 60 days each, or even 45.

Walin: sudden-death time (2005-12-23 16:40) [#781]

I'm not against a little byo-yomi. Something along one move every other day after three months main time would work for me. But I'm unsure which would be better for that, Japanese or Canadian byoyomi.

It's just that there was really not much opposition to current time system, so it has been up like that and is not likely to change.

There's always DragonTourney2007+, one of these years we will hit the right combination :)

-- Matti

Back to forum     Back to page

New reply

Forum for DragonTourney2006
RecentChanges · StartingPoints · About
Edit page ·Search · Related · Page info · Latest diff
[Welcome to Sensei's Library!]
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Login / Prefs
Sensei's Library