I like the simplicity of Fischer time and the fact that you don’t have to evenly spread your time usage to make the most of it. But if I play a Normal quick-match Fischer game, it seems to take twice as long as a Byo-yomi game. It doesn’t seem like 30 second fischer increment and 30 second byo-yomi should be treated as equivalent. Wouldn’t it be ideal if the quick match “Normal” and “Blitz” games resulted in similar total length of games regardless of which type of time control is used?
To accomplish that, I assume the normal Fischer increment would need to be closer to 20 seconds, but the developers should have access to data that they could base this decision off of, right? I think it would make sense to standardize other time controls according to the average length of a byo-yomi game right now, since that’s the default and most common.
I think you are making a false assumption. IIRC, the default settings in automatch were not devised to be “equivalent.” They were chosen because they were the most common or popular settings in each category. Indeed, I seem to recall one or more polls being taken at the time they were implemented.
If that’s an incorrect assumption, then I guess I’d like to propose that they should be equivalent. Them not being equivalent results in user experience problems. For one, the buttons say things like “~30 seconds per move” on them which becomes grossly inaccurate if one time control is twice as long as another. And secondly, the default setting is to have no preference for the time control. It’s not very welcoming to a new player to have the default behavior to result in games that take vastly different amounts of time. And I bet most people don’t know about those settings until they get confused about why a game is taking so much longer than their previous games.
I’d actually argue that in general, having the default settings just be “prefer X” is pretty confusing to players. Even though I prefer Chinese rules and Fischer time (if it wasn’t twice as long), I think it’d make more sense to have the defaults be Require Japanese Rules and Require Byo-Yomi Time, and so that it would only deviate from those when both players have changed their settings.
The time settings definitely should aim for equivalence in actual game length. A 19x19 blitz player is looking for a roughly 15-20 minute game, not a 15-45 minute game. Such inconsistency makes the system practically unusable. Time settings for automatch in general are all over the place and really hurt its (already questionable) usefulness.
However you’re unlikely to see any improvements in the automatch system as it has been essentially given up on and abandoned for years now.
Japanese Byo-Yomi: Clock starts with 20 minutes main time, followed by 5 30 seconds periods.
and
Fischer: Clock starts with 3 minutes and increments by 30 seconds per move up to a maximum of 5 minutes.
For a 240 move game say, 120 moves by each player, and they spend all of their time
Japanese: 20*60s+4*30s+30s*120 = 4920s ~40s per move. 164min game max = 2.7 hours
Fischer: 3*60s + 120*30s = 3780s ~ 31.5s per move = 2.1 hours game max
For blitz I think we’re comparing
Japanese Byo-Yomi: Clock starts with 5 minutes main time, followed by 5 10 seconds periods.
Fischer: Clock starts with 30 seconds and increments by 10 seconds per move up to a maximum of 1 minute.
So similarly for a 240 move game
Japanese: 5*60s+4*10s+10s*120 = 1540s ~12s per move. 51min game max ish for this length of game
Fischer: 30s + 120*10s = 1230s ~ 10.25s per move = 41 mins game max
Those would be for long ish games going to scoring. Obviously shorter games and resignations will be shorter on time.
If you look at the 10s increments and byo-yomi, the game is long just because Go is a long game. 10s per move is quite fast, but 240 moves is 240 moves. It’s just going to be pushed towards 40 mins no matter what.
I have some issues with your point. You’re comparing these time controles as though the players would use them to the fullest, when in practice it is simply not the case. First, nobody plays 0 moves in the main time before the byo yomi, and once in byo yomi, 90% of moves are done in 15 seconds. However, with a Fischer control that adds 30s at each move, spending 30s at every move is perfectly possible.
The empirical result is that Fischer games will last much longer than byo yomi games. (this is why the EGF, when requiring specific time controls in tournaments, has different scales for fischer and traditional overtimes.), and as such, the point of the post still stands.
I think this is an issue, and OGS’s developpers should fix this (it’s not really a hard change, is it? Just remove the possibiltity of choosing the time control when using auto-match, and set byo yomi as default.)
Well I think it’s important to know what in theory the upper bound is. In fischer it doesn’t actually matter because it’s all main time, your incrememnt gets added in.
Probably more practically people might run out of main time in byo-yomi by mid game, unless they’re managing their time quite well.
So sure the 240 game could be more like 5*60s+4*10s+10s*60~940s, say if half the moves are spent in main time or about 31 ish minutes of a game for blitz
For the live version it could be like 20*60s+4*30s+30s*60~940s~3120s or 104mins for a game ish. Both players using their main time is going to make it a 40 minute game anyway.
As in even if people play faster on average, I feel it’s important to know how long the game you’re signing up to “could be”, because people can use their time if they want to.
Maybe most live games will still finish in like 30mins because people will play fast and then think only when it gets complicated or they need to.
It’s probably not a hard change, but it might not be a change everyone wants. Some people prefer or want to play fischer and dislike byo-yomi.
I kind of prefer byo-yomi, less complicated time management, and I do agree about what you’re saying, in that you can theoretically use all the available time in fischer, and so it probably will be longer if both players do decide to use all of their time.
Spoiler alert: they don’t. Which is why you can’t meaningfully compare time controls by their theoretical maximum duration. The estimation overshoots greatly, especially for non-accumulating systems like byo-yomi. You’d either use something like EGF’s adjusted time formulas
Adjusted time (TA) calculation:
Japanese Time Control (Byoyomi)
TA = basic time + time equivalent to 45 moves
e.g.: basic time: 60 minutes, byoyomi: 30 seconds per move:
60 + ((45 * 30) / 60) = 82.5 minutes
Canadian Time Control
TA = basic time + time equivalent to 60 moves
e.g.: basic time: 60 minutes, byoyomi: 20 moves in 10 minutes:
60 + ((10 / 20) x 60) = 90 minutes
Fischer Time
TA = basic time + time equivalent to 120 moves
e.g.: basic time: 45 minutes, increment per move: 15 seconds:
45 + ((120 x 15) / 60) = 75 minutes
Absolute time (Sudden death)
TA = basic time
or fit your own based on empirical observed durations and their distributions. As a rule of thumb though, an equal duration game with Fischer corresponds to about half of that of byo-yomi (i.e. +15s increments = 30s byo-yomi periods).
As I’ve understood it the formula tries to find the absolute time equivalent. That is, the current byo-yomi corresponds to about 12.5 minutes of absolute time for each player.
If you wanted to use the formula, you’d select a certain time setting in one setting and then use it to determine the others.
The current byo-yomi would be TA to 5mins+45*10s/60 = 12.5 mins.
The equivalent Fischer time with similar 5 minutes main time would correspond to about +4s per move increments.
Edit: The equation is obviously underdetermined so you’d need to decide on some balance of main time and overtime.
It might be hard to convince people to play with a four second increment even if they like fast games.
My gut feeling is that what it feels like to be in your last byo-yomi or living off increment in Fischer is probably more important than the potential for slightly longer games. As in will it turn into a click fest to time out the opponent with 4s increment? Maybe you can bank a bit of time with 4s.
It’s probably important to pick a setting many people will play otherwise it might as well be removed as per
On the other hand that’s the catch with Fischer, isn’t it? You have more freedom in how to spend your time, which also means you have more freedom in shooting yourself in the foot by over-spending it.
Not that I’d mind giving +5s or alternatively lowering main time to 1 minute and giving +6/7s per move. Experienced Fischer players should be able to manage their time bank either way.
There are of course ton of different situations where the time managment is completely different with byo-yomi and Fisher
When you’re fighting a ko, with fischer you want to take the ko asap on your turn in order to gain more time, but with byo-yomi it makes sense to use 28-29 seconds on counting the ko threats before taking the ko on your turn. Of course most people dont actually do this, we tend to play quickly because we’re in byo-yomi! xD
If your initial time is same or close to the cap, you prolly “waste” big part of your Fischer increments early of the game. If the players keep playing at a pace thats faster than the increment, they can pretty much finish the game while ~always having the full cap at their use.
Short byo-yomi main time usually forced players into byo-yomis, thus creating constant “urgency” for each move. Tho many people use long byo-yomi settings almost like absolute time, and finish their games without ~ever using all of their main time.
btw i like using 15s fischer with my live games, idk if there was ever a vote on those defaults or if just missed that… Automatch defaults are quite long, even the corr settings are capped at 7d with weekend pauses, i think many people realistically speaking wont use even 50% of the time they could use
://
Is an inconsistency when chosing, a real problem?
What counts first is how players like to use each setting, maybe fischer time is more attractive with setting generating a longer global time?
Your opinion is interesting but there were so many threads about automatch settings with even a test once on the beta server, it’s difficult to find the consensus anyway. One more stone in the debates.
I don’t see any more simplicity in the fact that i have to play each move under a time pressure. I may lose on time more easely if suddenly i focuse more and forget the clock running… I may play too quickly trying to capitalize too. A player used to friendly games without clock may feel more confortable with the classic main +byoyomi. Anyway there is no perfect system to me, i can play both (and canadian too)
I remember a vote for the blitz setting on the main+byoyomi and fischer timing were considered too. No memory on a poll for live settings
To use “prefer” instead of “require” is linked with how many players are in the system.
Same as how many different categories you want to establish, splitting the pool and reducing the chance (longer waiting time) to create a game.
Custom challenge is more fitted for specific requirements (and quite popular on OGS)
[Edit: it depends on the given starting time + incident (Byo-Yomi) which time settings I’d prefer.]
Only for ultraslow time settings. If the main time is so much, it’s basically absolute time settings for me, than yes. If I suspect my normal speed will bring me into Byo-Yomi, I’m screwed.
As a newcomer to timed games, I feel more comfortable with Fischer time. It gives me a sense for how I manage my time (“You can spend 2:30 to think about your next complicated move without overstretching your time budget.”), while Byo-Yomi feels like “so, you were thinking for too long 30 moves earlier? You time is up. Here are a few extra seconds to end the game.”
When you reach the Byo-Yomi periods, Fischer timing would be more comfortable since it let you accumulate time to spend on a move later.
While in main time, you can potentially spend most of your time on a few moves with Byo-Yomi, while Fischer (with cap) forces you to spend the time more evenly.
At what move (on 19x19) should I’ve spent my main time for optimal usage of my time?
I’m not sure we really know what this is. Certainly if the currently settings were based on some kind of vote previously preferences might have changed by now…
For me the main thing that people probably want is fast matching. I think there are more threads/discussion about that than about the ins and outs of time control.
My opinion is that automatch should a single rule system (probably Chinese) and a single clock type (Fischer? But I don’t really know) and a few lengths like blitz, rapid, relaxed live and correspondence and that’s it. That would help with faster matching which is the top priority.
If people want something different then custom game is available.
We have explained the many downsides to not having equivalence. If you disagree with those, then it’d help the discussion if you were more specific about your disagreements. In many cases, there may be ways to preserve the community preferences that you’re referring to without having to stick to the status quo.
The current auto-match is a very poor user experience if someone has not taken the time to understand and modify their settings. Good design can accommodate both simplicity and flexibility in many cases, but simplicity is generally higher priority for the overall user experience. People reading the forums already represent a minority of the player base. Even a poll here to determine preferences should merely be used as a datapoint, not as something that should override good design. Design by democracy has never been successful at any sort of scale. If we want OGS to grow, it’s imperative that design flaws like this get addressed.
As has been stated, the easiest way to still allow flexible time controls with the current default preferences, would be to standardize the total game time across the different types of time controls. In the long run, the current behavior of a new player getting games with both different scoring rules and different time controls is not a good default experience, but I understand that at the current scale it’d mean anyone who really wants non-default rules or time controls would not be able to find games quickly. I still think that’s an appropriate sacrifice to make to improve the new user experience, but that’s not what I’m proposing in this thread right now. I just want to roughly standardize the total game time. If that means that typical Fischer games are still 10-20% longer because we think that 20 seconds increment feels more forgiving that 15 seconds (just throwing out random numbers as examples), that would be a fine compromise. But the current disparity is way too much.