Hello OGS!
The new Play page and quick match system has been up for a few weeks now and now that most of you have had a chance to use it we’d love some feedback, particularly around the time settings. Below the polls there are some stats from the past couple of weeks for those that are interested.
The Fischer clock
Should we keep the Fischer clock options on the auto match portion of the Play page? (These are the options on the left, such as 30s+3s
, 5m+5s
, etc.). This only applies to live games, not correspondence games which will definitely stay Fischer.)
- Keep the Fischer clock options
- Remove the Fischer clock options
Blitz settings (the fastest clock settings)
Is the 30+3s
fischer blitz setting too fast?
- Yes
30+3s
is too fast - No it’s not too fast
Is the 30+5x10s
byo-yomi blitz setting too fast?
- Yes
30+5x10s
is too fast - No it’s not too fast
Overall blitz settings:
- The blitz settings are good
- I would like to play blitz games, but the options are too fast
- I’m not interested in blitz games
Rapid Fischer clock settings (the middle clock settings)
- The rapid
+5s
Fischer clock settings are good - I want to play with a Fischer clock, but the 5 second increment is too fast
- I prefer to play with using a byo-yomi clock.
Rapid byo-yomi clock settings
- The
5x30s
byo-yomi overtime is good for the rapid clock setting - I would prefer a faster byo-yomi clock for the rapid clock setting for all board sizes
- I would prefer a faster byo-yomi clock for
9x9
but not for19x19
Long Fischer clock settings
- The long +10s Fischer clock settings are good
- I want to play with a Fischer clock, but the 10 second increment is too fast
- I prefer to play with using a byo-yomi clock.
Thanks to everyone who takes the time to vote!
Quick match game statistics from December
Total quick match games: 49324
Blitz: 2.0%
Rapid: 59.3%
Long: 38.4%
9x9: 32.7%
13x13: 10.7%
19x19: 56.4%
It should be noted that the default option is 9x9 rapid, however the ~60/40 split between rapid and live games is fairly consistent in 19x19 as well, so the new rapid settings seem popular.
Blitz games seem very unpopular though. I’m hoping the poll helps us figure out if the proposed times are simply too fast, or if the whole category is undesired.
Fischer / byo-yomi preference
Rapid:
9x9:
Fischer: 73.3%
Byo-yomi: 26.7%
19x19:
Fischer: 57.4%
Byo-yomi: 42.6%
Long:
9x9:
Fischer: 29.6%
Byo-yomi: 70.7%
19x19:
Fischer: 25.2%
Byo-yomi: 74.8%
Since 9x9 rapid Fischer is the default setting, we expect a significant bias towards Fischer for the rapid category in general, particularly the 9x9 rapid category. In this regard long 19x19 games are probably the most interesting in terms of trying to figure out how the general player pool is feeling about byo-yomi vs Fischer. Given byo-yomi is the incumbent time control system, the fact that we see ~25% games being played under the Fischer clock is a pretty strong signal that there is some significant interest in playing with a Fischer clock.
Game durations & timeout rate
Blitz:
size | time_control | clock | duration | timeout_rate | samples |
---|---|---|---|---|---|
9x9 | fischer | 30s + 3s | 02:38 | 30.9% | 1029 |
9x9 | byoyomi | 30s + 5x10s | 03:11 | 10.5% | 639 |
19x19 | fischer | 30s + 3s | 07:05 | 51.9% | 395 |
19x19 | byoyomi | 30s + 5x10s | 09:55 | 21.2% | 709 |
Rapid:
size | time_control | clock | duration | timeout_rate | samples |
---|---|---|---|---|---|
9x9 | fischer | 2m + 5s | 04:30 | 8.2% | 19506 |
9x9 | byoyomi | 2m + 5x30s | 05:09 | 1.5% | 5990 |
19x19 | fischer | 5m + 5s | 15:53 | 14.7% | 9219 |
19x19 | byoyomi | 5m + 5x30s | 19:08 | 5.2% | 6798 |
Long:
size | time_control | clock | duration | timeout_rate | samples |
---|---|---|---|---|---|
9x9 | fischer | 3m + 10s | 06:22 | 5.7% | 2817 |
9x9 | byoyomi | 5m + 5x30s | 06:39 | 2.6% | 6148 |
19x19 | fischer | 10m + 10s | 25:14 | 5.1% | 2912 |
19x19 | byoyomi | 20m + 5x30s | 26:40 | 3.1% | 8697 |
When initially picking what clock settings to use for each of the categories I did a lot of analysis on games in our database to figure out what Fischer time settings would result in games that lasted about as long as their byo-yomi counterparts, erring on less time if a value seemed to line up well with respect to the other settings. In that respect about half of the selections are pretty decent while the other half came up about shy, which is pretty much what we aimed for. However what wasn’t as measurable at the time was the timeout rate as people playing open challenges using similar settings were probably more prepared to manage their time accordingly for the given Fischer settings they were using.
Given the current data I think the Fischer times are probably too fast, resulting in too many timeouts. I would expect the timeout rate for Fischer games to be a bit higher than that of their byo-yomi counterparts as they tend to bring in the 90th percentile game durations significantly, but I don’t think we want to be seeing 3x the timeout rate. Additionally, the 9x9 byo-yomi time seems to be too long as on average folks aren’t experiencing much if any time pressure.