This is already the case. Players can only receive 3 challenges and send 3 challenges, so it’s already pretty low. When you’re in the top tier it goes up to 4 received, and very top 5, but for the majority of folks it’s only 3 incoming and 3 outgoing challenges.
Thanks for the reply. I was thinking more along the lines of 2 sent and 2 received per ladder. When you’re on multiple ladders (not to mention tournaments), the number of games can quickly add up. I should clarify that what I meant is that having less challenges makes sense if the time settings require people to act faster.
There’s very little point to changing the maximum accumulated time. The effect on the duration of long games is negligible (in the order of 1%). I’ve seen the same discussion a lot on another site too, and most people simply fail to understand that the impact is negligible. 3 days instead of 7 days feels like it should speed up the game significantly, but in effect it just means a 19x19 game must be finished in roughly 270 days instead of 274 days.
If you want to reduce the maximum game time significantly, you would have to reduce the increment. Reducing that from 1 day to 12 hours would mean games must be finished within about 135 days instead of 270 days.
The downside to reducing the increment is that it can be very unfair due to time zones. If two players can only play in the evening, and one of them is 6 hours ahead the other, he will always lose all his games against the other player (he moves, goed to bed a few hours later, his opponent goes online, moves about 6 hours after him, and he doesn’t get back online until 18 hours after that, so he loses 6 hours per move).
Effectively, an increment of 12 hours means that people that can only play evenings (or only mornings, or whatever) are effectively not welcome on the ladder.
Current evidence suggests otherwise, see my post above
The downside to reducing the increment is that it can be very unfair
due to time zones. If two players can only play in the evening, and one
of them is 6 hours ahead the other, he will always lose all his games
against the other player (he moves, goed to bed a few hours later, his
opponent goes online, moves about 6 hours after him, and he doesn’t get
back online until 18 hours after that, so he loses 6 hours per move).
Effectively, an increment of 12 hours means that people that can only
play evenings (or only mornings, or whatever) are effectively not
welcome on the ladder.
Fully agree with these concerns. Even with 16 hour increments, anything less than a 8 hour window of periodic checks on ladder games could hypothetically cause you to time out if you’re playing against someone who has just the right kind of mis-matching schedule. 24 hour increments guarantees that one move per day keeps you in the game regardless of timezone differences.
It’s my perception of this site that it specializes in correspondance games and a casual atmosphere. In that vein, I personally consider it important that “daily activity” is enough to avoid timeouts.
I was just about to quote that post myself. By the by, what’s the percentage of games timed out over each of those periods? I wouldn’t be surprised if it increased at least a little bit with a smaller maximum, but going by the number of games completed, they seem fairly similar.
Your hunch is correct, it does result in more timeouts - in that same time interval that i ran the previous numbers in, we’ve seen 158 non-annulled timeouts with the new settings vs 126 timeouts with the old settings, so about a 25% increase
This data is looking at only the faster games (ones that ended in less than 36 days). How has the max setting change affected the slowest games?
I’m talking about long games, not average games. I think most people are fine with games that last either 7 days or 11 days, but are mostly annoyed by games that last many months.
Also, I wonder how much effect there has been from the fact that the last 36 days have been in the middle of summer holiday. Ideally, you would want to compare to the same period last year.
I wonder it would be beneficial to give each player a daily pause for a time scheduled by the player. Live games obviously shouldn’t be affected, but correspondence game clocks could be put on hold during (for example) late night hours specific to a player’s time zone. If the pause is 8 hours and the players are 8 hours away from each other, then that would result in a 16 hour pause every day. In contrast, players in the same time zone might have pause settings that overlap; because both would presumably be equally capable of making a move in the same unpaused time frame of 16 hours, I think this is fair.
With such a mechanism in place, so long as the time increment is large enough to be sufficient for players from the same time zone, players from different time zones should have no problem with the same increment.
[quote=“Jamada, post:50, topic:1499”]
correspondence game clocks could be put on hold during (for example) late night hours specific to a player’s time zone.
[/quote]I like this; and it would be even better, I think, if ppl could set their personal regular 8 AFK hours. (I for one am a night owl.)
Ah yes, worst case they’re basically the same…
Over the same time last year there were only 246 ladder games that completed within the same interval (this was “The old OGS”), and the average completion time was 11 days 6 hours 1 minute and 3 seconds for those games.
And for the fun of it here are average move times of games that completed within the 36 day window during these time periods:
2014 Jan/Feb: 13 days 6 hours 19 minutes
2014 Mar/Apr: 13 days 0 hours 36 minutes
2014 May/Jun: 11 days 13 hours 57 minutes
2014 Jul/Aug: 11 days 18 hours 1 minutes (only 276 samples for 7d)
So it’d appear you’re intuition about games progressing quicker in the summer months is correct, though the “most recent” games that I’m pulling from reach back to 2014-06-13, so that’s still in the summer months, so I don’t think we’re seeing too much bias from that.
Does that mean games which only took up to 36 days? So, you’re only looking at the shorter games then?
Also, any statistics on the 3 day max ladder games are going to be biased, since the longer games that were started since the change haven’t even finished yet.
This is the point that I’ve been making about reducing the max. It does not actually force games to play out faster, and the slowest games still move at the same pace.
So if you read my process, I only considered games which were completed within a 36 day window for both 3d and 7d games, this eliminates the bias you’re referring to.
We only have 36 (now 37) days of real data to work with
If we want to consider non ladder games as valid, we can go back and look at games from earlier this year:
Games from January that have completed by now (so within 196 days):
3d+1d: 10 days 13 hours 59 minutes
7d+1d: 29 days 22 hours 15 minutes
And now you can speculate as to why there is such a variation, is it only the time controls? is it the fact that 7d+1d is a default value where as by the time you are setting 3d+1d manually you are going to be a more active player? It introduces a number of possible (and I’d say probable) selection biases which I feel are likely to make 3d+1d look more favorable than we can expect to see on the ladders.
So yes, we are working with the data we have available and trying to compare apples to apples as best as we can.
Ah great then we’re in agreement. The max (which isn’t a big factor since it doesn’t dictate the expected pace of the ladder as a whole, as these games are “rare”) is expected to be the same. The average game completion time (the dominating factor in the expected pace of the ladder) is significantly sped up with the reduction of the maximum time. Hence, the reduction to 3d from 7d notably speeds up the expected rate at which players can progress through the ladder, which was the objective of the last change, and thus was successful in that respect. Now we can move on to the next real questions, “Was it enough”, “Was it too much”, and “How can we make it even better”. Hooray for progress!
Note: the following stats are probably useless
But for the curious, I ran some stats on the available 3d + 12h and 7d + 12h games that have been run… (I didn’t find any fischer 16h in the database ) There really isn’t enough data for these numbers to be useful I don’t think, but here they are anyways
From Jan 1 to May 1
50 fischer games were found to have a max time of 7 days with 12 hour increments, their average move time was 32 days 21 hours 52 minutes
67 fischer games were found to have a max time of 3 days with 12 hour increments, their average move time was 19 days 17 hours 2 minutes
As usual these are only considering games that have completed without timeouts and without being annulled. I have no answer as to why those numbers are higher than their 1 day counterparts other than not having enough samples…
I also re-ran that same query filtering out all the games that took longer than 36 days and came up with
33 games were 7d + 12h, average of 13 days 10h
60 games were 3d + 12h, average of 12 days 23 h
Which is interesting since not only is the 7d+12h about the same as the measured 7d+1d we did before, but so is the 3d+12h! … but again, I don’t think we actually have enough samples to draw any real conclusions from this… but still kinda interesting. If we were to simply look at this data we’d have to draw the conclusion that neither the max time nor the time increment matter one bit!
At the same time, as you said:
So I would think the question is, is a decrease of the average short game thinking time worth the extra timeouts? Timeouts drop a player from the ladder altogether, it seems, which would significantly impede their rate of progress through the ladder…
A valid question, one I’d love to see discussed!
I would say with some critical modifications to the ladder rank forfeiture by timeout mechanism, the extra timeouts won’t be a serious problem.
@yebellz had a good idea in this thread: