Musings regarding activity, wins, points awarded
Posted: Sun Oct 31, 2010 6:47 pm
I've been keeping a close eye on some of the points usagi has been raised recently on appropriateness of activity rewards, and 3 of his prime tenets were:
1) People with poor records (I'm going to use < 50% for the sake of this post) shouldn't be promoted / awarded things
2) People with lots of wins should benefit
3) People with good win percentage records should benefit
I was in the same division as him last month, so as it was one with good familiarity to both of us regarding availability and games, I thought I'd assess some of these premises. 3 people were basically inactive, so for the sake of argument I'd say 6th is the middle of the division, 1-5 the top half, and 7-11 the bottom half.
1a) Only one player with <50% record was in the top half, out of 5 players.
1b) Four players in the bottom half had a record <50%.
This is as I would imagine it, and it doesn't feel like the distortion effect is as bad as usagi seems to claim (with people able to promote simply by playing lots and getting 40-50% win records.
2a) 4 people got 10+ wins. These players occupied 1st, 2nd, 4th and 5th.
2b) The top two players on wins alone finished 1st and 2nd.
Again, this fits nicely with usagi's hopes that people winning lots would finish highly.
3a) Best winning percentage came 1st. 2nd best came 5th. 3rd best came 7th. 4th best came 2nd.
3b) Most games game 3rd. 2nd most came 6th. 3rd most came 2nd.
So, best winning percentage doesn't correlate hugely strongly, and if anything total games correlates more closely. However, it is clear that winning percentage and total won games is at least as important overall as the number of games played, and for those that have played a large number of games winning percentage seems particularly important.
Obviously this is only a sample size of 1 (division), and is hardly significant or impressive - I just picked it because as we were in the division together I thought I'd use it as a pilot. I'd be delighted if other people did something similar if there was interest there.
However, while I genuinely understand where usagi is coming from, and understand his frustration, I do think he is hugely overstating the imbalance, and the size of effect the reward system is having on "punishing" stronger players. For what it is worth, the 2 promoted players have the highest KGS ranks on the division at that time, which I hope is reassuring.
The purpose of this post I suppose is also partly to ask: usagi, mathematically calculated theory aside, do we really have the problem you think we do in ASR, in practice?
1) People with poor records (I'm going to use < 50% for the sake of this post) shouldn't be promoted / awarded things
2) People with lots of wins should benefit
3) People with good win percentage records should benefit
I was in the same division as him last month, so as it was one with good familiarity to both of us regarding availability and games, I thought I'd assess some of these premises. 3 people were basically inactive, so for the sake of argument I'd say 6th is the middle of the division, 1-5 the top half, and 7-11 the bottom half.
1a) Only one player with <50% record was in the top half, out of 5 players.
1b) Four players in the bottom half had a record <50%.
This is as I would imagine it, and it doesn't feel like the distortion effect is as bad as usagi seems to claim (with people able to promote simply by playing lots and getting 40-50% win records.
2a) 4 people got 10+ wins. These players occupied 1st, 2nd, 4th and 5th.
2b) The top two players on wins alone finished 1st and 2nd.
Again, this fits nicely with usagi's hopes that people winning lots would finish highly.
3a) Best winning percentage came 1st. 2nd best came 5th. 3rd best came 7th. 4th best came 2nd.
3b) Most games game 3rd. 2nd most came 6th. 3rd most came 2nd.
So, best winning percentage doesn't correlate hugely strongly, and if anything total games correlates more closely. However, it is clear that winning percentage and total won games is at least as important overall as the number of games played, and for those that have played a large number of games winning percentage seems particularly important.
Obviously this is only a sample size of 1 (division), and is hardly significant or impressive - I just picked it because as we were in the division together I thought I'd use it as a pilot. I'd be delighted if other people did something similar if there was interest there.
However, while I genuinely understand where usagi is coming from, and understand his frustration, I do think he is hugely overstating the imbalance, and the size of effect the reward system is having on "punishing" stronger players. For what it is worth, the 2 promoted players have the highest KGS ranks on the division at that time, which I hope is reassuring.
The purpose of this post I suppose is also partly to ask: usagi, mathematically calculated theory aside, do we really have the problem you think we do in ASR, in practice?