Improving the rating system is basically improving its accuracy, because this helps competitive players to measure their progress, thus feeling rewarded and motivated. It also helps all players to choose the opponents they feel more comfortable with. All this implicates more fun for everyone and in consecuence more playing, instead of spectating/watching which today requires less wait time and have more alternatives. The more games are played in Voobly, the better gaming experiance and also the more strong the community is, generating a virtuous cycle that will make more people prefer the client every day.
Virtuous cycle:
Rating accuraccy => More games => Better gaming experiance => More games => ...
To improve rating accuracy lets enumerate the current system’s problematics to seek for solutions.
¿What problematics are affecting the accuracy today?
1 .- Pointsitters
A decay system is necessary but today it is producing a wall of flat 2k inactive ratings
2 .- The wall of flat 2k inactive ratings itself
It is imposing a psychological barrier, and downplaying the prior goals (16++, 17++, 18++, 19++)
3 .- Inactive ratings in general
The current rank# of a player below 2k is being downplayed by the bazillions of inactive ratings, whilst current rank# could be a nice way to categorize, acknowledge and motivate active users (e.g. current top 10, top 100, top 1000)
4 .- Smurfs
Smurfs are often getting in the way of average players, dragging their ratings down and in some cases causing frustration to players who are seeking for a balanced game
5 .- The mountain of game count stigma
Nobody mention this, obviously because nobody will recognize to feel ashamed or tired of this psychological bulk. This data can only be misinterpreted as “look at how many games played to get some skills”, because it have no other practical reason than showing that the rating does not belong to a smurf. It is a misinterpretation because the contrast that exist between the honest player who decides to play most of the time in the same account and the “prodigy” who have played a much smaller amount of times in the ladder, with a new account.
Proposal for dealing with these problematics
1 .- Hidding inactive ratings: this will normalize the current rank# based in active players and eliminate the psychological barrier of the wall of flat 2k inactive accounts. An inactive rating should be a rating that has no games during 4 weeks.
2 .- Lowering decay base to 1600 but decay in percentual value: this will produce a better distribution of the ratings, eliminating the abnormal tendence of ratings to concentrate around the 2k barrier. For example if a player is 1850, decay the 2% from the difference between 1600 and 1850, i.e. 250 * 2 / 100 = 5 points in a week of unactivity. A 2k player will lose 8 points in a week, a 2k2 12, a 2k5 18. In a year of unactivity a 2k5 will have 19++, which is too much decay, but it is possible to implement a mechanism that stops decay when it accumulates a 30% of the average difference (e.g. 1850 - 1600) of the last 10 games. Also it can stop when the rating reaches 1699, to put a reasonable limit to the number of calculations to be made. To optimize, all these calculations can be made just after the player plays a game in that rating, this way abandoned or forgotten acounts do not have to be updated (most will be hidden after the 4 weeks of unactivity).
2.1.- Require player to earn/lose 16 points in total to prevent decay. For example, if I am nearly a week of inactivity, and want to sit in my rating by playing a way lower player, I would need to win 16 times to keep my rating. Other example is win to 8 noobs and lose to 8 pros to sum the 16 points needed, that is a lot of work and will discourage most players from abusing in this manner.
3 .- When the count of games played reaches more than 100, display it as 100+. This eliminates the subliminal message of “the mountain of game count stigma”. This will prevent smurfings from people who want to take a break from it.
Please feel free to express your opinions and critizism about this proposal, and to express what you think should be a better rating system.
Virtuous cycle:
Rating accuraccy => More games => Better gaming experiance => More games => ...
To improve rating accuracy lets enumerate the current system’s problematics to seek for solutions.
¿What problematics are affecting the accuracy today?
1 .- Pointsitters
A decay system is necessary but today it is producing a wall of flat 2k inactive ratings
2 .- The wall of flat 2k inactive ratings itself
It is imposing a psychological barrier, and downplaying the prior goals (16++, 17++, 18++, 19++)
3 .- Inactive ratings in general
The current rank# of a player below 2k is being downplayed by the bazillions of inactive ratings, whilst current rank# could be a nice way to categorize, acknowledge and motivate active users (e.g. current top 10, top 100, top 1000)
4 .- Smurfs
Smurfs are often getting in the way of average players, dragging their ratings down and in some cases causing frustration to players who are seeking for a balanced game
5 .- The mountain of game count stigma
Nobody mention this, obviously because nobody will recognize to feel ashamed or tired of this psychological bulk. This data can only be misinterpreted as “look at how many games played to get some skills”, because it have no other practical reason than showing that the rating does not belong to a smurf. It is a misinterpretation because the contrast that exist between the honest player who decides to play most of the time in the same account and the “prodigy” who have played a much smaller amount of times in the ladder, with a new account.
Proposal for dealing with these problematics
1 .- Hidding inactive ratings: this will normalize the current rank# based in active players and eliminate the psychological barrier of the wall of flat 2k inactive accounts. An inactive rating should be a rating that has no games during 4 weeks.
2 .- Lowering decay base to 1600 but decay in percentual value: this will produce a better distribution of the ratings, eliminating the abnormal tendence of ratings to concentrate around the 2k barrier. For example if a player is 1850, decay the 2% from the difference between 1600 and 1850, i.e. 250 * 2 / 100 = 5 points in a week of unactivity. A 2k player will lose 8 points in a week, a 2k2 12, a 2k5 18. In a year of unactivity a 2k5 will have 19++, which is too much decay, but it is possible to implement a mechanism that stops decay when it accumulates a 30% of the average difference (e.g. 1850 - 1600) of the last 10 games. Also it can stop when the rating reaches 1699, to put a reasonable limit to the number of calculations to be made. To optimize, all these calculations can be made just after the player plays a game in that rating, this way abandoned or forgotten acounts do not have to be updated (most will be hidden after the 4 weeks of unactivity).
2.1.- Require player to earn/lose 16 points in total to prevent decay. For example, if I am nearly a week of inactivity, and want to sit in my rating by playing a way lower player, I would need to win 16 times to keep my rating. Other example is win to 8 noobs and lose to 8 pros to sum the 16 points needed, that is a lot of work and will discourage most players from abusing in this manner.
3 .- When the count of games played reaches more than 100, display it as 100+. This eliminates the subliminal message of “the mountain of game count stigma”. This will prevent smurfings from people who want to take a break from it.
Please feel free to express your opinions and critizism about this proposal, and to express what you think should be a better rating system.