Over the past few weeks, we have been trialing a new system for ranking users based on points. After a number of revisions based on suggestions from those enrolled in experimental features, we feel the system is stable enough as-is to make public.
Why a new system?
The old system over-awarded those who spent time "farming" a large number of easy problems, and under-rewarded those who spent time figuring out more complex solutions. The new system hopes to strike a balance in ranking between users who solve many problems, and users who solve fewer — but harder — problems.
How are points calculated?
In short, the new points system takes the weighted sum of the top 100 submission scores on distinct problems. The score of the 
problem is weighted by 
. Formally, where 
represents sorted scores and 
a user's base points,
![\displaystyle P = \sum^{\min(99,\ \left| S |\right)}_{i=0} 0.95^i \cdot S[i]](https://dmoj.ml/mathoid/f927323f8778bf63433f6cbcbe6c96b7bfba66d6/svg)
To not over-penalize users who have dedicated time to solving many problems, an additive adjustment 
is made to the base point value. Where 
is the number of problems a user has a fully-accepted solution on,
A user's point score is then the sum 
.
How can I see a problem's contribution to my points?
A new table has been added to the Problems tab of profile pages, which displays the 100 submissions used in the generation of their point score, as well as their weights.
