Will it be helpful to decouple aim, speed and accuracy pp?
opened this issue · comments
As title.
As any weighting of a specific skill basis set will be under/overweighted in some ways, is it just better to decouple them, and find a "minimum orthogonal basis set" with like 3 to 5 dimensions to describe the performance, with separate leaderboards?
While I think (aim, speed, accuracy) is small and orthogonal enough as a skill basis, and changes can be rolled out quickly while solving some "underweighted" or "overweighted" problems. The overall ranking may be then substituted by ordering the sum of the three dimension rankings, which I considered as a community-scale average rather than a personal-scale average, favoring all-rounded players. So that there will be less flaming in the community on "overweighted" maps.
If it (aim, speed, accuracy) is not suitable, I think Syrin's documentation is great, but it will need the reworking of the whole system and fitting it with suitable parameters.
And I'm not familliar with C++ or operation on databases, and the modification of pp system is an API-level change, so I opened this issue here rather than working on the code and opening a PR. My proposal may work like:
| Player | Aim | Speed | Accuracy | Overall rank |
|---|---|---|---|---|
| Player1 | 8000pp (rank 3) | 6000pp (rank 1) | 3000pp (rank 1) | 1 |
| Player2 | 9000pp (rank 1) | 5000pp (rank 3) | 2900pp (rank 2) | 2 |
Hi there, the original ranking system ppv2 is based on (osu!tp) did things the exact way you describe. The main reason ppv2 does not already work like this was technical. A separation of the 3 metrics is welcome and I fully agree that the final metric should at least by a large part be comprised of just the sum of the 3 metrics. I say "by a large part" because good scores on single maps with high difficulty in all 3 categories should also be worthwhile.
I'd consider adding further categories and finding suitable bases a different issue which we should investigate only once separation of aim/speed/acc is in.
Oh I got it. Thanks for the explanation :)