[TC_Halogen]

The premise behind deviance in timing windows is absolutely brilliant, because at the surface, we all see timing windows as chunks, but not as their physical distances in comparison to the central point of timing for a given note.

Using J4 to illustrate the point (even though it's not necessarily relevant data wise, it still is concept wise): if I get a perfect because I missed the note central point by 22.51 milliseconds, that perfect should be basically worth the maximum value that a perfect can be worth compared to someone who missed the note by 44.99 milliseconds, which should be the absolute minimum that a perfect should be worth.

I also like the idea of reducing freeze arrow weight to 0 -- it actually helps to add difficulty to charts that utilize them properly with regards to balance while mitigating the unnecessary advantage of getting extra DP due to micro-freezes whose lengths are trivial. You could initially apply a similar concept to life gain as well (not sure if mentioned).

Solid proposal overall; I could respond to this a bit more in detail tomorrow, perhaps.