Khelly wrote:
Now how much effort have you put into ranking compared to him? If what you claim is true then if you worked harder you should be able to easily be ranked higher.
Well, this is the whole point. He uses Tillerino or whatever (not sure exactly) to find very PP efficient songs and farms PP on easy songs that have high yield. Meanwhile, I don't really care about any of that so I just do the hardest/coolest songs I am able to do regardless of PP yield or whatever. This results in me playing more difficult songs that have lower relative yield than his, or just playing much more difficult songs that I can't necessarily FC resulting in lower rank. The variance for this is huge, and now that I check my PP yield after every FC, it seems like the actual difficulty of the song within a certain bracket is completely irrelevant to its PP yield. I've played very fast jumpy songs that took me many plays to FC and felt really good finally nailing only to find I got shit for PP. Other times I play slow boring songs that I FC on, like, the first try, and get tons of PP.
Obviously there is somewhat of a correlation. If a song is MUCH more difficult, it will definitely have more PP on an FC, but when it comes to song remotely within the same bracket of skill there is massive variance. You shouldn't be able to have a 20% PP advantage just because you played the "right" songs, that is WAY too much. In an "ideal" system, a player wouldn't need to go out of their way to play the "special" songs to farm for PP, just play any song and, if they are better, they will have higher PP. Otherwise, you might as well just split ranked up into seasons with a limited map pool of 100something for ranked so nobody has to worry about whether their map happens to be a PP goldmine or not.
I understand that osu! doesn't put much emphasis on an accurate ladder or rankings, and that's fine, I just play it for fun in between queues and waiting for in-houses or whatever, but if the discussion is going to be about whether the PP system actually works as an ELO-style skill determiner, it just doesn't work. It only works on a VERY general basis. If someone has double your PP, they are almost definitely going to be better, sure, but if two people are anywhere remotely near each other (I'm talking 20-30% PP margin), it's completely inaccurate.