The complaints all seem to be some form of "the PP system does not weight X enough", where X can be skills like consistency, accuracy, sliders, tech reading, low-AR reading, etc.
-the majority of these people just want THEIR skill weighted more, not necessarily because they genuinely believe that the skill is underrepresented
The essential difference between PP!Balance and PPv2 is that, unlike PPv2, PP!Balance doesn't try to directly define what it is exactly that is difficult in a beatmap (e.g. aim, accuracy, speed, reading, finger control, stamina and so on). Instead it lets the difficulty of a map be determined empirically by players' ability to play it.
-in principle, it’s a nice idea, but there are a lot of shorts in the idea once you get down to it
RANK BASED SUPERIORITY
N is the number of other plays made on that map (including your own plays and failed plays)
-the game is centered around high scores, I don’t think failed plays should be counted. I don’t think # of plays should be counted at all. If a high score was set but you took 1000 tries, should it be worth less than the same high score set with 100 tries?
The more plays-within-a-year a beatmap has, the more PP you get for getting a high score on it.
-this is pp farming, Reol No Title gets top PP, no argument
The reason why the PP of a play should depend on N is that the more people try a beatmap, the higher the competition is for getting high ranks on it. It is more difficult to win competitions where lots of other players compete
-so 4.5 star tv size maps should be worth disproportionally more PP than the 6* maps just because the 4 stars are played more?
k is a constant weight that determines the importance of N.
-how exactly will this be set? Manually determined by a person? Generated?
The second term (1/ln(stuff)) is supposed to scale inversely logarithmically with your rank on a beatmap.
-this is a problem because the usual PP system does not include things like spinner wars… but this will encourage it. For scores that have many HDHR FCs, you’re telling me that with a score difference of 1000, the second place should get 0.76x the PP of the #1?
SCOREv1 BASED SUPERIORITY
-There are major problems with this. The first is that score averages are dynamic, not static. It’ll require shitloads of calculations every day. Not only that, it makes the PP system dynamic and relative, which defeats the whole purpose of the PP system, which is to display your top plays.
The idea assumes that average score on a beatmap is a good proxy for how difficult it is.
-this won’t work. Let’s say there’s a map that’s really easy for tv size, but at the middle there’s a part that’s unreadable. God knows how that’s ranked, but let’s suppose that it is and EVERYONE misses on that part. Should the one guy that smashed keys, got all 100s and proceeded to FC get an amazing amount of PP compared to the rest? Ame to Asphalt is another good example, it’s easy until the very end which has 6.5* jumps for 16 measures iirc. So average score will be high, and the guy that hits all the jumps will get lower pp than he deserves.
If beatmap leaderboards will continue to stay ScoreV1, then I think the solution to this is just to keep two leaderboards for each beatmap, one for each scoring method, and then just make the ScoreV2 leaderboard hidden. That way, you can make plays that have a higher ScoreV1 value than ScoreV2 value without losing PP.
-this defeats the whole purpose of the scoreboard system. The leaderboards lose their prestige since people will try to datamine how to get PP and only focus on the hidden scorev2 leaderboard.
the more failed scores a beatmap has, the more PP should be given. One way of dealing with this could be to just submit failed scores (i.e. their ScoreV2 values up until the point they failed) into a hidden database for that beatmap, and then include those scores when calculating the average score on that beatmap.
-this can be exploited by spamming maps (I see multi lobbies with rank 100k players playing freedom dive all the time), which reduces the pp of the people that actually mastered the map. This again makes the pp system dynamic and relative, which I already wrote.
Anyway, here's the formula I think has the best scaling. I won't go into how I derived this, but if you need to know you can ask me.
-log is a unitless function…
We give 0 PP to all scores below it.
-doesn’t work, because every play requires some display of skill… you’re telling us that “plays under this score require no displays of skill, therefore no pp”
k_1 and k_2 are constants that determine the range of PP values.
-how will this be determined?
To achieve 1256 PP, you need to score a thousand times better than average.
-again… this makes the pp system dynamic
-the majority of these people just want THEIR skill weighted more, not necessarily because they genuinely believe that the skill is underrepresented
The essential difference between PP!Balance and PPv2 is that, unlike PPv2, PP!Balance doesn't try to directly define what it is exactly that is difficult in a beatmap (e.g. aim, accuracy, speed, reading, finger control, stamina and so on). Instead it lets the difficulty of a map be determined empirically by players' ability to play it.
-in principle, it’s a nice idea, but there are a lot of shorts in the idea once you get down to it
RANK BASED SUPERIORITY
N is the number of other plays made on that map (including your own plays and failed plays)
-the game is centered around high scores, I don’t think failed plays should be counted. I don’t think # of plays should be counted at all. If a high score was set but you took 1000 tries, should it be worth less than the same high score set with 100 tries?
The more plays-within-a-year a beatmap has, the more PP you get for getting a high score on it.
-this is pp farming, Reol No Title gets top PP, no argument
The reason why the PP of a play should depend on N is that the more people try a beatmap, the higher the competition is for getting high ranks on it. It is more difficult to win competitions where lots of other players compete
-so 4.5 star tv size maps should be worth disproportionally more PP than the 6* maps just because the 4 stars are played more?
k is a constant weight that determines the importance of N.
-how exactly will this be set? Manually determined by a person? Generated?
The second term (1/ln(stuff)) is supposed to scale inversely logarithmically with your rank on a beatmap.
-this is a problem because the usual PP system does not include things like spinner wars… but this will encourage it. For scores that have many HDHR FCs, you’re telling me that with a score difference of 1000, the second place should get 0.76x the PP of the #1?
SCOREv1 BASED SUPERIORITY
-There are major problems with this. The first is that score averages are dynamic, not static. It’ll require shitloads of calculations every day. Not only that, it makes the PP system dynamic and relative, which defeats the whole purpose of the PP system, which is to display your top plays.
The idea assumes that average score on a beatmap is a good proxy for how difficult it is.
-this won’t work. Let’s say there’s a map that’s really easy for tv size, but at the middle there’s a part that’s unreadable. God knows how that’s ranked, but let’s suppose that it is and EVERYONE misses on that part. Should the one guy that smashed keys, got all 100s and proceeded to FC get an amazing amount of PP compared to the rest? Ame to Asphalt is another good example, it’s easy until the very end which has 6.5* jumps for 16 measures iirc. So average score will be high, and the guy that hits all the jumps will get lower pp than he deserves.
If beatmap leaderboards will continue to stay ScoreV1, then I think the solution to this is just to keep two leaderboards for each beatmap, one for each scoring method, and then just make the ScoreV2 leaderboard hidden. That way, you can make plays that have a higher ScoreV1 value than ScoreV2 value without losing PP.
-this defeats the whole purpose of the scoreboard system. The leaderboards lose their prestige since people will try to datamine how to get PP and only focus on the hidden scorev2 leaderboard.
the more failed scores a beatmap has, the more PP should be given. One way of dealing with this could be to just submit failed scores (i.e. their ScoreV2 values up until the point they failed) into a hidden database for that beatmap, and then include those scores when calculating the average score on that beatmap.
-this can be exploited by spamming maps (I see multi lobbies with rank 100k players playing freedom dive all the time), which reduces the pp of the people that actually mastered the map. This again makes the pp system dynamic and relative, which I already wrote.
Anyway, here's the formula I think has the best scaling. I won't go into how I derived this, but if you need to know you can ask me.
-log is a unitless function…
We give 0 PP to all scores below it.
-doesn’t work, because every play requires some display of skill… you’re telling us that “plays under this score require no displays of skill, therefore no pp”
k_1 and k_2 are constants that determine the range of PP values.
-how will this be determined?
To achieve 1256 PP, you need to score a thousand times better than average.
-again… this makes the pp system dynamic