But that's the whole problem, isn't it? Consider how a P2P network operates. There are N users with a copy of the song. From this we know that there have been at most N uploads, for N users, so the average user has uploaded 1 copy. Really slightly less than 1, since at least one of them had the original so there are N-1 uploads and N users and the average is (N-1)/N.
There could be some users who upload more copies than others, but that only makes it worse. If one user in three uploads three copies and the others upload none, the average is still one but now the median is zero -- pick a user at random and they more likely than not haven't actually distributed it at all.
Meanwhile the low end of the statutory damages amount is 750X the average, which is why the outcome feels absurd -- because it is.
Consider what happens if 750 users each upload one copy of a $1 song. The total actual damages are then $750, but the law would allow them to recover a minimum of $750 from each of them, i.e. the total actual damages across all users from each user. The law sometimes does things like that where you can go after any of the parties who participated in something and try to extract the entire amount, but it's not that common for obvious reasons and the way that usually works is that you can only do it once -- if you got the $750 from one user you can't then go to the next user and get another $750, all you should be able to do is make them split the bill. But copyright law is bananas.