I agree that having your run beaten by a few frames is not fun. Heck, I wasn't happy when Blechy obsoleted my Umihara Kawase run by four or so seconds.
But I have to side with Phil on this one. There have been many points brought up that I agree with, so I won't repeat them, but I have some other thoughts. Disclaimer: I have not watched Phil's movie, so these are just my thoughts based on the information that I've heard. And I mean no offense to Sleepz.
Let's say that Sleepz had submitted Phil's run instead of his a few months ago. Would it have been accepted? Of course. And then suppose that Phil now submitted Sleepz's run. It would be instantly rejected. It would be slower, and even if it did look a tiny bit better, it would be not enough to change that it was slower. Think being obsoleted by a run 36 frames faster is bad? Imagine being obsoleted by one 36 slower.
So, if (back to normal) Phil's run is not accepted, we have now given an obvious advantage to the person who submits first. Even if someone submits a movie that is a little bit faster or a little bit more entertaining after, the movie will not be accepted. Someone working on a run can then slack off, knowing that even if the run is not perfect, it's good enough -- his or her run won't be obsoleted by missing that 2 frame shortcut. This sort of mindset is not what we want. I'm not blaming anyone of being like this, and I'm sure that Sleepz's run was not beaten because Sleepz was lazy, but the point stands. Time of submission should not be considered. I'm sure there are other reasons why giving the person who submits first an advantage is bad, but I can't think of them at the moment.
Okay, now let's pretend for the sake of argument that Phil has now submitted the perfect run; the game cannot be completed a frame faster. If Sleepz's run had never been submitted, then Phil's would be accepted. Nesvideos now has the perfect SMB2 run in its collection. But because of Sleepz's run, if Phil's is not accepted, we will never have a perfect run. Nesvideos is now short of a perfect video, because someone had previously sent in an imperfect one. I do not see the logic in this. Even if Phil's run is not perfect, he should not be penalized because of previous videos. Although that's sort of misleading, since Phil did use tricks of Sleepz.
How about if all the tricks that Phil had found for not for a few frames, but for a few seconds each? Another instant acceptance. "Wow, it's amazing how you found so many ways to save time!". But this run has just as many time saves as that unreal one, they just happen to not save as much as time. And since they're such small improvements, they would probably be harder to find than large ones (assuming they existed). Why is Phil wrong in finding these subtle improvements?
Plagiarism is also a big concern. People shouldn't take others' runs, hex edit them in a few places, and then submit them. But if the run is played from to finish, I see no problem. The movie is entirely that person's work, even if the ideas are not. But in papers, you are allowed to cite, are you not? Someone using a technique does not give him or her a monopoly on it. I think the idea of mentioning previous submitters is a wonderful idea. Although I guess it would be fair to let authors waive this right if they don't want people to know his or her run was obsoleted :p
In regards to the wobbling, I don't know, because I haven't seen Phil's run. If it really is that bad, fine, reject the run. And then accept it when he submits it without the wobbling. If the run is not accepted, and Phil is told that another run by him will not be accepted by him if it's the same length, do not do it on the pretense of the wobbling. Shifting the blame to that solves nothing. I'd rather that a decision be made (either one) rather than it be deferred.
