The categorization into instant mode and long mode glitches is fine by me. It's probably a more elegant solution with regards to the HoF than the categories I assumed (especially since it rids us of that miscallaneous category).
There are some things I wouldn't really refer to as a glitch, like the Roket-trick, but as you said, that doesn't really matter at this point.
With regards to the IPG vs. PCM distinction, here is my position: They are two different glitches and thus we have to treat them differently. The reasons are as follows:
- They are triggered differently. One (the PCM) is actually triggered outside the actual game, on the loading screen.
- They play differently. Cut has discussed the jumping problems, but PCM also disables the look mode.
- The IPG works on all platforms, the PCM most likely not. CC experimented with it for a while and couldn't get it to work at all. I'm sure he can tell you more about his attempts.
- PCM can be triggered in all parts, IPG only in very specific contexts
In general, the never ending powerups are sideeffects of triggering two different glitches. PCM transports you into a mixed state between camera mode and in-game and IPG in a mixed state between in-game and cutscene. In both cases, you are not directly triggering a glitch that never lets the powerup run out, but rather you change the game into a different state which has, in both cases, the sideeffect of the powerup never running out. The term "infinite powerup glitch" is thus a really bad term because it doesn't tell us anything about what the glitch actually is. Also, if you actually defined IPG = PCM and thought it through logically, you would arrive at the conclusion that the camera mode and a cutscene are the same, which just cannot be right. Therefore, they have to be treated as different glitches.
With regards to those three points, here are my views:
1. All IMGs and the IPG and SD. PCM is not allowed.
2. I agree about IMG and LMG both being represented. I'm okay with two columns but I am pretty much indifferent towards the exact representation (could be two columns or the tooltips we discussed earlier or something different, as long as both are represented equally).
3. I think we have to differentiate more with regards to that. I agree that the method you suggested is the best for a new glitch type. However, I don't think it's a good method for a discovery of new instance of a known glitch (like a new launching glitch). You describe it, as if every new instance of a glitch would require validation from the community. I fully disagree with that. If a glitch type is allowed in the HoF in one instance, it has to be allowed in all instances. That means, a player should be allowed to enter into the HoF without such validation. I have no problem with an agreement that we post all new instances of known glitches, shortly after they are found, but this shouldn't be tied to the HoF. Also, all of that only applies to glitches that can be relevant for scoring. I really don't want this thread to be swarmed with videos documenting every single useless superjump (and there are tonnes of those, just look at the speedrunner-videos).