I remember back when the issue of ending input early was controversial.

That time is back! Wow this movie is a Goldberg machine. It simply gently pushes the ball that then takes 12 minutes to RICH while presenting you a hilarious comedy. I laughed several times, it's the funniest TAS I've seen in a long while. The main feat is gameplay that looks primitive but does in fact have variety for almost the whole time. You never know who loses next and when, and how long the opponents take to defeat themselves. At first it even looks like one of them is winning and maybe the whole thing desynced. And it's much more ridiculous to watch this in the emulator with input display, when you are already sure the actual input ended minutes ago, but keeps delivering. This movie is not trivial, even simply judging by the fact that it was revised a few times after considered finished, and quite a bit of time was saved since then. While the current input may be trivial to execute, it was not trivial to come up with it. Technically the same situation as with [4173] SMS Zool: Ninja of the "Nth" Dimension "game end glitch" by The8bitbeast in 00:21.61, the only difference is this game doesn't have a published movie yet. I also remember a movie where it was a meaningful question whether you aim for shortest input or fastest ending: #7123: Jigwally's NES Solitaire in 00:40.93.

• You can potentially delay the ending indefinitely with your shorter input. We wouldn't want t to watch a 10-second movie that makes the game end 10 hours later.
• But you can't indefinitely speed it up in your ending-based movie. There's only so much you can do to end the game faster, and it gradually gets harder and harder to come up with new timesaves. You can't make it happen instantly.
• And you can't indefinitely delay movie end to make the game end faster. After some point it becomes unnecessary.

So in terms of obsoletion criterion, ending-based movie is clear cut. I think for this game, judging movie optimality by when the game ends is a better idea.

There's only so much speedrun challenge to this game if one only aims for shortest input. When the only room for further optimization is those 17 frames, it's not hard to just go through them all and test them, checking if either of them leads to an ending at all. Arguably if there's some obscure combination of inputs that makes the game end even later, how do we judge it? Optimality becomes a moot concept when you sacrifice one important aspect to gain another. But if we aim for quickest ending, there's still a lot to optimize in the future. Aiming to actually win as soon as possible gives you minutes to work with, and after some time you can't make it end sooner by extending your input any further. Now the really hard question is whether we should prefer some approach to a given game or should we just start allowing both to co-exist. For me this movie is absolutely Moons, but I'm sure the ratings won't agree with my sense of humor. I'm not ready to allow both for Standard because I don't agree that they are de facto standard goals already. If you only have 2 options, and for every single movie you have to decide which to pick, it may become a part of your actual in-game goal, but it can't become a separate goal in and of itself. It can be a second version of any goal you set, depending on the game. I don't know if many people here would want to see every movie in 2 variations of how it ends. At the same time I can see unique value in both if they look distinctly different. How about this: If aiming for a different kind of ending can make the whole movie look different enough, then it's a proof it can work as a goal on its own and deserves a place in the Standard class. If it's just a minor variation of the main goal, it's up to the author and the audience which one to prefer.