i judge replays by 2 scales. execution and idea. idea refers to what you are doing, execution refers to how well you are doing it.
i can only evaluate execution if the idea is somewhat clear, since objectives and goals are not universal and etc. (for example i could make my goal to make a replay where i fall to the ground and break my neck while looking stiff an symmetrical. i would contract the hips and hit my head on the ground, breaking it. the execution would be a 10/10 perfect score in that case or something very high atleast).
the biggest problem for judging replay (for me atleast) stems from this. i always have to speculate what the idea of the replay is, unless it is clear or the replay is trying to immitate something (eg: Binklawz).
i give somewhat 50/50 importance to both the idea and the execution if they are both equally as good. i usually give more importance to the idea the worse or the better it is. for example, if someone made the replay that i hypothesized above (breaking your neck like a dumbass) then i would obviously nitpick that it is not a good idea at all, because it ruins the replay. if they made a replay with an insanely good or creative idea, i would praise it and give it a lot more importance than the execution (not complete importance though, like 75% idea and 25% execution at max). if someone made a replay with a so-and-so idea that does not "define" the replay, but that doesn't ruin it either, i would care little about the idea and try to talk more about the execution.
so the function for my criteria would be something like y=(x - 50)^2, domain [0,100], being y=importance given to the idea and x=idea quality in percentage (0-100%)
execution, i usually judge that one pretty straight forward. when looking at execution i usually try to see if the idea is achieved (objective), and if the idea is achieved in a way that looks good (subjective).
if an idea is not very good, then i will not really care if you managed to achieve the idea (in a way, quality of the idea=how much i care about if you achieved it or not). but in that case, i will care more if you achieved it in a way that looks nice. of course, that criteria is completely subjective in function of my taste, which is another problem for judging replays accurately. some days i will be in the mood for something different than usual, etc..
i don't know what makes a replay great. therefore a great replay is a replay that shows me something i didn't know/i didn't know was possible.
THAT DOES NOT MEAN THAT A GREAT REPLAY NEEDS TO HAVE A COMPLETELY ORIGINAL IDEA THOUGH. for example you can make a replay where you kick off every single joint off of uke in a such a fashion so that it looks very elegant, and i would be like "wow this replay is great, i didnt know you could do that". if you've read David Hume's empiristic argument on the origin of knowledge, you know that he defends that there are no innate ideas (ideas that we are born with, necessarily true ideas that serve as paradigms. empirism defends that all knowledge comes from experience, your human senses). he explains the fact that we are able to imagine stuff that doesn't exist such as unicorns or mermaids by associating different ideas absorbed by life experiences. for example, i have seen a horse and i have seen wings, so i can associate both and imagine a pegasus. in that same way, i can imagine a replay that hasn't been made yet and that i am not sure is possible to create. i have seen a replay where every joint gets dismembered. i have seen someone kick joints off. i have seen someone look elegant. therefore, i can imagine a replay where someone kicks off every joint in an elegant way, even if such replay does not exist. because of that, i can imagine replays that do not exist but that i wish to see. and that is what a great replay is to me. a replay that breaks that barrier, that shows me something that i did not think was possible.
of course, this implies that i do not think that everything in toribash is possible. i used to be that "endless possibilities WOW" guy, but not anymore really. i take into account people's ability to dedicate time and effort into replays, which brings in a less accurate factor into the mix. it is not really possible to accurately determine if a replay is "great", but i try.
next time on tl;dr, dont lose our discussion on boomhit ethics and the ontology of openers