Friends of mine and I were talking about the idea of a canon or pantheon of great movies, and how to select them. We had a fun debate about the criteria, and eventually I drew up a list.
For me a "great" movie needs to satisfy at least one of the following:
It isn't hard to follow this list and draw up, say, 50 or 100 movies of your own that fit. That said, many movies I like or even love do not really satisfy any of these things; I just happen to love them for some highly personal reason.
Still, I find it useful to come up with rules like this as a way to consider why we end up with so many of these things in what amounts to a hierarchy anyway. I think it's because we know life is short, that there are only so many things one can experience in that short life, and we want to know what's most worth it. But we also want to know how those recommendations were made, and where our own curiosities might and might not overlap with those recommendations.
One of the things I've had to wean myself of is obscurity as some criterion for greatness — that we should be looking towards all that time and attention have ignored first. Yes, there are quite a few truly great movies unjustly ignored by time (Son Of The White Mare; Killer Of Sheep), but the mere fact something has not been given notice is not always a sign of greatness ignored. Many things I love to wave under other people's noses are not well-known, but I also know I can only make a partial argument, or sometimes no argument at all, for greatness — merely that they are distinctive, and that they deserve a moment under someone else's sun if only for that reason. And again, just because something is unlike its peers doesn't grant it superiority, or longevity, or even true distinction. Sometimes it's just an odd man out.
I suspect some variation of the above criteria can be applied to most any art form.