>So you're telling me you can only tell if it isn't a bubble if the market steadily declines eventually?
No, that's just its defining characteristic. You can "tell" by other signs too -- but they could be misleading.
>Whatever "bubble" means, it should probably only be used as a description of an abrupt crash in retrospect, otherwise we're always in a non-falsifiable bubble state.
That's the case with every phenomenon that depends in the final result for its definition. E.g. is an act X "beneficial"? We might have some heuristics, but we can only certainly know from its effects after it is completed.
Whatever "bubble" means, it should probably only be used as a description of an abrupt crash in retrospect, otherwise we're always in a non-falsifiable bubble state. Saying "We're in a bubble" really just means colloquially "I think this market segment is overvalued", probably not much more than that.