You know you’re in trouble when the main result in a paper you’re reading takes up half a page to state. Not prove; state.
You’re in bigger trouble if you’re writing said paper.
Well, more or less writing. One of my collaborators prefers to have as little to do with the typesetting process as possible, so whenever we’re putting a draft together I get a manuscript sent to me through the post. Said manuscript is almost invariably written in pencil, on the back-sides of whatever half-blank paper was lying around his office: old print-outs of other manuscripts, tests his students had written, CVs of former grad students… Also, much scribbling and crossing out, but that’s to be expected. Anyway, I’m transcribing such a manuscript now, making corrections and edits along the way. And oh my, this one theorem.