That feels like it could be addressed with client side hashing of the note's contents and then deduplication. It might be a little more processor intensive but I'd not imagine it to be TOO bad on performance. Especially if it's only done on notes that are marked as reposts.