I really wish they'd have provided their false positive rate over several lengths of document, rather than an overall estimate. Because if it dives after say, 1

I really wish they'd have provided their false positive rate over several lengths of document, rather than an overall estimate. Because if it dives after say, 1,500 words, that's a relevant piece of information for its use. I'm pessimistic, given they chose not to do so.