I'd like to make a prediction market on bot vs human activity online, but, while there are interesting papers on the subject, nobody seems to publish yearly reports with consistent methodology that could be a prediction target.