Right. That was our biggest deviation from their numbers. There's quite a lot of things that could account for it -- both of us are using imperfect heuristics t

Right. That was our biggest deviation from their numbers. There's quite a lot of things that could account for it -- both of us are using imperfect heuristics to classify them, and they didn't exactly explain their methodology with regards to what counts as a unique question. In our case, we counted rephrasings, RTs, and quotes as retweets, not just the native ones. This may account for the difference. We may make the data set we gathered public some time, so people can draw their own conclusions about which categorizations are more accurate.