TweetDeck can be used both as an organic Twitter client and to schedule future tweets. It turns out that it's possible to estimate the percentage of tweets that were scheduled in a given set of tweets sent via TweetDeck.
(formula explained in thread)
cc: @ZellaQuixote
(formula explained in thread)
cc: @ZellaQuixote
Humans tweeting organically don't show a preference for tweeting in the first second of any given minute versus the fifth, thirtieth, or whatever. Twitter traffic sent via the website reflects this; tweets are distributed randomly and more or less evenly across all sixty seconds.
TweetDeck's scheduling feature has minute accuracy and the scheduled tweets appear to almost all post during the first second of the minute for which they are scheduled, resulting in the spike shown in this figure. How can we use this?
Here are examples of individual accounts. All of @DrunkAlexJones TweetDeck tweets thus far have been scheduled, and all fall within the first second of the minute. @propornot's usage is organic/live, resulting in a random distribution.
We can use this to come up with a (simplified) metric for estimating the percentage of scheduled tweets in a given set of TweetDeck tweets. Assuming that A) all scheduled tweets fall within the first second and B) 1/60 of nonscheduled tweets yields the following formula:
Footnote: we've seen networks that tweet via TweetDeck and appear to be bots but don't flag this particular test. It's possible the operators of these accounts are automating TweetDeck via means other than the (fairly basic) standard scheduling feature.
