But their criteria for user was anyone who had signed up. Twenty percent of them didn’t even have a single follower.
The fact is that loads of people dip their toe into the twitter pool and decide the water is too cold. They join and quit in droves.
I re-ran the experiment looking at a sample of the last 1000 tweets from the people I follow. I regularly prune dormant users so the vast majority of the 350 may be considered active ie they have tweeted at least once in the past month. 200 of them were represented in my 1000 tweet sample.
The distribution curve of tweet contributions looks much less skewed than above. In fact Pareto’s Principal looks well on track with 24% of users account for 79% of tweets:
It’s worth noting that while 40% of those I follow did not contribute a tweet, the next third (33%) contributed 3 tweets or less.
The weakness of this analysis is, of course, my sample may not be representative. I may have a bias to unfollow hyperactive tweeters with a high “follow cost” (ie they are noisy and take up a lot of bandwidth).
So which view is most accurate? The truth probably lies somewhere in between the two.
Suffice to say the Twitter dead should be buried before doing a census.