Twitter is all the rage lately. Media personalities in journalism, sports and Hollywood have started tweeting, and the masses have followed. Taking a break from our regularly-scheduled research, we wanted to see what effect these media personalities have had on Twitter’s growth. The answer is: a lot.
We randomly selected about 70,000 Twitter accounts* and plotted them by when they were created. The x (horizontal) axis is the age of the account, so “0” (far left) is the most recent account (as of Tuesday night) and the far right represents accounts created about 1000 days or nearly three years ago.
About six months ago, the number of Twitter accounts exploded. However, two spikes occur: starting about two weeks ago and about 10-11 weeks ago. These coincide with two important media events surrounding Twitter.
The first comprises the celebrity trifecta of Ashton Kutcher, Oprah Winfrey and Larry King. Ashton and Oprah’s accounts were created in January (about 6 months ago, coinciding with the increase in new users), and King followed in March. Steady growth continued throughout the first part of the year, but things became more interesting in April.
On 4/13, Kutcher appeared on CNN and they announced a challenge: who could reach one million followers first? Oprah jumped on this bandwagon by having Kutcher on her show and sending her first tweet on 4/17. This coincides with the first (right-hand) peak. The peak was short-lived, and account creation decreased until nearly two months later when the Iran crisis developed, but that’s another story.
The question is, did Oprah/Ashton/Larry cause the spike in new accounts, or did they go along for the ride?
It is too early to see the after-effects of the Iran peak, but it will be interesting to see whether people who create accounts for news/geopolitical reasons are more or less “sticky” for Twitter than those who joined for entertainment reasons.
This post is sort of an advertisement for our group; several academic papers will be coming out of this summer’s work on Twitter, and we have already released a draft version of “Tweet, Tweet, Retweet” (boyd, Golder and Lotan), which describes Twitter users’ retweeting practices. Stay tuned for more.
* Methodology: We approximated the most recently assigned user ID and generated user IDs uniformly between zero and this maximum user ID. We ignored any user IDs that mapped to a nonexistent or suspended user account.