## Twitter follower semi-log graphs

[20 Dec 2010]

Here are some interesting distributions, and another “real-life” application of semi-log graphs.

The information comes from Hubspot’s *State of the Twittersphere report*, dated Jan 2010, which is unfortunately no longer available.

One of their conclusions:

The vast majority of Twitter users have networks under 100 people, [with] 82% of Twitter users having less than 100 followers, and 81% of Twitter users are following less than 100 people.

This makes sense to me. If you follow thousands of people, how can you hope to read and respond to even half of them?

This reminds me of Dunbar’s Number, the idea that humans tend to cluster in groups of around 150, since that is the limit of reasonably expecting to know everyone in the group.

But the low number of following/followers is more likely explained by the large number of people who would start Tweeting, then give up.

### Twitter – most followers

As I write, the top 3 people on Twitter with the most followers (according to Twitaholic) are:

Followers | Following | |
---|---|---|

Lady Gaga (ladygaga) | 7,394,362 | 145,898 |

Britney Spears (britneyspears) | 6,416,696 | 416,140 |

Justin Bieber (justinbieber) | 6,355,747 | 95,167 |

Barack Obama is #5 on this list, with over 6 million followers.

### Twitter – follower distribution

Now to the graphs. They have used a **semi-log scale** for these graphs so we can see more detail at the higher end (a million plus) and for the more “normal” number of followers.

This graph reminded me of Zipf Distributions, another place we find semi-log graphs.

### Twitter – following distribution

And for the distribution of the number of people Twitterers are following:

There is a “blip” in the middle of that last graph. It is explained by the fact that Twitter imposes a temporary limit on following at the 2,000 level.

### Twitter – patterns of use

Other findings from Hubspot’s report were interesting:

- Thursday and Friday are the most active days on Twitter, each accounting for 16% of total tweets in our study.
- 10-11 pm is the most active hour on Twitter, accounting for 4.8% of the tweets in an average day.

And finally, lots of people try to use every one of thier 140 character limit in their tweets.

This graph has linear axes (it has linear scaling on both the horizontal and vertical axes). The vertical scale is presumably the number of tweets.