Today I was pleased to see how the two Research Information articles that have been published up to now [there are more in the pipeline] compare with other articles in the same journal.

According to Topsy 81 of Research Information’s web pages have been mentioned on Twitter, and the most popular two are by me: Programming skills could transform librarians’ roles (78 mentions); Web 2.0 fails to excite today’s researchers (67 mentions).

As would be expected, at least by the biblio-/webo- metricians amongst you, the distribution of Twitter mentions broadly follows a power law.

Which means third place is quite a distance behind on 43 mentions, and the vast majority of pages have only been mentioned once or twice.

Metrics are only the start of a conversation rather than the end, as there are a host of possible reasons for the difference in Twitter mentions, e.g., my papers may be considered more controversial by some sections. Nonetheless, it’s nice to start the conversation from the top of the pile .

From the 1st of December it is virtually impossible to get any work done as everything revolves around Christmas: the parties, the break, the present buying. If you were visiting Britain from a country that didn’t celebrate Christmas you would be forgiven for thinking that Christmas was a month long festival rather than a single day.

Then finally January arrives, but just as you think you can throw yourself back into work you find that everyone else is catching up on the work they should have done in December!

It is only now that things finally seem to be getting back to normal. On Friday I finally received the Jan/Feb issue of Online with my article What are Libraries doing on Twitter? and the book with my chapter Social Network Sites: An Exploration of Features and Diversity that should have been published at the end of last year.

Nb. Unfortunately my Twitter article has not been made freely available online, so (unless you have a subscription) you will have to wait the 90 days before I’m allowed to put a preprint online.

The network diagram clearly shows that MPs are as bad as the rest of us when it comes to only listening to what we want to hear. The arrows point from the follower to the friend; so the nodes with the most arrows pointing to them are the most followed. [Click here to see full-sized picture].

Each of the nodes in the above diagram represents an MP, coloured according to party affiliation (those MPs that are not in one of the three main parties are coded white).  The layout of the nodes is based on the Fruchterman-Rheingold algorithm, which tries to place connected sites close together. Party affiliation plays no part in the positioning of the nodes, although it obviously plays a part in who MPs listen to on Twitter.

So who is the MPs’ MP on Twitter?

Whether taking into consideration neighbouring connections, or looking at the network as a whole, the most central MPs are all Labour MPs. This is hardly surprising with such a large number of Labour MPs and the tendency for party links.

Degree Centrality (i.e., only taking into consideration neighbouring MPs):

1. Kerry McCarthy (@KerryMP) -33 MP followers
2. Tom Watson (@tom_watson) -30 MP followers
3. Sadiq Khan (@SadiqKhan) -27 MP followers
4. Kevin Brennen (@KevinBrennanMP) -26 MP followers
5. Jim Knight (@jimknightmp) and John Prescott (@JohnPrescott) – both 25 MP followers

Betweenness Centrality (based on the shortest paths between all the different nodes):

1. Kerry McCarthy (@KerryMP)
2. Tom Watson (@tom_watson)
3. Jim Knight (@jimknightmp)
4. Tom Harris (@TomHarrisMP)
5. John Prescott (@JohnPrescott)

Obviously this network analysis has only looked at the MPs in isolation, and we’d hope it was also being used to communicate more effectively with constituents, but as with all social media metrics they are the starting point for the discussion. The next stage is looking out how the networks are actually being used to share information, i.e., through RTs, rather than just potential avenues for sharing information.  Unfortunately we have to wait a bit longer for the API to gain that functionality.

UPDATE October 7th: Curiously enough I just received an email highlighting another study of the political Twittersphere by Sysomos…curious because I was emailed at my other blog which had nothing about the political blogosphere!

A multiple regression analysis based on the data from the people I follow (taking out the outliers identified in the graphs above) produces the following equation:

Expected Followers=1.191*Following + 0.108* Statuses

At the time of collecting the data I should expect to have:
1.191*127+0.108*1,449=307.749 followers
Unfortunately my actual number of followers is only 220, 71.5% of my expected number of followers.

By no means a perfect equation, but it does provide a simple way to compare streams which have been posting for different lengths of time and are following different numbers of people.