Revised version of “The role of reciprocation in social network formation, with an application to blogging”

Caterina and I wrote a thorough revision of our paper on the role of reciprocation in social network formation that was first released in March. It is available on SSRN at

We re-organized the whole of the first part by cutting off some of it to make room for more discussion of the literature. We added many references and the whole reads better overall. The theoretical model was not changed, but we expanded the econometric section to discuss the long-term impact of sustained changes in a blogger’s activity.

Sustained increases in a blogger’s level of activity translate in higher audience, but the process is front-loaded in terms of costs. In short, blogger A who becomes more active will not at first get many more readers. In fact, A will start reading more blogs before they read A back. A’s audience thus increases, but only after a while.

The total effects of changes in activity settle only slowly over time and differ from the short-term effects. That is, a blogger will have to go through a period where she adds more blogs to her reading list than add her before this translates into a more than commensurate increase in her own audience.

We summarize our findings as follows:

  1. Writing more entries will not make your blog more popular, but writing more comments on others’ blogs will.
  2. Getting more features for your blog, if necessary by paying for them (horror!), will boost your readership.
  3. Sustained effort will increase your audience, but don’t expect payoffs to come soon. In fact, you might have to read more blogs than read you for a long while before this pays off. Things will even out over the long-term though.
  4. Adding others to your reading list will get others to add you to theirs. But only about 10% will, according to our estimates. We did not go into what to do about those who do not add you back.
  5. Joining communities of like-minded bloggers is a good way to find new blogs to read. Yes, this may sound obvious, but think that through reciprocation, they might read you back! Hint, hint!
  6. Successful bloggers tend to reciprocate readership more freely (though whether they read back those who read them… we have no way to know…). This could be a reason they are more popular, but we cannot prove this with our data.
  7. LiveJournal bloggers in Russia were supposed to be more media-oriented while those in the US are more into social networking, but this did not translate into differences in our results however. The differences in culture between those two types of bloggers may have been overstated…
  8. There is very strong persistence in term of readership, that is, people tend not to tire from reading blogs they are subscribed to (or at least, your readership will not decrease much over time on its own).

So all in all, from my 10+ years experience of blogging, the results certainly sound quite intuitive, but I am sure that it would have been easy to argue for results going the other way.  It is nice to have some robust results and a clearer idea of the properties of blogging.

I was surprised by how similar our results for Russian and US blogs were. Those are really two different cultures on LJ, with US being mostly teenagers and grad students, while in Russia, they are more like the standard WordPress user of the US, i.e. thematic blogs. I would like, in a further iteration of this research, to classify blogs by type so as to examine further differences in their dynamics.

Second, the low apparent rate of reciprocation, around 10 to 15%, was lower than I expected. Bloggers reciprocate only about 10 to 15% of the time and this does not seem to vary much from stages to stages in the life of a blog (i.e. how long they have been established). I wonder if we would find the same if we gathered data on the establishment and breakdown of links at the individual level. This is possible, but much more data intensive and having access to company data would be helpful so we know who started the relation.

For more details, the paper is available at SSRN and RePEc.


5 thoughts on “Revised version of “The role of reciprocation in social network formation, with an application to blogging”

  1. Cool that you have a blog Alexia. Seems a bit heavy for me (at least the mood I am in write now). I will have a good read later.

    1. Excellent! I like your idea for a blog, and your research looks interesting. From the point of view of social network analysis, nerd culture is very interesting as their networks are so tightly knit and yet generate so much passion, disagreements and debates. Perfect so as to obtain a bit of dynamics in the establishment and destruction of links among people!

      I am thinking of selecting a community on LJ for study along the same lines as this paper, except that I would also monitor the establishment of links among members. Very data intensive, so not sure I will really have the time or budget to do this. Did this for one period, and it took me two weeks to process the data!

      1. Data processing is always so lengthy and expensive! When I did some research on fanfiction it took much longer than I had anticipated because of that. I had to code, and then translate my codes, and then compile my codes, and and and…

        I think LJ would be a great focus for your research. There are so many communities though! How would you pick just one?

        1. “I think LJ would be a great focus for your research. There are so many communities though! How would you pick just one?”

          I don’t know, I would probably start by choosing at random among communities that have been recently updated with a post (cf. community section) and then work from there in a pre-determined fashion so as to eliminate selection bias. I used to be quite active in some communities there, but they are long dead or have horribly devolved.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s