How much online influence do you have?

A number of years back, I read Malcolm Gladwell’s book The Tipping Point which presents his thesis on why things “go viral”. Its a very powerful book, even if it has received some interesting criticism. Bells started ringing for me as soon as I started reading a recent publication from a group of students at Oxford University titled The Dynamics of Protest Recruitment through an Online Network. The paper is a bold attempt to monitor the spread of information about a protest on Twitter, and it draws some interesting conclusions based on the results.

Perhaps most importantly, these sorts of studies will help the marketing army to re-evaluate how they carry out their campaigns, but I also hope that it convinces some people that they should just give up tweeting because nobody is really that interested.

The researchers behind the paper focussed on the surge of protest mobilization that took place in Spain in May 2011 and resulted in a fairly large camped protest in the centers of many of the major cities in Spain, including the Plaza del Sol in Madrid. During the period of a month, starting on the 25th April 2011 (20 days before the mass mobilisations started) and finishing on 25 May 2011 (3 days after the national elections), over 87 000 Twitter users were tracked for nearly 600,000 protest messages. Through the Twitter API, it was possible for the researchers to determine who each user follows and who in turn follows that user. By following how messages spread through this network of relationships, they determined which nodes were most effective at user ‘activation’.

There are some interesting results from the research. Naturally, the group found that the first people who received initial messages about the protests were very quick to be ‘activated’. This is because these messages were most likely initially spread between people who were relatively close and probably already involved in the recruitment process. Moving on from this, the researchers discovered that various recruitment bursts would take place as messages spread.

This came down to a process known as ‘complex contagion’, which effectively takes place when an individual finds that many of their closer neighbours within the network join in a short space of time. This is a bit like a form of online peer pressure. When a bunch of your friends start tweeting about something, you’re likely to start tweeting about the same thing pretty soon.

A central finding in the report is that while users with many followers are important, the spread of information very much depends on how central these users are within the network. If users with many followers are closely connected, they act as bridges of information because they connect multiple ‘local’ networks. This facilitates complex contagion. This means that in order for a meme to spread effectively, a core group of highly connected individuals need to be sharing the information, and each of these core groups need to have groups of followers that are also interconnected. One of the researchers states that reciprocal connections are more relevant to the recruitment process because they are more likely to reflect genuine offline relationships.

This all seems to make some sense, but another interesting point arises. The paper finds that people who seed an idea, do not need to be core to the network across which the idea spreads, they just need to belong to a local network that is connected to a core user. This means that as long as a local network is well activated, and that network is connected to a core user, the idea is likely to spread.

While much of this research may seem fairly obvious on the surface, it very much confirms many of the fundamental tenets presented in the Tipping Point. For one thing, it re-affirms some of the attributes of the different types of people involved in spreading a meme. Gladwell has defined ‘The Law of the Few’, which focusses on three types of people that are critical for the spread of a message. Certainly this research picks up on the notion of “Connectors”, “Mavens” and “Salesmen”.

The research paper will be very interesting to marketeers and may help to better understand interactions within social networks. However, it is important to remember that the paper is really built around some fairly limited parameters.

To begin with, it only studies one particular incident inside of one social network. In reality, the number of messages studied is still fairly small, and other messages within the network are essentially ignored. This makes me think that the research is very open to confirmation bias. That’s not to say that it offers a valuable insight into a particular incident. I just think that you really need to be cautious about drawing too many conclusions from a paper like this.