Singapore.
The next presenter at ICA 2010 is Jonathan Zhu, and his focus is on the dynamics of blog participation. The big change of Web 2.0 is that it threatens the traditional division of labour in publishing, and that more people can become active participants in online publishing - and many have, but how sustainable is such activity? What predicts whether people who start blogs continue their blogging activities?
In other words, how long do bloggers stay active before quitting? How do we define 'active', how do we measure the duration of their activities? Jonathan conducted a survival rate analysis which took both these questions into account - examining two time windows: a stringent criterion which required blogs to be active at least once per month, and a more lenient one which required a post at least once in six months, to be counted as still operational.
And if blogs are found to be live or dead in this way, is there anything in their track record which may explain this? For example, what year they started (i.e. during the early blogging boom, or more recently), the frequency of a blogger's previous posts, the time they spent on blogging, or the feedback they received? The present study examined this using data drawn from Sina Blog in China, whose URL system made it particularly easy to engage in a random sampling from a large population of blogs (the URLs are numerical, so a random digital search approach similar to random number dialling on telephones can be used).
From found blogs, the number and length of posts, number of embedded images, number of comments, and date and time of posting were extracted. The study found a standard power law distribution of the survival rate: of the blogs which had at least one post within the one-month time window, some 70% of bloggers had stopped blogging after just one day (or more precisely, one post), while only 5% of them had been active for three years or more.
The effort spent on blogging (length of posts and number of images) was shown to be a significant predictor of continued activity, as was the time available for it (measured by examining whether bloggers typically posted at all times of day, or just in specific timeslots); feedback from readers was not a predictor, interestingly. Such an examination could also be conducted for Twitter or other social networking sites, of course.
