4 Mar 2009
Many areas of Internet scholarship make strong – and often erroneous – assumptions about patterns of Web traffic. Still, there has been little comprehensive research on how online audiences are distributed, and even less work on how site traffic changes over time.
Using three years of daily Web traffic data, and new models adapted from financial mathematics, Matthew Hindman examines large-scale variation in Web traffic. These data show that Web traffic is highly heteroskedastic, with smaller sites having orders of magnitude more variation in the relative number of visitors they receive.
These consistent patterns allow us to provide reasonable estimates of how likely it is Google will still be the most visited US site a year from now, for example, or the odds that a new site currently ranked 50 overall will break into the top 10. Despite constant churn in online traffic, the audience distribution for both the overall Web and for subcategories of content is extremely stable, limiting the number of prominent outlets.
These results challenge many accepted notions about online life. In particular, the talk will discuss what these traffic patterns mean for the openness the online public sphere.
This public lecture was the opening event for a two-day workshop ‘Internet and Democracy: Lessons Learnt and Future Directions’.
This lecture was co-organised with the Berkman Center for Internet & Society
and the Reuters Institute for the Study of Journalism, in partnership with the Journal Policy and