Predicting the future popularity of a web page

Balloons in Trafalgar Square
Creative Commons License photo: wili_hybrid

New Scientist reports this week that a new tool developed at HP Labs could potentially predict the popularity of a web page in 30 days time. Essentially they say that by looking at the rate at which a web page picks up views in the first few days can predict the subsequent popularity of the page 90% of the time. It doesn’t seem too radical an idea – after all the pages which are more popular in the first few days are likely to get bookmarked more, linked to more, higher place on Google, etc.

The research focused around the sites Digg and YouTube so it would be interesting to see how it could be applied to other sites. You can download the paper online at arXiv.org.

On a similar note, I’ve found that I’ve been able to get some incredibly stunning useful information from the popularity of webpages on my site. For example, one of my posts about MSN Messenger downtime gets a lot of hits whenever MSN Messenger goes down. When the number of visitors for that page is significantly above normal, I know that MSN is actually down. If the number of visitors is normal, it’s typically just an issue with my connection or my local server. In fact, I’ve found this method much more reliable than using Microsoft’s own service status page for the Messenger service. Similarly, I found a huge spike in the number of visitors to my post on the possibility of VAT cuts straight after the recent pre-budget report. If only there was a way of exposing these statistics in a useful way!

Leave a Reply

Your email address will not be published. Required fields are marked *