Ryan MacMillan

We’ve installed the (quite useful) Popularity Contest plugin. It tracks all sorts of variables, and combines and weights them to come up with an overall “popularity” score. Here’s how it’s currently set up:

This is leading to some peculiarities. I’m not convinced that Iain’s “p-books are still cool” article is, in fact, the most popular post on the site. But this could be jealousy.

My reason for playing around with this comes - in part - from my interest in blog assessment.

I’ve come up with a simple (very rough-and-ready) model of influence, as illustrated in this slide. As I understand matters, people (or “blogs” or “information sources”, or “vertices”, or “nodes”) derive their influence (and exert it through) popularity and/or authority. There’s a similar (but less-well defined?) relationship between popularity and authority.

Current systems for influence analysis use a variety of (much more complex and thorough) methods to understand networks of influence, but doesn’t it come (ultimately) down to something like this?

RMM’s website is growing increasingly popular: simple log analysis shows our monthly visits growing from around two thousand when we launched to around twelve thousand last month.

This is all fine and dandy (and, like most low-to-high graphs, rather satisfying). Popularity is good. If we get the numbers a little higher, we might be able to sell some advertising, or sponsorship on the site (or, at the very least, sign up to Google’s Adsense programme).

But popularity on its own isn’t really enough. We want to exert some influence. After all, some of you reading this are - when it comes to it - rather important people. If you trust us, and (by extension) what we say, we can be said to be exerting some authority.

Does all this make sense? I wonder. Feedback gratefully received and argued with.

In the meantime, consider this:

Edit: 3 May 2007

Well. The plugin never really worked for us — What we were seeing on Google Webmaster Central, Yahoo Site Explorer, our log analysis and our Google Analytics and ClickTale and our eyes & commonsense were telling us, and what the Popularity Contest software was saying were completely at odds with each other.

The idea is good. Something in our implementation may have been wrong.