Wednesday, January 28, 2009

Virtual Corruption Ain't So Virtual

I've been playing with something called "The Corruption Calculator." It's a tool that can translate data into corruption scores for entities large and small.

I've been thinking about how Webcomicland would score. Why not? It has defined borders, identifiable citizenry and plenty of metrics.

It even has government types: "patchwork authoritarian districts," "authoritarian city states" and "tribal," which is the best match I can find for collectives.

There are two challenges. Defining corruption, and data collection.

For data collection, it looks realistic to collect data on any given unit, or a bunch of them. To do all of them would be impossible without a budget to hire and train staff. To do just one web site, it would require that enough data be available. That would leave out about 9000 titles, roughly, but leaves in over 500.

Defining corruption is interesting. Here is a list of candidates I dreamed up:

  • Playing fast and loose with the truth, where the truth can be verified;
  • Playing black hat games with your SEO;
  • Gaming ads through click fraud;
  • Using traffic generators to make it appear you have substantially more traffic than you do (surprisingly common: see fake traffic ). (Getting easier to detect, too.)
  • Involvement with lists and contests which have not been designed for fairness, but which present their results as definitive (when they are merely subjective)
  • Fabricating lots of Twitter followers to make you look "popular" (I include this because so many cartoonists use Twitter. Many well-known ones have been dragged into schemes by allowing the fakes to follow them too despite knowing what is happening.)
I can't predict if this will go anywhere. It's a fair bit of work to put a "score" on things I mostly know, but measuring different sub-populations might be instructive. Still, I'm interested whether these or other metrics would be considered valuable. It's an opportunity to express the standards you value most.