Anyone have experience with Zipf Distribution?
 

Specifically when it comes to sample data models?

I'm trying to work out the basic sample data required for somewhat realistic results in terms of clicks vs movement/rankings.

I'm currently basing it on a 1000 click ratio with a 1.5% variance buffer (to avoid jitter) and then applying a 15% threshold a link should meet vs. the link above it, for that link to shift higher in the list.

Would this work well do you think or should I refine the numbers? :helpme