Unfortunatedly when I looked today at the rating pages, I saw: The comment for Geo::Shapelib from Jive Chang. Gosh!... I even agree with your calculation to some point, but there are also mathematical models, that clearly show, that a too low Signal:Noise ratio will kill the signal.

I am a regular reader of cpanratings, and while there are some unuseful ratings and some ratings which I wished I could rate themself, the majority of the ratings is quite useful and informative.