I'm ranking movies by critics ratings. Most of them have too low numbers, and thus naive bayesian ranking by avg does not work. IMDB gets away with it, but I cannot. And you should be able to see a good preview of the expected ranking even with low numbers.

So you need to check the confidence interval with Wilson, but you also need to check the quality of the reviewer. There are some in the 90%, but there are also often outliers, i.e. extremities. Mostly french btw.

I updated the c and perl versions, compiled and pure perl here: https://github.com/rurban/confidence_interval