The trouble with five-star rating systems is that, in aggregate, there’s no way to get an idea of the distribution of ratings. So here’s a great idea: rendering those ratings as a color-coded distribution, in the manner of a sparkline. Now I’m looking for an excuse to implement this idea somewhere.