This is a bit inside-baseball, but I've decided to change how I score novels and have gone back and re-balanced all my old scores based on the new system. In the old system, I was trying to fit all of human writing into a 1 to 5 star scale. So at 1 star you would have a things like NyTimes editorials and crude anti-semitic comments on YouTube videos, while at 5 stars you would have works of transcendent genius that set off cascades and feedback loops in your skull. Realistically though, I don't read or review much 1-star material since it's pretty easy to identify as bad/useless and then avoid. And on the other end, there are not that many works of genius that exist and are readily identifiable from afar. So the scoring scale ends up being between 2.0 (terrible stuff that managed to make it into my reading queue) and 4.0 (really good stuff that I've found and really enjoyed/admired). It's similar to the problem that video game reviewers ran afoul of, where most reviewed games would end up being between 7 and 9 stars out of 10. They don't really go above 9 stars, since that would imply a perfect game, and they don't really go below 7 stars since that is the land of games that don't run at all and have technical problems or are obviously terrible. (Also, companies/fanboys would bitch and moan if there game got below a 7, so that was another incentive not to).
Anyway! To solve the problem, I've copied Tom Chick's solution and decided to use the whole range of stars available to me. A bad book is going to get 1-star, and a really good book will get 5 stars. And if there are things which theoretically should be off the scale, (the NyTimes editorial, the book of genius), well, I will the deal with the outliers as they come either by going off the scale or just bounding the outlier to the ends of the scale.