The scale from 1 to 10 blows:
Numbers are convenient for generating statistics (Anime News Network calculates a nifty Bayesian estimate), but if the stats are based off crappy data, they won't mean much. These flawed stats end up having no absolute meaning, so they're only used for relative comparison a.k.a. ranking, a total waste of time. If you've ever tried making an top 10 anime list, you know exactly what I'm talking about.
Grouping is much less futile than ranking. At the end of the day, you just want to know if a show is good. That's why I threw out numbers and introduced a more intuitive system:
Anything with a + is recommended, and anything with a ~ isn't. Could it get any simpler?
For various reasons I may occasionally use a numerical or alphabetical rating scale. In those cases the ratings translate strictly by this table:
My grading scheme is not open to interpretation. It has consistent definitions, and if you ignore this, you may end up missing a great show or worse – watching a really crappy one.
9052 statisticians agree that my system rocks