The Math of the 5-Star Rating System
I’m not comfortable assigning star ratings to books, but I find it helpful when other people do it, so I’ve decided to partake (especially on Goodreads, if not on Amazon, though they’re now under the same umbrella, which is its own problem). But I’m generous with stars, which I’ve managed so far by rating only books I can assign four or five.
I may change my mind about this, but at the moment I doubt low ratings are helpful. For one thing, except for classics, I rate mostly picture books. Should I try to ward people off with my one disgusted star? I think most people see picture books in person before buying them or act on friends’ recommendations or the reputation of the author/illustrator. They don’t need a warning from me.
And I’m not Kirkus Reviews, which serves a top-down role in the public discussion of children’s books. Kirkus Reviews and other journals are important, but I see myself as a bottom-up contributor, and I would rather signal which books grabbed me than complain about those that didn’t.
So I rate books I like. Those, for me, are the only ones worth thinking hard enough to write about. I know I’ve been talking about stars and not reviews, but I feel like a poor star rating should be backed up with a review showing cause. Even simple reviews require me to organize my thoughts, and I have no desire to tackle the work of organizing my thoughts to back up a critical rating.
I’ve decided to describe my approach to rating with this graph:
The curve is exponential. Since it’s not linear, the area under the curve between the fourth and fifth star is greater than the area from zero to four. If your rating style is exponential, more books deserve high ratings.
Still, generosity can be problematic: Just the other day, I gave five stars to a really fun, inventive picture book whose quality can’t possibly be compared to Middlemarch, my Platonic form of a 5-star book. How many books are as good as Middlemarch? Five? Ten? More? Certainly not the picture book I gave five stars to the other day.
Rather than worry about whether a book is so good that people will be talking about it 200 years from now, I’ve decided to give a book five stars if I don’t wish it were different than it is, if it strikes me as perfect on its own terms (and I like those terms). If I really like a book but wish it were a little different, I give it four stars. If I wish it were a medium amount different, I don’t rate it; I look for another 4- or 5-star book. There are enough of them, and finding them is fun.
But I said at the beginning that I find other people’s star ratings helpful. Will my liberal system help others? Sure. My ratings basically say, “This book is worth paying attention to.” And actually, maybe my graph is incorrect. I heard once that the best chess players don’t even see the bad moves. I’ve been visiting picture book walls lately to see what other picture book writers are doing. There are a lot of great picture books, but, to be honest (and at the risk of sounding self-congratulatory), there are many books I don’t really see, books that don’t enter my consciousness.
So maybe my graph should look like this:
It’s still exponential but in reverse. In this graph, the exponential curve shows an increasingly small number of off-the-chart ratings, books that deserve even more than five stars. Let’s say I don’t really see the one- and two-star books, and Middlemarch is off the charts. It gets a nearly impossible ten.