Book Ratings Considered Harmful

“I don’t think there is such a thing as a bad book for children. Every now and again it becomes fashionable among some adults to point at a subset of children’s books, a genre, perhaps, or an author, and to declare them bad books, books that children should be stopped from reading… It’s tosh. It’s snobbery and it’s foolishness. There are no bad authors for children, that children like and want to read and seek out, because every child is different. They can find the stories they need to, and they bring themselves to stories.”

—Neil Gaiman

I’ve been thinking a lot about how we talk about books. Partially because I’ve been reading a lot of books, partially because I’ve started listening to a lot of authors (especially marginalized authors) talk on Twitter, and partially because I’m reading so many truly excellent books and I need everyone to read them so I can discuss them with you. I have So Many Thoughts.

And, as is my nature, I have So Many Thoughts about how we engage with rating, reviewing, and recommending books. And I’m noticing some things I think we could, honestly, do a bit better.

I’m toying with the idea of writing up some thoughts on a bunch of books I read in the last few years, so I wanted to talk a bit about how I think about reviewing, recommending, and rating books, because I think that’s a useful foundation for understanding me talking about books. Or maybe it doesn’t matter and I’m overthinking it. I’m known to do that. I may also be imagining people to care more than they actually do about nuances I care about. I’m known to do that, too.

And I guess the proper place to start is to describe what I think I’m seeing, or at least how I’m interpreting what I’m seeing.

What’s the Point of Reviewing/Rating Books?

Why do we have sites like GoodReads or TheStoryGraph? What’s their point? Why use them?

I think the point is to help people find books they’ll enjoy reading. I keep thinking about it like product reviews: hear from an unbiased (or at least, with different biases than the author/publisher) party about something before you invest time or money in it.

I think they exist to help us identify, surface, and share “good books”.

We do this by assigning star ratings to them; a one star book is to be avoided, a five star book is a strong endorsement. We recommend books by saying “if you like X, you’ll probably like Y”–sometimes human-curated, sometimes algorithmically determined based on what people who bought that also bought. We review things to tell people if they’re a good book, and what we enjoyed about them, or if they’re a bad book, and what was wrong with them.

Everything distills down to that binary distinction: good book, read it; or bad book, avoid it.

What’s the Problem?

Surfacing “good books” is great and all, but it begs the question: what’s a good book? Well, a book you’ll enjoy. And that’s a tricky thing to navigate, because it turns out people are different and like different things. A million years ago, in high school, I was taught that the story you read can be as much about what you bring to it as it can be about what the author wrote. And I think that’s a neat way to think about reading; it’s personal, it’s individual. Two people aren’t going to get the same thing out of the same story. Reading a story, by necessity, changes it. Talking about it changes it further. I like that idea, the living story.

But that idea is incompatible with how we’re talking about books. We try to distill the entirety of a book into a number of stars, or trying to say what people will like based on liking things we consider similar, but those discard so much important context that the results are pretty hit-or-miss. You never quite know what someone else brought to the story that made them rate it that way, or what made them decide it was similar to another book.

Consider, for example, that I like Cory Doctorow’s Little Brother; you may be tempted to recommend Orwell’s 1984. Little Brother is heavily inspired by 1984, after all! There are even 1984 references in it. I’ll love it. But that recommendation fails to consider what I actually enjoyed most about Little Brother: it wasn’t about the anti-authoritarian message, or the dystopia, or anything like that; it was the rhythm of surfacing problems and then finding solutions to them. That’s what drives the plot forward: problem, finding solution, repeat. So a more appropriate recommendation would be Andy Weir’s The Martian, even though that has so much less to do with Little Brother’s main themes than 1984 does. It turns out I wasn’t in it for the main themes.

Similarly, consider that I love queer books. I read so many queer books. But I love reading about certain parts of the queer experience much more than other parts, and depending on what the author likes or wants to write about, that may or may not be in their book at all.

Does that make those books bad? Well, by some definition, yeah. I didn’t enjoy them at all. If one star is “I didn’t get any enjoyment from this at all” and five is “I read this over and over again because it makes me happy”, they would get a single star.

But that’s not useful, because someone else may love those books. We’re allowed to like different things. It’s not a bad book, it’s a bad book for me.

This mostly would be fine, if suboptimal, except it turns out ratings and reviews are how a lot of people buy books. And so a lot of one star ratings on, say, GoodReads can impact your sales.

So if a lot of homophobes read your book and are surprised it has Gay Stuff in it, you’re going to be in for a bad time. Or if racists are surprised that Black people exist in your fictional world.

Things can also get mean. I’ve heard some authors just won’t read GoodReads reviews anymore because who wants to hear people talk shit about something they worked hard on? It is astounding the lengths people will go to to tell authors that they hated their book. And I never really understood that impulse. What good does that do?

And then I started generalizing that a bit further. What good does it do to tell people that you hated a book? That you got no enjoyment out of it? I can see the benefit in letting people self-select out, so people who are unlikely to enjoy something know that up front, but that’s not “this book is bad”, that’s “this book contains X”. People can decide whether containing “X” makes a book enjoyable or not.

So, to recap: ratings and recommendations don’t surface enough information to be the best solution to the problem they’re trying to solve, and authors can pay the price for that, especially marginalised authors.

So How Should We Talk About Books?

I’m going to propose that there’s a better system for talking about books. It has already been trialed, you can find it in most libraries. The fine people at have been executing on it splendidly. Letterboxd sort of does it for films.

It’s the good ol’ fashioned “list of books on a theme” approach.

It is a way more productive signal for me that the book is in the “YA Books with Supportive Parents” list than it is that the book is “like Simon vs. the Homo-Sapiens Agenda”. It tells me what part of the book you’re highlighting, and if that’s a thing I care about (or care to avoid), I can select or select out of that experience appropriately. And when I’m looking for something specific, say found family stories you can conveniently index on that. But it also means that if, for example, a book is on a list of a topic I know I won’t enjoy–say, “Protagonist Unlearns Toxic Masculinity”–I know I can avoid it and that’s an equally useful signal for people who actually enjoy reading books on that theme.

What I love most about this is that it acknowledges the complexity of stories, and builds in the understanding that stories can mean different things to different people. There’s no attempt to reduce stories to a single data point, a binary “buy” or “don’t buy”. But a pleasant side-effect is that it’s at least a little less geared towards inherently negative interactions about books. Yes, you can make a “Books I Hate” or “Books Not Worth the Paper They’re Printed On” list, but you have to actually try to be negative. Whereas clicking 1 star can feel like you’re saying “I didn’t enjoy this” but be interpreted as “this book sucks, don’t buy it”. It’s much easier to accidentally be negative without meaning to.

Obviously, because I’m me, I have a domain purchased and some notes and sketches about how to build out software that would aid people in curating these lists. It turns out the trickiest part is having a reliable catalog of books that people can search through to easily add to their lists; capitalism’s weird. Whether I’ll actually ever build the software remains to be seen, but it at least has made it onto the ever-expanding list of “things I would do if I had the time and energy”.