Quantitative Book Rankings
This is a response to Autumn’s post about Murakami.
I don’t disagree with the limitations of the five-star system; in fact I have some half-baked alternatives. Full disclosure: I’m the Murakami pusher in the original post.
The problem is trying to solve everything with computational thinking, hence the weight that star reviews have everywhere — whether a taxi driver will pick you up, if a small business’ products will be “discovered” and what information algorithms display to you.
Goodreads, an Amazon owned platform, is always going to tend towards the quantitative. But that’s not how books have been sold in real life. Once upon a time you could walk into a local bookstore and have a conversation with a knowledgeable employee about what you wanted. “Try this for a lazy Sunday afternoon read, or this for something challenging yet worthwhile”. The conversation was never, “this book got 4.7 stars!”.
You can replicate this on an online platform with a tagging system, but it takes a lot of work to get the information architecture right. The first tag would be how strongly you recommend a book, perhaps: yes, yes with reservations, kind of, not really, absolutely not.
The obvious objection is that such tagging maps directly to a five-star rating system. It doesn’t. With stars, there’s no baseline. Is 3-star review good and 5 reserved for a masterpiece? Semantic tagging avoids that.
Other default tags would be needed for context: easy read, challenged my worldview, unpleasant but rewarding, entertaining, etc.
There should also be custom tags. Maciej Cegłowski has an excellent talk on tagging. It’s worth noting that this talk is ten years old and used not-new-at-the-time tech to achieve great results from community tagging.
This makes it nearly impossible to computationally rank and recommend books. But for dedicated readers willing to apply filters, search and dig a bit, you can get context-rich and meaningful reviews.
Another feature, I’d like to see on book review sites is an automatic reminder to revisit a review every year. Writing a review right after you finish reading a book is fine, but the true mark of “serious” books is whether they age well.
I don’t know of a large-scale platform that does all of this. I do something of the sort for myself in my personal notes. It’d be fairly easy to build something in Jekyll that other people could easily copy and paste, but that’s still a small audience that is both tech savvy enough to customize a Jekyll setup and into books enough to bother with it.
A commercial solution seems fleeting. You’d need to either have paid memberships or sell books via Amazon affiliate links. Once that starts, the temptation to start inflating every review and hiding overly negative ones becomes too much to ignore.