How about finding new books by mapping who thanks who?

14.48, Tuesday 5 Jan 2021 Link to this post

I remember reading The Red Men by Matthew De Abaitua, which is near-future sci-fi (and, in 2021, barely sci-fi at all) about AI and robots, simulated worlds and algorithms, cults and, um, marketing agencies - it’s an amazing book, highly recommended - and all the way through I was thinking: what’s amazing is that I don’t know this guy. The themes and the twists on ideas were just so electric-yet-familiar, like it was directly for me and the people I knew.

Then I got to De Abaitua’s Acknowledgements on the final page and there, thanked in the first line, was a key person at his publisher: artist and writer James Bridle, very much responsible for shaping what my world was and is thinking about. At the time Bridle and I were working in physically adjacent studios. That makes sense, I thought.

I’ve just had a similar experience:

After reading The Institutional Revolution by Douglas W Allen (see my list of favourite non-fiction read last year), I’ve been recommending it far and wide. And I just looked at the back cover, and the book is blurbed by Tyler Cowen of Marginal Revolution – not someone I know personally, but the blog is legit one of my favourites reads. It wields economics like a scalpel for a incisive perspective on all kinds of topics.

Can we flip this?

I would love book recommendations based on acknowledgements and endorsements, rather than reader ratings.


TWO BRIEF ASIDES.

The best line in The Red Men is this one:

‘I’m giving you a direct order. Take the drug!’

‘This is not the military, Bruno. We work in technology and marketing.’

‘We work in the future!’ screamed Bougas. ‘And this is how the future gets decided.’

Technology and marketing and hubris. You can see why it spoke to me.

Secondly, Matthew De Abaitua was kind enough to contribute to my old 3 Books Weekly newsletter back in 2016. Here, with blurbs, are his three book recommendations.


CiteSeer indexes scientific papers, and lets you explore by listing papers that also cite the one you’re currently reading. (And then ranks the papers by how many times they are cited, and so on.)

CiteSeer first appears in my notes back in 2001, and it was a revelation to be able to research this way. Search for terms, rank papers by reputation, explore the siblings in the citation tree to discover alternate points of view, etc. There used to be a graphical explorer were you could tap and zoom around the whole citation graph. I can’t find it now.

Google Scholar does something similar. For example: here are the 86 publications that have cited Mind Hacks.

There is a whole field of citation analysis, and you can imagine being able to look for high-level patterns in the scientific literature like bubbles of groupthink, or multi-year controversies and schisms. Or perhaps you could build an early warning system for new scientific paradigms emerging – or spot old ones creaking.


What I’d like is CiteSeer but for books, making using of the softer signals.

Who does the author thank? It’s a highly meaningful signal: if they thank someone I’ve heard of and rate (and especially if it’s someone I know), I am going to look at this book in a different way. Debut novels often thank other authors who have helped them along the road, but it’s the non-authors who are thanked that I find particularly interesting. I always like checking the Acknowledgements page.

Who blurbs the book? Book blurbs aren’t blind marketing. No author wants to endorse something terrible, and that reputational risk means that blurbs are meaningful. But also, an author blurbing another author’s book means that they know each other in some way. They are in the same idea space.

(It would also be neat to extract the bibliography for non-fiction books. I think the utility of bibliographies goes beyond the citation graph. A bibliography tells how close this book gets to primary sources, which is a clue to its originality or how academic it is. And I also get a sense of the overlap with ideas that I’ve already encountered.)

I’d like ways to explore the network of influence and gratitude.

  • Who else thanks the people who are thanked in books that I love? And who do I know?
  • What books are blurbed by people whose blogs I read, or by people who appear in my email inbox?
  • Can this be joined to the annual publishing schedule, so that I get a live feed of upcoming books with which I have a strong connection?

More interesting data to be revealed: Who are the “dark matter” agents, commissioning editors, and mentor authors, i.e. the people who crop up the most, in the most popular books?

Recommendations based on this wouldn’t be based on the taste of the average reader, as star ratings are, but instead reflect the book’s position in idea space, and what is adjacent to - or deliberately distant from - what I’ve already read.

It feels like the kind of thing Google Books or Amazon/Goodreads could do.

Google and Amazon both have a massive corpus of scanned books, an existing database of known titles and authors, plus the requisite machine learning chops… it would be a good side project for someone there. Let me know if you have a go and want some early product feedback.