Better Book Reviews

I think the book review ecosystem is broken. Here's a quick story explaining why.

My day job is co-leading Roote, an online school for systems thinking. We'd love to get more students to apply to the school, so we're thinking about ways to get more eyeballs on our site.

One way to do this is through content marketing like writing book reviews. For example my book review of The Sovereign Individual has 20,000 views and gets roughly fifty views a day.

Sounds easy, right? I should just write some (hopefully great) book reviews and direct folks to Roote.

I don't think it's that's simple.

I. Problems With Writing "Yet Another Book Review"

Let's say I decide to write a book review about Thinking in Systems, a delightful book on systems thinking by Donella Meadows. The traditional way to do this would be to write the book review and try to beat the other book reviews on SEO and distribution.

Lots of internet bloggers do this (partially because they want to write a great book review and partly because they want to drive readers to their site). Here are examples from two internet bloggers with names that start with J and C: James Clear's three-sentence book summaries or many of Jason Crawford's posts. On the startup side, apps like Blinkist have raised $35M to provide book summaries-as-a-service for $8/month. SparkNotes was the original summary site (and still exists today!).

This doesn't sit well with me. Is it long-term helpful for the epistemic commons to have a google search page that looks like this?

A bunch of isolated summaries that don't reference each other, and don't create something bigger than themselves. Hooray.

I call this issue "Yet Another Book Summary" (YABS) after YAML, Yet Another Markup Language. Instead of creating some sort of evolving shared artifact, we're left with dozens of separate summaries. What does the world need more of? Yet Another Book Summary of course!

Why are YABS a problem? In the short-term, they're not. Most of the book summaries of Thinking in Systems are fine. But we don't know what we're missing. Imagine if all book summaries built on each other in some way? Our collective book summary artifact would get better over time, like a good Wikipedia page. Instead, book summaries exist in isolation of each other. It's like scientific papers not referencing one another. Or if ants just got food for themselves without leaving behind pheromones. We're creating knowledge in isolation without contributing to the hivemind.

Why do we create YABS? It's mostly a function of three things: incentives, technological affordances, and coordination.

First, incentives. We're incentivized to pull eyeballs to our sites and monetize them. Of the links in the google search above, tries to get us to pay for their exclusive community and wants us to sign up for their newsletter. (Don't forget to sign up for Roote. 😉) Plus, there are all of the social incentives. If I create a new Wikipedia page, I get no credit. If I create a book review that is shared widely on Twitter, I get more followers and my social standing increases.

Second, the current state of digital tech forces us down this YABS route. If we want eyeballs, we need to have high SEO, a large newsletter list, and lots of followers on Twitter. The content gets optimized around those constraints. In addition, there aren't many good alternatives to writing a review on your own website/Substack/Medium. The only other option is to edit a Wikipedia page. (As part of writing this article, I tried to create a new page for Thinking in Systems but didn't succeed at making it through their review process yet.)

Third, coordination. It's difficult to coordinate outside of yourself (and outside of Wikipedia) to write a shared review. Who makes the call on what gets into the final article?

We know we have a problem. Book reviews exist in isolation of each other. This happens because of poor incentives, old technology, and weak coordination. We end up with an internet full of Yet Another Book Review (and Yet Another Internet News Story and Yet Another X and...) instead of building knowledge for the long-term.

What can we do about it?

II. Solution: Not Another Book Review

What does a good solution look like? What would be a good Book Review process?

  • It needs to have positive externalities. (It should provide at least a bit of value to the epistemic commons.)
  • It needs to be a living artifact. (It should be updated over time.)
  • It needs to work. (It should align with human incentives of meaning, connection, and basic needs. It should get lots of views, for example through good SEO. It should be easy to coordinate.)

Concretely, this could look like a couple of different things:

  • A Wikipedia page. From my experience, Wikipedia's constraints might make this difficult, but at least some of the Not Another Book Review (NABR) output should likely live there.
  • A Wiki-style page on a different wiki, like P2P Wiki. I think this is a good balance of wiki-style editing with more freedom.
  • A page on GitHub or GitBook.

Or, more creatively:

  • A meta page that contains links to a prioritized list of reviews. (Like something in between Goodreads reviews and Page 1 of a Google search.)
  • An internet norm that encourages people to add to a Wikipedia page whenever they write a book review.
  • A deeper link mapping between an rBook representation of the book and a Wiki representation.
  • A review with embedded Orbit prompts that pay the reader BTC as proof-of-memory.

Whatever it looks like, there's another outstanding question: how does NABR leverage the crowd? Is it something that individuals chip-in to as they go about other knowledge work? Or is there more of a formal community based around it, like a #BookClub Discord group for each page?

I'm not sure the answer to any of the questions above. To learn, I'll start with an experiment: I'm going to run a Monthly #NotAnotherBookClub that will produce a  networked Not Another Book Review as an output each month.

We'll going to start with Julia Galef's new book Scout Mindset. It doesn't have any summaries yet! #NotAnotherBookClub will meet from 9am-10am PST on Thursday, May 27th. Please read the book before then. Register here.

Thanks and hope to see you there!


  • I hope that solutions or ideas explored here will have reach. I hope that solutions for the Book Review Commons will apply to News Commons, Recipe Commons, and Medical Commons and other forms of information.
  • In addition to extracting knowledge from books, we can think about structuring books in a way so networked knowledge is easily extractable from them. Books should change too.
  • News is full of Yet Another Syndrome. Mainstream media outlets just repackage the same information for their subscribers. Medical advice and recipes are pretty bad too.
  • It would be nice if a solution had reach for non-information domains as well. For example, we have Skype and Zoom and Microsoft Teams. Wouldn't it be better to have more protocols aligned around a Shared Outcome (of good video calls) instead of constantly reinventing the wheel?
  • There are many forms of Yet Another:
  • There's also an issue with the mindset that Google SEO creates in me. It pits me against Sloww. Either they get clicks or I do. But in fact, I like Sloww! They seem to be doing good work and I like their summary. It would be nice if this coopetition was more legible.
  • Youtube video has even worse problems than text. We get a bunch of different Youtubers repackaging similar content, competing on their lighting and camera setup. Womp.
  • Wikipedia is such a beautiful part of this puzzle. I see it as a shared third rail that left/right/sideways can always turn to for information. I looked at it often during BLM protests.
  • Wikipedia is also so different than the rest of links on your average Google search page. It's collaborative created, not owned by some org who is secretly trying to shill their stuff.
  • I'm not exactly sure how this should relate to Building a Second Brain and Progressive Summarization, but it should. One initial language change: Building a Third Brain—the internet's brain.
  • It's helpful to think from the perspective of an ant colony. What small, low-cost "trace" can we leave and build on as we navigate the epistemic commons? Computer tracking of links and page views is a super cheap version of this.
  • We need more digital librarians to do this work. Like Facebook's 40k content moderators, but for positive psychology instead of negative psychology.
  • Google itself is likely an issue here. Search Engine Results Pages are a platform, not a protocol. How much should SEO be based on recency? idk. But for now, they choose, not "us."
  • One revenue stream for Book Reviews is affiliate fees for book purchases.
  • I've started to try to see knowledge at two levels: As the object-level knowledge, and as a through-time "ordering" of those snippets of knowledge. A book is both. But more than anything, it's defining an ordered path through a set of nodes.
  • The mechanism for solving this (constantly buildable networked knowledge), will be valuable for humanity.
  • This is fundamentally a curation problem. I'm beginning to think of curation as a form of computation. How can we convert information from one form into a more usable form?
  • The design space for an epistemic commons hasn't been explored much. It's pretty constrained by a fitness landscape that is warped by financial incentives.
  • LessWrong has a good book review policy: Book Reviews on LessWrong are different from normal book reviews; they summarize and respond to a book's core ideas first, and judge whether you should read it second. A good book review sometimes distills the book's ideas so well that you no longer need to read the book.
  • One way I think about this problem is that Google gives us tons of results for a given search. But how can we (as a hivemind) get really really good at making the top 10 links?
  • Any system of collaborative governance will run into Fukuyama's balance between a modern state and accountability. In this Wiki context, decisions need to be made while still allowing for the crowd to have voice.
  • In this article, I don't give enough credit to the positive impacts of book reviews as is. For example, book reviews help get ideas into brains, which then allow for future knowledge production. It's not purely in isolation.
  • Similarly, some of the incentives are actually just desired signaling systems. We want good creators to receive karma for their posts. We want them to receive financial rewards too.
  • Similarly, there are some worse is better dynamics where a bad book review ecosystem forces more folks to do the "hard work" of actually reading and reviewing the book themselves.
  • Not sure how GPT-3 and Yahoo Answers dying relate to this, but they do.
  • One of the dominating narratives right now is the Passion Economy/Personal Monopoly. It incentivizes these isolated verticals. It says: let there be a top person in a niche and let everyone read/watch them, because they're the best. I think this is the simplest answer to a coordination game to produce good content (because it just involves one person). But I don't think it's as good for our epistemic commons.
  • I think of "contributing to the epistemic commons" as part of our Distributed Bill of Duties.

See Also