The walled marketplace of ideas: a statistical critique of SSC book reviews

Introduction

Scott Alexander is an extremely popular online blogger, who for many years ran a blog called SlateStarCodex (SSC), and now runs a similar blog called AstralCodexTen (ACX). He is one of the top two most influential figures in the rationalist movement, along with Eliezer Yudkowsky, with his blogs and associated communities acting as rationalist hubs. This influence extends to effective altruism, where he is often cited and which he frequently advocates for and defends, although he has no official position within the movement.

One of the most common type of blog posts put out by Scott is the book review. These do not consist of star ratings, but of long deep dives into a book he is interested in, highlighting key quotes and ideas, and agreeing or critiquing the views of the author. He has done many of these reviews over many years, and now has almost a hundred publicly available reviews over the two websites.

This is enough reviews that we can analyze the statistics of them, and not have the results be total garbage. We can actually look at some approximation to the media diet of one of the most influential people in EA spaces, which can also give some indications of the blind spots of the community as a whole.

Who’s ideas and perspectives are being heard? What impact might that be having on the many people that read this influential blog?

To be clear, I’m not pretending to be unbiased here. I’ve read almost everything Scott Alexander put out, and I find him to be quite a skilled writer with occasional good insights. But I also strongly dislike the guy, especially regarding his anti-progressive writings. You can read this blogpost for a list of controversies by someone who’s more forgiving than I am. In particular I find his flirtations with “HBD” to be weaselly and extremely harmful, but that subject deserves a post of it’s own.

I do recognize the well has been poisoned on critiques of Scott, and I have recently written an article on why critique of EA-adjacent people in general is often substandard. In this work I aim to be as honest and accurate as possible: I will save the ranting for the end of the article. I have seen people in EA claim to be scared to criticize Alexander. I hope this fear is misplaced.

Boring methodology stuff

Feel free to skip this section if you only care about results.

The spreadsheet containing the analysis is contained here. I finished off my analysis about September 2024, but a lot of it was done much earlier, so things like review counts might have changed a little.

I went through every book review I could find written by Alexander himself on both Slatestarcodex and Astralcodexten. This was easy to do for SSC, but a royal pain on the much worse ACX website, so I can’t guarantee I got every single review. I caught some reviews that didn’t have “book” in the title (like the series on dictator biographies), but may have missed other. Overall there were about a hundred book reviews to look at.

I skipped three reviews: The malleus mallecifurium, the Greek alexander romance, and the Arabian nights. These were all historical texts from hundreds of years ago, and thus not particular useful to this analysis. I also did not analyze “practically a book reviews”, the results of the “your book review” contests, or any reviews that Alexander has hidden behind a subscription paywall. I am confident that I got all the book reviews on the initial SSC website, but can’t be sure I didn’t miss something on ACX, as the ACX archive is substantially more annoying to interface with.

The most subjective judgement was judging the reviews for being critical or positive. Most of the reviews he did not give his own judgement, so I went for the general vibe of the review, and how much pushback Scott gave. For reviews where Scott didn’t seem particularly enthused about the book itself, but also didn’t give major pushback, I put down as “mildly positive”. I put mildly negative if there was a reasonable amount of pushback or criticism, and negative if it seemed like he genuinely disliked the book. This will be the most subjective category.

For popularity and quality, I looked at goodreads. Obviously it’s an imperfect measure, but it’s at least one that can be compared evenly. Note that the bulk of this analysis was done back in June or so, so the exact numbers may be different now.

For subject matter, I described the book, and then put it into several broad categories, like sociology, health/medicine, or history/biography. Obviously, these are arbitrary categories, and plenty of books are sort of on the border of multiple categories, but this just gives a general gist.

Checking Wikipedia for the authors was fairly straightforward. I can’t rule out that I may have missed one or two by misspelling the authors, but I’m pretty sure I got it right.

To determine the political views of the author, I took a look at the arguments being described in the book, their descriptions on Wikipedia (sometimes an authors political leanings are stated there), and their twitter feeds if available. I eyeballed it from there and rounded it off to common political categories. This is highly subjective as well.

Ethnicity is a bit of a minefield, as there is no set definition of race and the whole concept is often quite blurry. However, race still affects peoples lives, so I couldn’t ignore it entirely. I ended up with a simplistic metric of “are they white-passing?” with the mindset of, if I saw the person on the street, would I assume they were white? This wouldn’t past muster in academia, but for this analysis it will have to do. Most cases were unambiguous, but there were also cases like Nassim Nicholas talab, who is Lebanese, but has ethnically Greek ancestry. I counted him as non-white to be generous. I’ll note that there were quite a few people of Jewish background, I still counted these as white if they looked Caucasian. I have seen some controversy about whether white-passing people of Jewish ancestry count as “white”: as someone who is unambiguously white and non-Jewish, this debate is out of scope, but will note this in the relevant section.

The gender of the authors was fairly straightforward as well. In the case where the author didn’t have a Wikipedia page, I sort of judged it from their name and picture, there weren’t really any ambiguous cases. There is one author who is listed as nonbinary on their Wikipedia page, but I didn’t see any other non-cis people, although I can’t rule out that maybe there other trans or gender non confirming people that merely pass for male.

When analyzing authors, if multiple books by the same author were reviewed, I counted them multiple times.

Result analysis

Sentiment:

Let’s start off with sentiment. Does Scott like the books he reads?

Scott doesn’t give ratings, so these are all very subjective. But in general, he is mostly positive or at least unchallenging of the claims made in each book, with 61% of them being positive or mostly positive. But there was still a decent amount of times when he felt the book was bad and said so, or where he was very opposed to the views advanced in the book. So it’s clear that Scott prefers to review books he likes, but being bad/oppositional is not necessarily a disqualification.

Year of release

I’m being very lazy here, and graphing this loosely as a function of review order. Unfortunately, the ACX website is hot garbage, so those aren’t really in order.

Scott explores a decently wide time period, with a decent number of book from the 60’s to 80’s. I’ve excluded two books from the 1500’s from the analysis, which would make the range even longer. The majority of the books reviewed are roughly contemporary, with a median release year of 2013 (and seemingly getting more contemporary over time), but Scott is willing to look back at books published many, many decades ago.

number of ratings

Next, I’ll graph each book by how much they are read, based on

How popular are the books? Here we see a large scattering over orders of magnitudes of number of goodreads ratings (our gauge for how read the books are). We have a range from 400 thousand goodreads ratings (on the road by Jack Kerouac)… down to just 2 (The arctic hysteresis). For context, the most popular goodreads non-fiction book on the entire site is the diary of Anne Frank with 3.8 million ratings. This means that Scott’s reviews cover nearly the whole range on goodreads, only excluding the most famous of famous books.

From this we can see that Scott is not really sampling from the most famous and bestselling books: He’s just picking any book he thinks would make for a good review.

Rating on goodreads:

I acknowledge this is a poor metric for quality. Take this more as feel for how the average reader feels about the books.

The median rating is 4.08 out of 5. This may seem high, but the reviews on goodreads are strongly skewed high: 80% of reviewed are within 3.5 and 4.2. The books selected by Scott seem a little better than average (if you buy the goodreads reviewers), but there are plenty of stinkers in there, and it’s a wide range.

The highest rated book is “evolutionary psychopathology” with a 4.65 although it comes from only 55 ratings. This is a psychotherapy book arguing that demons are real, in case you need an excuse to take these rating with a grain of salt. Lowest is that arctic hysteresis book with a 3.5, coming from only 2 ratings. If we wanted the highest and lowest with over 100 ratings, at the top it’s “phikal a chemical love story”, a book about hallucinogenic drugs, while at the bottom it’s “the cult of smart” by Frederick deBoer, who makes a leftist case against the education system.

Subject matter:

I had to round a lot of these out, see the data for a little more detail.

Scott is into “big picture” nonfiction, with sociology/anthropology being the most common topic. There is a decent amount of historical/biography stuff, greatly boosted by the ACX series discussing various dictators. As is to be expected, Scott commonly reviews topics associated with rationality and EA, and as a psychiatrist, reviewing health and medicine makes sense as well.

All in all, this is a fairly broad slice of the non-fiction book market. However, it’s impossible for one blog to cover every topic, and there are some notable subjects missing:

When it comes to science, Scott focusses almost entirely on social sciences, not the physical sciences or engineering. He might call himself a “left hereditarian”, but you’ll learn very little biology from reading this blog. Whether from a lack of interest or qualifications, you are not going to learn things like the current state of nanotechnology from reading SSC. This is probably the biggest change from early Lesswrong to SSC, as “the sequences” were very big on all types of pop-science.

Similarly, there is fairly little on the finance or business side of things. He does not seem particularly interested in imparting financial or organizational advice. As can be expected from an atheist, religion is fairly absent, with the exception of a book or two on Buddhism, mainly focusing on meditation. One overrepresented area is the general AI/rationality/EA sphere, for obvious reasons.

One notable absence is the lack of books on sex/dating/marriage. He reviewed the book “seven rules for making a marriage work” by john Gottman, but nothing else as far as I can tell, although I’m sure other books might have touched on the subject. I highlight this because Scott highlighted the plight of romanceless lonely men as a serious issue, but is seemingly uninterested in presented these men with actual practical advice.

The major absence I see is when it comes to social justice. I see almost zero books specifically about social justice topics from a pro-social justice perspective. No books advocating against racism, sexism or homophobia. But he did review a book by former white nationalist and still pretty racist author Richard Hanania, arguing against social justice. I will discuss this a lot more later in the article.

Political views:

Okay, what is the skew of political views of the authors? Not all of them could be categorized, and the ones that could had to be rounded off. Take a look at the methodology questions.

This is another case of high subjectivity, so see above for how I categorized this. But it seems in general there’s about a 50:50 split between left-ish and right-ish sources. I want to point out that academia skews quite left wing and make up a decent chunk of the uncategorised people, so I expect if you interviewed every author, the “liberal” category would be bigger.

This definitely supports the idea that Scott is reviewing people with a wide variety of political leanings. This is especially true of economics, where Scott reviews liberatarians, marxists, mainstream progressives and conservatives. But of course, this is still merely a slice of the opinions out there. Anarchists, for example, have to be content with two books by James c Scott that are not focused on the economic system of anarchism.

I do want to point out that the left-right variety in authors is primarily about their economic beliefs. When it comes to social justice, for example, there is no balance: there are several books that are mostly about how social justice is bad, and almost no books about how social justice is good. To the extent that social justice is advocated for, it is as a subset of books talking about left wing economics.

Gender and racial diversity:

The gender diversity of subject authors can be seen in this section. I will just note the facts in this section, I will explore more about these figures (including possible explanations) later on.

The authors of Scott’s books are overwhelmingly men, with 92% of authors being men. The prevalence of women in society is approximately 50%.

Let’s turn to race. I note some of the difficulties with categorising this in the methods section, so you can explore some of that there.

In the united states as a whole, only 57.5% of people are classified as white american.

There were few enough non-white people that I can just list them out here:

shou ching jaminet is Asian. nassim nicholas taleb is lebanese-greek. Lenora chu is chinese (although he technically reviewed somebody elses review of her book?). Soner capagtay is Turkish. Wang Hunin is a chinese CCP official. Ananyo bhattacharya and Venkatesh Rao are indian (or maybe pakistani?).

Black people make up 12% of the american population (and a higher percentage of the world at large). I checked three times, and out of 99 authors in the publically available book reviews, I was unable to find a single black person. Similarly, I couldn’t find any hispanic people (18% of the population), although some hispanics are white-passing.

The only bright side for diversity is that a decent proportion of the white-passing people are Jewish. I am glad to see “eichmann in jerusalem” on the list, for example.

We can combine the two analyses above to just ask how many of the authors are white men:

White men take up 86% of the authors reviewed. For comparison, In the US, only about 28% of the population are white men.

If you are already forming objections to these diversity statistics, I would ask you to read the next few sections first, where I address some of the anticipated counterpoints.

My takeaway:

Reading through Scott’s reviews, you get the feeling of being immersed in a vast marketplace of ideas. One day you’ll be looking over a leftist historical treatise on the ripple effects of the states need for legibility, then the next a book on nutrition and health, and the next a semi-obscure book by a chess GM talking about how he raised his children to be geniuses, all from different decades and differing popularity. You’ll learn about the economic views of socialists, libertarians, communists, and anti-communists. This is a great feeling, and I’ve been exposed to a ton of interesting ideas from all over the place from these reviews. I like that obscure ideas are being brought up, and that you never know what you’re getting next.

But while the topic space explored is large, the actual world, is far vaster. There are just too many views to cover: there are no anarcho-syndicalists, Buddhist nationalists, or flat earthers. While the book reviews are long, you will not get the proper case for each position from a single blog post on the subject. And of course, this is all filtered through scott’s opinions and choices: For example I have seen anarchists be very annoyed at how the anarchists James Scotts “seeing like a state” is presented.

With these reviews, you get the feeling of exploreing a vast free marketplace of ideas, but in fact there is a wall around it, and only the ideas Scott wants to explore get let in, filtered through his own knowledge and his own blind spots. This is not a Scott specific observation: anybody who tried to do the same thing would run into the same problem.

But that does not let him off the hook for the choices he does make about whose voice he puts forward as worth hearing, and what ideas he puts forward as worth exploring, and which people and ideas he chooses to leave out.

Who he puts forward is mostly white men, including vile ex-white nationalists like Richard Hanania and anti-woke crusaders like Jordan Peterson. Who he leaves out is Black people and anybody trying to argue in favor of social justice.

The phrase “listen to minorities” is sometimes taken to mean “you must agree with the last minority that talked to you”. This is obviously insane, because minorities are human beings. There are black liberals, black conservatives, black socialists and black fascists. Male and female PoC will have vastly different experiences, and so will those who are Queer, disabled, poor, etc.

Similarly, under the vast umbrella of “feminism”, you have Rad-fems, liberal feminists, socialist feminist, intersectional feminism, Swerfs and Terfs (boo), first, second, and third “waves” of feminism. And these are just some of the labels.  Many of these are at war with each other. Some of them are very insightful or interesting (yay bell hooks!) Some of them are garbage assholes (boo terfs again).

If you read Scott alexander, you would barely know any of these groups actually exist. When he describes himself as “trekking to Bangladesh and slaying demons” to investigate feminist texts, he’s just lying. I looked through his multiple anti-feminist screeds, and the only feminists he ever seems to talk about are BuzzFeed-style clickbait journalists and internet bloggers. Even when he occasionally argues in defense of social justice, like his admirable defense of Trans people in “the categories were made for man”, his primary citation is to the non-trans Eliezer Yudkowsky, with no indication given that there is a giant literature where trans people discuss these exact topics out there.

What I take “listen to minorities” to mean is to not do this. In this world, non-white people have, on the whole, been shit on for centuries, colonized, exploited, enslaved, killed, been subject to pseudoscientific forces claiming they are dumb and inferior, discriminated against, . Women have been unable to own property, been forced into submission, legally and illegally raped, harasser, assaulted at a disproportionate rate. A lot of this is still going on in large parts of the globe. In light of all that history, the very least you can do is to let them have a voice about their own affairs. To understand why such a large percentage of them think the problems are ongoing. To understand the ongoing legacy of these past events.

Scott alexander once defended his engagement with neoreactionaries by saying he was looking for “nuggets of gold” in trash (although the “nuggets” he mentions are generally pretty shit). I don’t think it’s fair that dumbass white racists have their trash sifted through for “gold”, while black people get pretty much entirely ignored. If you don’t think there are bold, controversial opinions within social justice communities, you are not familiar with them. If you don’t think there are Black geniuses who are worth listening to, you are delusional and probably racist. If you don’t think, among the 30-60% of women in America who identify as feminists, there aren’t any who are worth hearing out on the topic, you are not a brave “decoupler”: you are a intellectual coward.

Scott Alexander once said part of his community was trying to host “the national conversation”. If Scott book reviews are his version of the national conversation, it’s one where most of the population isn’t in the room. And furthermore, the people who are in the room are bitching about the people outside the room behind their back, while slapping each other on the shoulder about how smart and rational they are.

Pre-debunking the anti-diversity stances

Now, for daring to bring up the topic of racial and gender diversity I’m sure a certain segment is going to write me off as a rabid woke SJW. Indeed, whenever the lack of diversity in rationalist spaces is pointed out, people are quick to point out potentially harmless explanations. Scott himself excused the lack of diversity in rationality as a matter of different interests, like women doing more yoga.

But this is a different situation. There are no complex socioeconomic factors here that we have to entangle. The only variable at play is “what does micro-celebrity blogger Scott Alexander deliberately choose to spend his time and effort to review”.

I already anticipate some common defenses, which I will try and argue against here to save everyone some time:

Maybe less diverse authors are just more popular?

Doesn’t matter. As we established, Scott reviews books over a wide range of popularity, and has reviewed books with less than 10 ratings on Goodreads. Obscurity is clearly no excuse here.

Are less diverse authors better? This is a bit offensive to say, but regardless, it doesn’t matter. As we established Scott reviews books over a wide range of quality, and has reviewed books with atrocious ratings on Goodreads.

Nonfiction books overwhelmingly non-diverse.

Yes they are, but not that non-diverse. And regardless, Scott isn’t taking a random sample here. There are way more books on rationalism here than in the NYT nonfiction bestseller list, which is generally more diverse than Scott’s list.

Older books are less diverse, so that’s skewing the numbers.

Most of the books Scott reviews were written in the last 20 years, so that won’t really affect much.

Maybe Scott’s niche interests just happen to be in non-diverse spaces?

A little bit, but not really. Take sociology for example, which is one of Scotts favourite subjects, with like 15 books reviewed loosely in that area. 75% of sociology students and half of sociology professors are women. And yet almost every single one of the sociology books that Scott reviewed was written by a white man.

Maybe Scott isn’t interested in social justice?

This is simply false. Scott just reviewed a Richard Hanania book arguing against social justice, and has published posts against social justice on numerous occasions. He is clearly very interested in attacking social justice.

This is identity politics, the author of the books shouldn’t matter!

Well, I guess there are probably good books written by pro-social justice white people about the black experience. But Scott didn’t review those, either. Do you think Richard Hanania was the only person qualified to discuss the matter?

Sexism and racism aren’t actually major problems anymore.

Yes they are, but even if you disagree, they were problems in the 1960’s, and Scott has had no problem jumping back that far to review, say, Malcom Muggeridge, a fringe reactionary. Do civil rights leaders not deserve the same treatment?

Everyone already knows the pro-social justice side, Scott is just providing balance.

This is simply untrue. People know the pop culture versions of social justice, and a few random slogans here and there. Everyone loves Martin Luther king, but how many people know his actual beliefs outside of “racism is bad” Newsflash: Scott’s overwhelmingly white male audience probably does not have an accurate picture of social justice and the experience of women and minorities!

You’re just trying to cancel Scott!

I have like 50 subscribers, Scott has his own subreddit with 60 thousand members. I have no power to “cancel” Scott Alexander.

“woke” books that Scott could have reviewed:

The easiest debunking here is simply to point out the many books that Scott could have reviewed, had he chosen to do so.

Note that I’m not saying that any of the books below are totally correct, free of unproblematic elements, or even any good. After all, there are plenty of dodgy and problematic books on Scott’s list. I’m saying that they would fit in fairly comfortably with the books he’s already reviewed, and would fill in the blind spots in his coverage.

Invisible women: exposing data bias in a world designed for men by Caroline Criado Perez:

This book is an overview of the harms caused by gender bias in various aspects of society, surveying various academic literature on the subject.

Delusions of gender by Cordelia fine

This book is a neuroscientists attempt to debunk the idea of hardwired brain differences between men and women. Reception by academia has generally been positive, although the author has been accused of leaving out some studies to favor her argument.

feminist theory from margin to center by bell hooks

If you want to know what intersectional feminists actually believe, then bell hooks is a great point to start. hooks is black feminist writer who often critiqued elements of mainstream feminism at the time, such as anti-male sentiments and the focus on upper-middle class white women.

Women, race and class by Angela Davis

This book attempts to analyze the role of women and racial discrimination in US history from a Marxist perspective.

The new Jim Crow by Michelle alexander

This book presents a comprehensive account of the extreme and brutal levels of imprisonment in America, which disproportionately affects black Americans, as well as poor white Americans.

The wider problem

If you were to say that effective altruism has a groupthink problem, the typical EA’er might respond with a scoff. “How could we have groupthink? we can’t figure out where to go to lunch without a 10 page debate on the counterfactual ethical importance of sushi vs indian food. What a crazy accusation!”

The mistake they have made here is that groupthink is not just a function of the amount of arguing you do, it’s also a function of the range of things you argue about, and who you argue with.

Someone who thinks “the singularity will kill us all”, and someone who thinks “the singularity will usher us into a utopia” can exhaust each other arguing 24/7. They will both walk away thinking they’ve explored so many interesting arguments and arrived at their final opinion from good faith, informed argumentation of vast swathes of opinon. But they have still only explored a tiny part of the idea space: they have never considered the arguments that the singularity is not imminent, and may never happen at all.

Intentionally or not, Scott has set up an intellectual space where the opinions of black people and feminists are disproportionately left out of the picture. And I don’t think it’s just his blog that does it, although he might be one of the worst offenders. You can compare to this list of notable EA people prepared by an EA forum administrator: Sam Harris, Jeremy Bentham and Carl Sagan get cited as external heroes: Martin Luther king Jr does not. I would be unsurprised if more rationalists had heard of Curtis Yarvin than bell hooks.

I’ve always been bothered by the diversity problems in EA and rationalism, but I’ve never thought it was a pressing issue for society until recently. Some EA and rationalists are trying to set governmental policies about new technologies.  Some people influenced by EA, and by Scott Alexander and the rationalists, are straight up trying to build a machine god that is “aligned” with human values. Whose values we are aligning with? In the worst case scenario (which is thankfully very unlikely), the flawed values of a small group of AI creators could end up locked in until the end of the universe. Even in the more mundane situation that AI just makes a few people very rich and influential, their values still have a large effect on the world at large

It matters who is listened to, and whose ideas are taken seriously. A world where we listen to Richard Hanania, but not to black people, is a dangerous one indeed.