On the Rationalist Community and its Ugly Shoes

 

Here's an article that has been a long time coming.  And there's so much to say on this topic that I don't really expect to cover it all, even in a cursory way, in a single essay.  So perhaps this is the first of many.  

I reference the "rationalist community," either directly or indirectly, in some of the earliest blogs on this site, and I've been promising to write an article about the "rationalist community" for a long time - for instance here and here - two articles on two completely different subjects - Ayn Rand's "Objectivism" and geopolitical "Realism," respectively - in both of which I make fun of movements that give themselves names like "Accuratism" and "Truthism" and "Goodism," with the implication that to oppose them is to be against goodness and truth and rationality. 

There's another problem with referring to themselves by the term "rationalists" - or "aspiring rationalists," as they like to call themselves - namely, that there was already a quite different, huge, incredibly important philosophical movement by that name.  If you mention "rationalism" to just about any professional philosopher in America, and quite possibly the world, the historic philosophers that will immediately come to mind for them will be people like Rene Descartes, Baruch Spinoza, Gottfried Leibniz, Nicolas Malebranche, Antoine Arnauld, and many others.  One could indeed trace "rationalism" back further, to figures like St. Thomas Aquinas, St. Augustine, Aristotle, Plato, Pythagoras, and perhaps even earlier thinkers.  These are, of course, some of the most revered names in the history of western philosophy.  "Rationalism" in this sense refers to seeking pure reason, deductive reason, as the source of truth, "a priori" truth, as Kant would put it, as opposed to "empiricism," which seeks to ground truth in experience - inductive reasoning, "a posteriori" judgements to use Kant's terminology.  But none of that has anything to do with the so-called "rationalist community" (or does it? Put a pin in that).

So what is the "rationalist community"?  Of course you can google it, and I encourage you to.  It's definitely a rabbit hole, and I don't want to rehash everything here that you can just as easily find in a thousand other sources.

For now, I'll just say: it's a blog.  Or perhaps it's a few blogs, and the fans of those blogs.  The main blog in question is "LessWrong," which is mostly written by blogger Eliezer Yudkowsky.  There's also "Slate Star Codex," now renamed "Astral Codex Ten," written by "Scott Alexander," (the blog titles are both anagrams of his name) but Yudkowsky's stuff is more relevant to this article so I'm going to focus on him - though I will mention that Yudkowsky began as a frequent contributor to another blog called Overcoming Bias in 2006, and then started LessWrong in 2009.  Both were nominally collective blogs, but Overcoming Bias was dominated by Robin Hanson and LessWrong by Eliezer Yudkowsky.  Several blog posts were put into an order, which became known as "The Sequences," and then was published (with some editing) in book form, as "Rationality from AI to Zombies" in 2018.

Unlike traditional philosophical rationalists, like Rene Descartes, members of the "rationalist community" tend to see themselves as rigorously empirical (i.e., the opposite of "rationalists" in the traditional sense) and scientific, and the P.R. of the "rationalist community" is that it appeals to people in STEM fields.  But does it really? I'd like to see some solid statistics on exactly what proportion of "aspiring rationalists" are scientists in the sense of having a job doing scientific research or having advanced degrees in the hard sciences - and, conversely, what proportion of scientists are "aspiring rationalists".  Yudkowsky himself famously did not graduate from high school or attend college, and I'm not aware of many "aspiring rationalists" who are physicists, say, or chemists, or even mathematicians.  (Nothing wrong with that, of course - I hate credentialism as much as the next guy, and it took me pretty long to get any kind of college degree.)  It is widely reported (though again, I'd like to see some statistics) that the rationalist community is popular in Silicon Valley, though that hardly means that most of its adherents are computer scientists, or even software engineers.  If anything, it simply means that it is well-known in the San Fransisco Bay Area.  It does seem to appeal to some podcasters and youtubers, people who are terminally online, and a few very prominent venture capitalists.  But none of those words is synonymous with "scientists".  (Perhaps its most famous representative these days is Aella, who is most well-known for being a sex worker.)

Let's quickly get over some of the stuff most people say about it.  

They'll say the "rationalists" are nerds, which: guilty.  100% guilty.  1000% guilty.

They'll say "rationalists" are a cult, which... uh, maybe?  A little bit?  Personally, I think the cult-like behavior is a bit exaggerated, though I can't deny that it is a little culty.  Cult-lite.  Then again, I love cults. 

They'll say that "rationalists" are polyamorous and that the history of the rationalist community is the history of infighting within polycules.  I don't really know anything about that, but it sure seems like some of the "rationalists" are polyamorous - likely more than the national average. 

They'll say that "rationalists" wear ugly shoes.  No comment.

Of course, none of that is an essential characteristic of the "rationalist community". 

So what's it all about?  Well, that's a long story.  In the early years, the site was centrally focused on Bayesian statistics.  More on that in a moment.  Although that was the focus, sometimes it strayed into other topics, such as cognitive biases and how to overcome them.  This is a fun, interesting topic, well worth learning about.  Everybody should take some time to learn a little about this.  Sometimes it took on almost a self-help kind of tone, which I don't love, but it's fine.

But as time went on, the subject of the blog tended to wander - as blogs often do - and sometimes it got to the strangest places: Yudkowsky's extremely quirky ethical beliefs, his speculation about interpretation of quantum mechanics, his long-held extropian beliefs, such as his commitment to life extension and especially cryogenics, thoughts about philosophical zombies, and especially, eventually, speculation about the dangers of artificial superintelligence (ASI).

I have to admit: I agree with Yudkowsky about a lot of stuff.  I, too, love science.  I'm interested in the topic of cognitive biases and how to overcome them, and I think that Yudkowsky often has genuinely useful advice.  Also, as a layman, I think Everett's "Many Worlds" is the interpretation of quantum physics that makes the most sense to me.  I'm also really concerned that we will develop ASI without first ensuring that it is aligned to human values, and humans will no longer be the most intelligent or powerful beings on Earth - and I'll confess that it was largely through reading LessWrong that I came to that opinion (and then I read Nick Bostrom's book).  

Sometimes I see people on the internet complain about Yudkowsky's writing style.  I disagree.  I actually really enjoy his writing style.  I think he's charming, witty, and genuinely funny.  People say he's longwinded.  (Nate Suares has said, about the 300 page book that he collaborated with Yudkowsky in writing, that Yudkowsky wrote 500 pages, and Suares wrote negative 200 pages.)  I actually don't find him longwinded at all.  I think he has a remarkable knack for explaining complex ideas quickly and efficiently, with clever, illustrative story-telling.

In short: I like Yudkowsky!  And I find his arguments, though a little out-there, quite compelling and even convincing on a wide range of topics.

But none of you are here to read that, are you?  So let's get to my disagreements with Yudkowsky and the rationalist community.

First let me start by giving some background.  I was exposed to Bayesian statistics back in the 90s, when I took a class on the subject from Professor Clark Glymour of Carnegie Mellon University.  I'm tempted to call Professor Glymour an "early adopter" of Bayesian statistics - which is a bit of a silly thing to say, since Thomas Bayes came up with his famous theorem back in 1736.  But throughout most of the 20th century, Bayesian statistics were sidelined in favor of more traditional "frequentist" theories of statistics.  Bayesianism didn't really become all that popular outside of obscure academic circles until, well, until the 2000s, partially through the proselytizing of websites like LessWrong.  But yes, back in the mid-90s, the charismatic Professor Glymour applied the full force of his infectious enthusiasm not only for Bayes, but also for combinatorics, which he traced all the way back to Ramon Llull in the 13th century.  I'll never forget his tales of Llull.  (Actually there's more to this story, and it's relevant - but I'll skip over the nuances now, for the sake of the narrative.)

And let me tell you: I ate that shit right up.  I was thoroughly convinced that nothing could be considered genuinely scientific unless it rigorously applied Bayesian statistics.  The truth is, I didn't really understand Bayesian statistics all that well at the time, yet I understood it (or misunderstood it) well enough that it electrified me - it seemed to light up something deep and powerful and solid within me and gave me a meaningful answer.  I had already been a person with a fairly strong skeptical personality - I was not religious, I didn't believe in astrology or new age medicine or conspiracy theories, etc..  (Actually I was extremely skeptical in high school, but that's a story for another time.)  Now I had a peg to hang my skepticism on.

But then my life went on, I took other classes, I developed other interests, I became a musician, in short I lived my life.  And Bayesian statistics gradually faded in importance, and sank to the back of my mind.  So when I heard about Bayesianism again, many years later, through LessWrong, it was like becoming reacquainted with an old friend.  I chuckled to myself, as if to say, "Oh yeah, Bayes... what has he been up to all this time?"

I'm not trying to be a hipster and say that I was into Bayesian statistics before it was cool, but... well, yeah, I guess I am saying exactly that. 

And when I returned to Bayes, it was no longer with the zeal of a convert.  Instead, I looked at Bayesianism more dispassionately, more critically, more accurately, and perhaps ironically, more skeptically.  I've continued to think about Bayesianism, including on this blog.  For instance, you can see some thoughts I've had about Bayes here. 

I wrote the above thoughts shortly after reading William Poundstone's "The Doomsday Calculation" - a great book that I highly recommend for anyone interested in these kinds of topics.  (Many of the ideas in the book I had read about before, but some of them were new to me.)  You can enjoy it on two levels - first, straightforwardly, as an interesting, fun book with interesting ideas that are worth considering.  But secondly, as a kind of cautionary tale, not to take any of these ideas too seriously.  Seeing all the bizarre notions that people eventually come to when thinking about these topics too hard can be a great vaccination against getting too carried away with this nonsense.  I would say that reading that book was my final kiss-off to the whole Bayesian genre.

The kinds of things they talk about in that book are fun mental exercises, to be sure, but that's all they are.  There isn't a shred of evidence for anything that these people are asserting.  And that brings me back to the philosophical rationalists in the original sense of people like Descartes, Spinoza, Leibniz et al..  Perhaps the contemporary so-called "rationalists" aren't so different from the classical rationalists, after all.  They both assert things - wild things, bizarre things - without going out and doing any actual experimental research to find out if any of their assertions are true.  They simply have a logical argument, or a mathematical-logical argument, or the barest sketch of a mathematical argument, without any of the actual math - with either completely made-up numbers or no numbers at all - and to them, this argument shows why their assertions must be true, making actual, hard-won experimental evidence completely unnecessary.  In fact, at this point I have come to have more respect for the classical rationalists like Descartes and Spinoza than I do for the contemporary "rationalist community."

In fact, long before I read "The Doomsday Calculation," the "Rationalist community" had become a joke for me.