In Defense Of The Casual Polymath

It's Possible To Figure Stuff Out

I recently read an article that made me feel bad. Here’s me defending myself.

The article is called Beware the Casual Polymath written by the hilariously named Applied Divinity Studies (ADS). ADS argues that the internet creates incentives for people to be polymaths—writing about a whole host of topics one doesn’t know much about. People may be a bit interested in anthropics, which is something I happen to know something about, but as cruel, tragic experience has taught me, when I stick to what I know about, and write 5 billion back-to-back articles about anthropics, no one cares. As a result, writers on the internet are incentivized to spread their articles across a huge range of topics. Because one can’t be an expert on everything, people write about things they don’t know about, and there’s a race to the bottom for who can write the punchiest article that sounds good to a generally uninformed public. Experts are replaced with uninformed generalists.

This is mostly right. The top substacks that talk primarily about ideas rather than, say, fashion, mostly either just cover the news—generally through a specific ideological slant that appeals to a particular kind of person—e.g.. disillusioned liberals or illusioned liberals—or come from smart polymaths who write about a lot of things. There are very few substacks that come from, for example, people who are experts on microbiology, and write exclusively about microbiology.

Substacks are, in that respect, quite different from books. While many people might read a book providing a detailed account of the history of the civil war, microbiology, or anthropic reasoning, few will read a substack exclusively about it. Books have a more organized structure and concretely distill information—people aren’t as interested in a weekly barage of random new facts about niche topics as they are in sitting down, reading a book on a niche topic, and then only revisiting it if they want to read another book about it.

But it’s hard to be a genuine polymath. It’s hard to know about immigration, the lab-leak hypothesis, criminal justice reform, eating-meat, gun control, and many other topics. As a result, we’re being informed by polymaths who know a bit about a lot of topcis but aren’t experts on anything.

Reading this made me feel slightly guitly. To quote Taylor Swift “It's me, hi. I'm the problem, it's me.” I am, in my writings, the sort of polymath that Mr. Divinity studies was railing against. I write about tons of topics that I have studied a bit, but not super extensively—immigration, Judaism, Christianity, HBD, criminal justice reform, educational polarization, the negative impact of social media on mental health, the Camp David proposal, insect farming, pharmaceutical price controls, and more. While there are some things I know quite a lot about—arguments for and against the existence of God, anthropics, normative ethics—I also write extensively about topics that I’m certainly not an expert on, ones that I’ve spent, say, 8 hours thinking about rather than 1,000.

I could stop doing this. In the early days of Bentham’s newsletter, I wrote almost exclusively in defense of utilitarianism—my first 10 articles were about why Huemer’s objections to utilitarianism were wrong, I wrote two random articles about other things, and then 18 back-to-back articles defending utilitarianism. I could go back to doing that, writing a 500 articles about anthropics and calling myself SIA’s Spaniel, or writing 500 articles about utilitarianism.

So, am I bad? Should I stop doing my frivolous dabbling? Should I just write niche articles about anthropics rather than trying to be a polymath? I don’t think so.

First of all, I think it’s not immoral to write articles that you’re not super confident in. I’m not an expert on race science, but I wrote about HBD, while being quite clear to make lots of caveats about not being an expert. Even non-experts can read through disputes between experts and come to conclusions.

The more important point, however, is that I think it’s generally not too difficult to figure things out if given enough time. I think the average smart person could, if they put in a few dozen hours reading through the literature on google scholar, figure out that pharmaceutical price controls are a bad idea. It doesn’t take long reading about factory farming to realize that factory farms are grotesque torture-chambers.

I recently wrote a post steelmanning Judaism—making a maximally persuasive case for Judaism, even though I am not religiously Jewish. It didn’t take that long to get a pretty decent sense of the main debates surrounding the case for Judaism—involving the Kuzari, the survival of the Jews, and so on. Overall, I probably did about 5 hours of reading on the subject.

This wasn’t completely comprehensive. With 5 hours of research, I wasn’t able to extensively investigate the various archeological claims made by both sides, nor was I able to do extensive research into alternative explanations for the survival of the Jewish people. But I was able to get a decent sense of the case and see what the critics generally said in response.

A lot of topics are like that. It isn’t hard to see that, at the very least, most of the objections to immigration are wrong, as most of them rest on straightforward, demonstrable errors. Some of them are more complicated, and those require more research—when addressing those sorts of topics, I try to do adequate research.

But I think most topics can be figured out decently well by a smart person willing to put in the time to analyze the issues. Some require a great deal of time—I wouldn’t write about, say, the impact of criminal justice reform on crime absent doing a ton of research—but many topics aren’t too difficult to figure out.

I think, for example, after reading both Jonathan Haidt and his critics on whether social media negatively impacts mental health, one can see that Haidt is more convincing. Haidt’s theory best explains the available evidence, and while it’s far from certain, it’s quite probable.

I think Scott is pretty accurrate in the topics he investigates. He has an advantage over me—knowing statistics—and as a result of being a smart, pretty unbiased guy willing to do his homework, I think he’s substantially more accurrate than even people who have written books that he disagrees with.

One should, of course, not be too certain of the conclusions I reach in my articles. While I did do a pretty solid amount of reading on the dispute between Haidt and his critics, I’m not an expert. But even experts are wrong a lot, as Phil Tetlock has shown.

The problem is, then, not with polymaths but with human error. Sure, an expert would be able to rattle off studies more convicningly, but it’s not clear, across many domains, whether the experts are actually better than well-informed amateurs. As Tetlock’s research has shown, the superforecasters—generalists who don’t specialize on any particular topic—do way better than subject matter experts in predicting events. In fact, it is the people who specialize too much—devising elaborate models and theories, rather than being more general—who do worse than so-called hedgehogs, who rely more on rough heuristics than elaborate, highly-detailed models.

When one specializes too much, it can be easy to put too much trust in one’s assumptions. Once one feels that there aren’t unknown unknowns, their views become less convincing.

For this reason, I plan to keep writing about things that I’m not an expert in. While I try to be informed, I think a reasonably intelligent amateur can get a pretty good grasp on most topics. The ones I feel I can’t get a handle on, however, are ones I don’t write about!