Discover more from Razib Khan's Unsupervised Learning
RKUL: Time Well Spent 07/07/2022
The Survey Edition
Your time is finite. Your phone and the internet stand ready to help you squander it. Here are my latest picks for spending it well instead. Feel free to add more in the comments.
The first Unsupervised Learning Survey
In the 20 years I’ve been writing on the internet I’ve been doing polls and surveys every few years. I have a general quantitative sense of my readership already, but I wanted to see what the breakdowns were for Substack specifically. I asked for some demographic information, posed some questions relating to this website (how did you find me?), and solicited thoughts on future post topics.
More than 850 people contributed to the survey, a mix of free and paid subscribers. I sent the two tiers to distinct survey interfaces, so I have the information disaggregated, but in some cases, I wanted to merge them. For example, above is some basic demographic information relating to generations for all respondents. You can see the generations are pretty balanced overall, though the real “greatest” generation, we Xennials, are overrepresented.
No big reveals here, at least for me. For the question regarding religion, I requested even atheists put down a religion if they identified with one. If my earlier surveys are the guide I think they are, that would mean most of the 10% of readers who are Jewish are not very religious. Unfortunately, I haven’t diversified the regional distribution of my readers over the last 20 years, it remains a mostly European and North American audience. Long, long ago I thought I’d get a lot of Chinese readers, but they have their own internet behind the great firewall. I also had both African and Indian readers inform me that Substack (and so I guess Stripe?) is basically impossible in terms of payment options for them. I hope they’ll sort this out for everyone but in the meantime if this is you, feel free to email me directly. We’ll sort something out. This was one of the main reasons I resisted ever having a subscription model on my long-time blog; it seemed too likely to throw up barriers for international readers.
In relation to other basic data, the modal reader of my content is a secular white American male. These numbers are very familiar, as my male readerships have consistently floated between 80 and 90%, and 80% of my respondents are always non-Hispanic whites, with South Asians almost always number two.
In terms of politics, we’re all over the spectrum in significant percentages, but readers lean more right. I asked readers about how socially and economically liberal (Left) and conservative (Right) they were on a 1 to 5 scale, with 1 being more socially liberal and economically conservative. In this framework, an extreme libertarian would give 1 and 1 as responses, and an extreme fascist would give 5 and 5. Paid subscribers were 2.89 social liberal (so just to the liberal side) and 2.31 pro-free market (more strongly economically conservative). Free subscribers were 2.89 on social liberalism, the exact same average as paid. But for economics, they were 2.48, so a little less pro-free market. The political variation among readers is high, but there is a right-libertarian tinge, which perhaps reflects many of my own biases.
When it comes to education, nearly half have some graduate degree, with nearly 20% holding Ph.D.s. This compares to about 14% of Americans 25 and older having some graduate degree, but I know for a fact that a nontrivial number of my readers are graduate students or university students who will join that 20%, so this might not even be a fair comparison. About 1% of respondents have a high school diploma or less (the value for those age 25 and older in the US is 37%).
When it comes to socioeconomic status, whether free or paid, about half of the readers judge themselves as upper-middle class, and more consider themselves upper class than the lower class. In short, respondents are wealthy and educated compared to the average person.
How did readers find this Substack? 70% of paid subscribers and 45% of free subscribers found it through my blog or my Twitter (though I am always grateful for genuine endorsements from unusual suspects that introduce new audiences to me).
What about why subscribers upgraded to paid? 135 paid subscribers answered this question, and “steppe” came up 17 times and “Ashkenazi” 8 times. But the most common response was basically “no particular post” or “all the content.”
While unsurprisingly 64% of free subscribers did not read the archives, 31% of paid subscribers didn’t either. I’d love to tempt that minority of you who are paid and found me after the first couple months; most of my in-depth content is paywalled and will remain so. And it’s easy to miss something. With this in mind, we do have a better guide to past series of themed deep-dives in the works. Here is what someone who just subscribed two days ago messaged me (this is a moderately public person):
Where has your substack been my entire life?
I actually used to have a free subscription, but I got it a bit later (late 2021?), and I had *no idea* what there was in the archives.
In general, I manage to get out between one and three substantial paid pieces (in addition to the handful of free ones) every month, so this Substack has been accumulating a backlog. The archives await you! If you want something different from the Substack-provided archives, that includes my other copious, less edited content, I’ve been working with a startup, Dry.io, on delivering everything more effectively to readers. They pulled thousands of my posts into one site (recent content is surfaced first) and even set up a nice search engine to crawl the posts. Their goals are to allow you to build your own search engine, social network, or collaborative software platform in hours, with no programming, so I’ve been brainstorming with the CEO for a few months on what they can do to help me. I can’t write faster, but I can arrange what I have more efficiently, and I know even long-time readers often underestimate how likely it is that with twenty years of content, I may have already addressed their latest question.
Next question, how long have you guys been reading me? There are of course people subscribing to this Substack who have been reading me literally since 2002 (Hi Martin!), back when I was the same age as today’s median Zoomer. But no surprise that free subscribers are far less familiar with me. About 39% have been reading me for only one or two years, while the equivalent value for paid subscribers is 21%.
I got a lot out of the responses on questions of what people do, or don’t want. What people most commonly do not want is more “Culture Wars” content, and that shouldn’t be a problem. The national genetic histories (Hungarians as the ghost of the Magyar confederacy), general history (Made in China), and population genetics (Under the skin) are quite popular, and the top three categories selected by paid and free subscribers. For the free-form responses, a lot of respondents took time to contribute to a running wishlist of topics I work from. But I was gratified how many readers were happy enough with what I’ve been exploring to basically just say, “you do you.”
There is a lot more survey we could go over (you can find some memorable direct quotes at the end of this newsletter), but these are the primary things that jumped out at me. If you want to see the topline results, I’ve posted gdocs for free and paid subscribers online.
On a fallacious argument about ethnicity and genetics. Wyclif’s Dust has a title that always makes me think it’s going to be written by a 90-year-old Oxford don with a fascination for Reformation thought, but the author is actually David Hugh-Jones, coauthor of works like Genetic correlates of social stratification in Britain, and his opinions on ‘social genomics’ are always worth checking out.
A Guide Map For Reading the East Asian Canon. Tanner Greer reads and it makes all the difference to his content. I’ve read about a third of his list, and can recommend that subset for sure, but may check the others out (I’ve read F. W. Mote’s Imperial China twice and consider it indispensable if you are interested in Chinese history).
Red States Are Winning the Post-Pandemic Economy. Austin, where I live, along with Florida figure prominently in this piece. The real issue isn’t a massive Red-Blue divide, it’s that if you are a professional, Texas and Florida make it much easier to have children and buy a house than California and New York. Just look at the map.
The Circulation of the Elites: The counterrevolutionary ethic of the PMC. Leighton Woodhouse is trained as a social scientist, and that shows in this piece. It’s a must-read; he diagnoses how the professional-managerial class (PMC) is co-opting class rhetoric to entrench its status, and how this was anticipated by Christopher Lasch in the 1970’s.
What Caused The 2020 Homicide Spike? I agree with Scott and have long agreed with him, it’s de-policing. That being said, I don’t see that anyone is going to do much about this for a long time, leaving us to just hope everyone we love has the economic means to not live in a “sketchy neighborhood,” as they used to say in the 1990’s. Crime annoys the PMC, but most of us don’t know many people who were murdered and are not at much risk ourselves.
More Regulation Please! A few of my friends joke Nick Patterson can only write in LaTeX, so I’ll quote from the PDF: “I was going to post a piece about why I now believe that COVID-19 originated as a leak from the Wuhan Institute of Virology (WIV) but the case has been laid out with great clarity by Matt Ridley.” We may be seeing the beginning of a preference cascade in the next few months.
Genetic substructure and complex demographic history of South African Bantu speakers. This paper digs deep into the nature of the Khoisan assimilation into Bantu populations, as well as the earlier Nilo-Saharan assimilation into the Khoisan (and the emergence of the Khoekhoe).
Genomic evidence for ancient migration routes along South America’s Atlantic coast. “To further add to the existing complexity, we also detect greater Denisovan than Neanderthal ancestry in ancient Uruguay and Panama individuals. Moreover, we find a strong Australasian signal in an ancient genome from Panama.” The first part is a really weird result. The second part is probably robust, but no one gets what’s going on with the Australasian signal.
Evolutionary models predict potential mechanisms of escape from mutational meltdown. As they say, timely. “The conditions and potential mechanisms of viral escape from the effects of mutagenic drugs have not been conceptually explored.” We should all be catching up on viral evolution…
Complex Traits and Candidate Genes: Estimation of Genetic Variance Components Across Modes of Inheritance. Complex traits are important because a lot of disease genetics today (type 2 diabetes rates) can be thought of as complex traits. Work in this area is essential to improvements in well-being on the margin.
Widespread epistasis among beneficial genetic variants revealed by high-throughput genome editing. There are theoretical reasons people are skeptical of the importance of gene-gene interactions in the long-term evolutionary processes, but they are clearly important in proximate trait architecture, for example, in the context of diseases.
My Two Cents
There’s still no free lunch, free subscribers; my most in-depth pieces for this Substack remain beyond the paywall (although this month, my big two-part series on the history and utility of haplogroups was free for all). Since the last “Time Well Spent” I’ve posted one in-depth paid piece, but it looks like it will become one of my most-read regional genetic histories, like Ashkenazi Jewish genetics: a match made in the Mediterranean and Red hair is about as recessive as St. Patrick was Irish.
Though proudly Arab and steadfast in their adherence to Islam, modern Egyptians are still conscious of that hallowed history and deservedly proud of their nation’s past and uniqueness. And in the end, their pride and patriotism are not for an adopted history, like mine as an immigrant devoted to America; they are literally born to this storied lineage, carrying it forward unbroken for millennia. As if Egypt’s legacy to humanity wasn’t potent enough to begin with, genetics now confirms that the ancient Egyptians’ children truly walk among us today, from Aswan at the first cataracts of the Nile in the south to ancient Alexandria at the Mediterranean’s edge.
Here are my guests since the last Time Well Spent:
And here are the currently ungated podcasts all in one place.
Over to you
Comments are open to all for this post, so if you have more reading/listening suggestions or tips on who I should be talking to, what I really should be writing for you, or just missed your chance to sound off in the survey, lay it on me.
Many of you have long since noticed that Substack remains down-home enough that you can simply reply to my mailings and it comes directly to me. I love those informal little notes people fire off upon finishing one of my deep-dive posts. Please feel free to keep them coming. The free-form sections of this survey were fun though because people I don’t think I’d usually hear from sounded off, including some who might not have said what they did with their reply email attached. For your entertainment, here are a few memorable notes and recurrent themes:
Follow your instincts. What I like is that you seem unfazed when your priors lead you to the unpopular side of controversies. I suppose that irascibility goes hand in hand with being opposed to following the diktats of cancel culture.
Love your great work, your fearlessness and your general friendliness Razib.
Keep up the good work! I appreciate your fearlessness to discuss topics that send most people cowering in terror.
Multiple free subscribers informed me that my value to them was more on the order of $1, $2 or the $5 per month I originally charged. A couple of notes. I’ve heard Chris Best explain that Substack had to force the minimum floor of $5 per month because writers are so prone to undervaluing their output. They had writers wanting to charge 25 cents per month.
One respondent said Eight dollars a month is a lot for me - I pay every couple months and binge the archive. I want to explicitly endorse this approach. Many subscribers do this and this is absolutely how I would have done it at other points in my life. As for the person complaining that it is no longer $5 per month like it was when they first subscribed, the cool thing in Substack’s current model is that for the early adopters who haven’t bailed, it actually still is $5/month now and forever.
Some of the input was just enjoyable for the specificity of the detail included.
Come to Australia!
Being 84, I didn't know how to get your podcasts going, but thanks to a tip from Razib I'm chewing them all up starting from the beginning.
Love your blog and podcast but your Twitter is annoying
Petty technical detail: I need to be able to download the MP3 file before I can listen to your podcasts (due to the obsolete technology in my old Dodge pickup), so I patiently wait for the ungated versions.
Of course many had a bone to pick with me, some exceedingly personal, others general enough to apply to multiple cheerfully abrasive figures at once:
No [suggestions]. but you unfollowed me, and I will always remember that :)
during interviews you talk too much; I want to hear your guests. BUT, yer a good writer.
Razib (and Richard Hanania, who is worse) could practise completing a sentence without jumping to the next thought when speaking.
Naturally, there was lots of supportive input, some of it even earned by me personally.
l love your kid's voice at the start of the podcast.
I LOVE the [podcast] intro with the music and your kids talking. So cute. Look forward to the intro. Makes me want to start up a new podcast when I finish one.
The interview with Alina Chan from early 2021 was excellent and proved you (and she) did not cower in fear.
You're doing a great job. I'm a retired professor. The quality of your work is light years beyond the nonsense I had to deal with in my career.
its like reading old time newspapers with better depth.
I very much love and appreciate how you maintain a scientific view on genetics and history (as much as there can be). You really have no idea how much pleasure I've received from your posts and podcasts over the years. Thank you so much.
One respondent entertained me with the idea of a superabundance of Razibs. I’d started following your Twitter at some point for the smart takes on culture war crap, and it took way too long for me to realize “Holy crap, he’s THAT Razib Khan!” (.....from your genetics writing, which I had enjoyed for years.)
My family’s favorite was the slightly gnomic Razib is officially a bad person but tries hard to tell the truth, We need more such. My eldest narrowed her eyes at this and informed us, It’s kind of true.
And finally, one of the first respondents simply closed by informing me, I would like shorter surveys. Amen.