Main Topic: The podcast episode discusses various topics related to genetics, history, and evolution, with a focus on the intersection of these fields. The conversation covers topics such as the long-term equilibrium of intelligence, the genetic basis of endogamy, and the potential for gene editing and brain-computer interfaces.
Key Points:
Participants:
Takeaways:
Additional Insights:
Razib Khan is a writer, geneticist, and blogger with an interest in history, genetics, culture, and evolutionary psychology.
Watch on YouTube. Listen on Apple Podcasts, Spotify, or any other podcast platform.
Podcast website here.Follow Razib on Twitter. Follow me on Twitter for updates on future episodes
Thanks for reading The Lunar Society! Subscribe to find out about future episodes!
Time Stamps
(0:00:05) Razib's Background
(0:01:34) Dysgenics of Intelligence
(0:04:23) Endogamy and Genetic traits in India
(0:08:58) Similar Examples of Endogamy
(0:14:28) Why So Many Brahmin CEOs
(0:19:55) Razib the Globe Trotter, Geography Expert
(0:25:04) Male/Female Genetic Variance
(0:30:04) Agricultural Man and Our Tiny Brains
(0:34:40) The Church of Science
(0:42:33) Professorship, a family business
(0:44:23) Long History
(0:52:42) Future of Human-Computer Interfacing
(0:56:30) Near Future of Gene Editing
(0:59:19) Meta Questions and Closing
Please share if you enjoyed this episode! Helps out a ton!
Dwarkesh Patel
All right. Today, the pleasure of speaking with Razib Khan. He's one of the top science bloggers in the world. He writes about genetics, history, and evolution on his blog, Unsupervised Learnings. And he has a podcast of the same name. You can find it at razib.substack.com. So, Razib, thanks for coming on the podcast.
Razib Khan
It was my pleasure, man.
Dwarkesh Patel
So, can you give my audience a little bit of background about you, how you got into all this stuff?
Razib Khan
Yeah, I've always been interested in topics like history, demographics, etc. I've also been interested in science, have a scientific background and training. Over the last 20 years, genetics has become a really big deal as a tool to do various things, whether in the biomedical space or historical inference. Obviously, I'm interested in demographics and historical inference, and genetics is a tool I can use as a geneticist. As we're recording right now, I decided to do a bunch of pairwise genetic distances between populations just because I could, for a post. I do a lot of things by myself, why I replicate what's been done. That's a lot of what I do.
Dwarkesh Patel
Okay, interesting. I'd like to jump into it. My first question is, assuming there's no gene editing in the near future, what does the long-term equilibrium for intelligence look like? There are multiple visions. One view is like Charles Murray's *Coming Apart*, where you have fat tails because of assortative mating. Another is a slight dysgenic effect because there's lower fertility among higher intelligence people. What does the equilibrium look like if there's no gene editing?
Razib Khan
More like the second in terms of not an equilibrium yet. We're not going to have an equilibrium until the reproductive differentials equilibrate. They will at some point, but it could be centuries. At this point, people with genes for educational attainment tend to delay childbearing to the point where a lot of them do not have children because they invest in educational attainment in the short term. They don't have as many children, and their generation times are longer. The math isn't difficult there. Right now, there's a strong negative selection. Not strong, but there is negative selection on genes for educational attainment. Everyone who's looked at it says that, at least in the developed world.
Dwarkesh Patel
Is this something we can expect in the long term? Naively, I would expect that people who are more intelligent, as long as there's some sort of selection pressure in the long term, there should be selection for educated, smart people because they will have the cognitive tools to actually reproduce, survive, and thrive. As long as some smart people want to survive and thrive.
Razib Khan
Survive and thrive is one thing, but having offspring is a different thing. The incentives in our society are such that a lot of people believe that thriving is being child-free. What usually happens, I think, is people want to establish themselves in their 20s. At least professional managerial, college-educated people. Then in their 30s, they start thinking about it. Sometimes people wait too long, there are fertility issues, or they just can't find someone. In the long term, there's a limiting principle, but you don't need to be that bright to survive and have a lot of children. On the contrary, there's clear evidence that not being bright is good for your reproductive output. There's a movie about that in 2006 called *Idiocracy*.
Dwarkesh Patel
What explains the level of endogamy you see between Indian jatis, the subcastes? You had an excellent blog post about this. Apparently, as you say, there's genetic evidence that for thousands of years, these jatis living in the same village are not intermarrying or having kids together. Even within the context of slaves in America, this is not a thing that happens. You have Sally Hemings, Thomas Jefferson's mistress. How is it possible for thousands of people? What kind of social structure could lead to this?
Razib Khan
Nobody really knows is the short answer. The math is like this: there's evidence from Andhra Pradesh, South India. David Reich looked at it, and if you run the math, their endogamy rate is like 99.5 percent per generation, super high. When I was younger, the endogamy rate for Black Americans was like 95 percent, which is high, and today it's like 85 percent. But five percent is 10 times bigger than what I'm talking about. Average Black Americans have 20 percent European ancestry. There are really high barriers in the Indian subcontinent in terms of how it can be maintained. One thing I wonder about is infanticide, perhaps.
Maybe there are social taboos. Reproductive fitness is really low. For humans, it doesn't make sense, but the data is what it is. Indians are just really good at endogamy for some reason. In other populations, the general pattern is you see someone and you're like, "Oh, they're fine." It's not rocket science; it's universal human nature. But somehow Indians were able to escape that. No one really knows why. I've had multiple geneticists come up to me and be like, "What's up with this? I don't know." And I'm like, "Why are you asking me?" They're trying to figure out whether there's a secret sauce here because it doesn't make any sense for a mammal where the males in particular are highly polygynous.
Dwarkesh Patel
Are there any hypotheses out there trying to explain this?
Razib Khan
Not really. People say, "Oh, caste system, blah blah blah." But sexual exploitation of lower caste women by upper caste men has been a thing, so I do wonder what's up with that. There are some cases where you see things. For example, the Nair group in Kerala. Many of their women traditionally had relationships with Kerala Brahmins, Namathiri Brahmins, that weren't marriages but were consorts. The Nairs also did polyandry and other things. In the Nairs, you see a range of genetic distance to Namathiri Brahmins because those are their biological fathers. So there are exceptions to this. But in general, most Indians can figure out what their community is from. It's not typical. Most of the world is not like that. It's basically like all of India is populated by people like Ashkenazi Jews, very endogamous people. There are other examples like Ashkenazi Jews and the Roma, who themselves are of part Indian origin. The issue is just having a whole society like this is pretty weird. That is the innovation: "Let's have a whole society that's stratified." It's very interesting.
Dwarkesh Patel
Speaking of Ashkenazi Jews, I thought your post on that was very interesting. You talk about how before Jews were liberated and grew up in the 19th century, there just wasn't that much Jewish achievement. It made me wonder: are there other population groups in the world today that were bottlenecked by a similar process and are also very endogamous? Once they get to a point of prosperity and liberation like Jews went through in the 19th century, in the future, we'll be talking about how they're outputting a greater portion of the world's cultural heritage. Like parts of the world that are just going through industrialization now and might have small populations. Is there potential for a new "Ashkenazi Jew" in the next century or two?
Razib Khan
What you need is something like that. Ashkenazi Jews are highly endogamous and emerged in the context of Central East Europe as a middleman minority. The Haredi Jews dressed like Polish nobles because they worked for these Polish nobles as factors, tax collectors, and administrators. You have to look for something like that. One example, though not totally equivalent because endogamy is not a big issue here, is the Fujianese. Chinese from Fujian have traditionally done better on examinations going back a thousand years to the Song Dynasty. There were affirmative action quotas on people from Fujian. A lot of the rich Chinese, not all, but traditionally in Hong Kong, the elite families are Shanghainese and some Fujianese. These coastal southeast coastal people in China have been extremely enterprising. The central government in China has often clamped down on them, but the modern economy cannot. I think these populations might come into their own.
Dwarkesh Patel
Didn't you write somewhere else that the Chinese government for a long time, not just the CCP but in Chinese history, there have been many instances of the government trying to get rid of genetically distinct groups by breeding them into the larger stock? Potentially that reduces the odds of some outlier endogamous group.
Razib Khan
In China, the only equivalent is the Hakka in South China. The Hakka are descended from northern Chinese migrants and speak a dialect of Mandarin in the south, like in Guangdong where the Cantonese and Taishanese are. They still tend to intermarry because they're spatially isolated. But the Hakka are not like Ashkenazi Jews in having an ideological reason for their endogamy. Chinese lineages, to some extent like Indian lineages, are paternal. Your identity and clan are determined by who your father is. You might have lower status if your mother is an ethnic minority like Zhuang or Uighur, but informally.
Officially, you're part of the clan. That's how assimilation has happened genetically. The Cantonese have a minority of indigenous or south ethnic group ancestry. Some of their practices are clearly not Han Chinese, especially certain marriage practices. Most of the gene flow probably came from females, from non-Han that were assimilated in the area. The Han identity is very assimilative. North of the Yangtze, pretty much every Han sample I have has a little bit of West Eurasian ancestry. South of the Yangtze, none of them have it. I think most of that West Eurasian is probably assimilated Mongols and other groups.
The Mongols are about 10 percent West Eurasian. About 1 percent of Northern Chinese Han men have R1A, maybe 0.5 percent. R1A is mostly found in Iranians and Slavs, and Mongols have it because they assimilated Scythians, Sarmatians, and other Iranian steppe people. I think that's where that comes into the Chinese. You can go back to the Toba Turks and other groups after the fall of the Han Dynasty, 1,700 or 1,800 years ago. That's when they started introducing that genetic element to northern China, north of the Yangtze.
Dwarkesh Patel
Going back to India, there's been a lot of talk about how a lot of American CEOs of big tech companies are Indians, specifically from Brahmin jatis. Is there some particular reason that seems to be happening?
Razib Khan
The guy from Twitter is not a Brahmin, he's Bania. The Indian explanation, which you probably know, is that Brahmins are literate, they're symbolic manipulators. If you're working at Microsoft or Google, they tend to be particularly South Indian Brahmins, as opposed to North Indian Brahmins. There aren't that many of those. This goes back to the colonial period. South Indian Brahmins would migrate to the cities of North India to work in the Indian Civil Service. The reverse would not happen. This is a longstanding phenomenon of South Indian English-speaking Brahmins availing themselves of technology and higher education.
Tamil Brahmins, for example, are very well represented in engineering and software. That pipeline is what Indian Americans are going into as CEOs. They are highly overrepresented. The CEO of Microsoft and the CEO of Google are both South Indian Brahmins, both Telugu Brahmins. There are some debates about whether the guy at Microsoft is a Brahmin online because I can't tell these sorts of things. But anyway, I think Brahmins are like Ashkenazi Jews. They analogize themselves, particularly South Indian Brahmins. We have to distinguish that because you don't hear about a Guju Brahmin or a UP Brahmin.
Those people just stay where they are. They're local landed elites but not well known outside of the Indian subcontinent, or to be honest, within the Indian subcontinent.
Dwarkesh Patel
What explains why these North Indians were complacent and these South Indians were availing themselves of the resources?
Razib Khan
In UP and Bihar in particular, the elites tend to like to be big fish in small ponds. There aren't Rajput Thakurs all over the world from UP. Punjab is different. There are a lot of Punjabis all over the world of various groups, a lot of Jats, agricultural farmers in Central Valley countries all over the place. In contrast, in UP and Bihar, these North Indian states, there's less dynamism, less cultural dynamism. The behavioral economic literature shows a really strong preference for zero-sum gains. They want to be at the pinnacle of the local power structure rather than taking a risk going somewhere else where they might not be at the peak.
They might be way more well off in the aggregate, but they wouldn't be at the peak. For example, someone like Chandrasekhar, the Chandrasekhar Limit, he's a Tamil Brahmin by background. He settled in the United States eventually, but he was born in Lahore. His dad was working for the Indian Civil Service. If you read his biography, they experienced discrimination and prejudice being South Indians in the North. Chandrasekhar went to the United States during the time of segregation. They tried to put him in the Blacks-only area in St. Louis for some sports game. He experienced prejudice at the hand of Arthur Eddington in particular, who was pretty prejudiced against Indians and their ability to contribute to physics.
Dwarkesh Patel
Is that the guy who proved Einstein's relativity right?
Razib Khan
Yeah, empirically. At least that's Chandrasekhar's take. You don't know if it's 100 percent true that Eddington was really like that. Sometimes it turns out there are personal beefs going on. I don't think Eddington ever told his side. He died a long time ago. Chandrasekhar lived until about 1995, so 25 years ago. He was still around when I was in high school. I remember someone did a report on him, and it was hard to find information back then, but he was still giving quotes.
Dwarkesh Patel
Does the work you do involve a lot of traveling? You're writing about all these different areas of the world and their anthropological and genetic history. Does it require you or help you to travel to all these places, or are you able to do that from here?
Razib Khan
I do most of it in the United States. I've traveled a little bit, but not too much. I'm not a big traveler internationally. Some people do. Spencer Wells, who I worked with, my former boss, has traveled all over the world for National Geographic. That adds a lot of local color in terms of things you see and know. Whenever we talk about the Eurasian steppe, he's been there a lot, so he can add a lot. In a few places, I've been to Finland, so there are certain things I know about Finland. I've been to Italy. England just seemed like the United States but whiter. There wasn't a "whoa, I really understand the British people now" moment. It was just like, "Okay, they're drinking a lot." I'm not surprised by that, judging by all those British sitcoms and TV shows where they're drinking in the morning. I get it now.
Dwarkesh Patel
Knowing all that about the history of these different places, do you feel that when you visit a place or learn more about it, you think, "Oh, what they're doing today makes sense given what I know about the roots of what happened thousands of years ago"? Or does it feel random?
Razib Khan
No, it's not random. Sometimes someone does something or their family does something, and I'm like, "Oh, it's because of this." They're like, "What?" They don't know the antecedents of everything they do. A lot of times I do. For example, Americans are really ignorant about geography. In 2019, I was at a scientific conference in America studying human genetics. I was meeting people, networking. I met this Chinese geneticist, I think she was in grad school in the United States. I asked, "Where are you from?" She said, "I'm from a city between Beijing and Guangdong, exactly in the middle." My train of thought: I immediately blurted out "Wuhan."
She was shocked. "How'd you guess that?" Most Americans don't know Wuhan. Shanghai's in the middle, but if she was from Shanghai, she would say Shanghai. So it had to be another city. I happen to know there's a high-speed rail line between Beijing and Guangzhou, and its middle point is Wuhan. So I knew Wuhan was exactly in the middle. For an American, that's super amazing because we don't know any geography. Her friend was there, and I looked at him and said, "You're pretty tall. Are you from North China?" He said, "Yeah, I'm from the East." I said, "Shandong." He was like, "Whoa, how'd you know that?"
I said, "What's the easternmost province?" It's just an educated guess. If someone's talking about chowder and drinking tonic and says "wicked smart," I'm like, "Are you from Boston?" They're like, "Whoa, that's wicked crazy, how'd you know that?" Everybody outside of America rightfully assumes that Americans do not know anything about where they're from. So it's an incredible party trick with an American accent to be like, "You are from Prague."
Dwarkesh Patel
By the way, can you guess my jati?
Razib Khan
Well, I know you're Guju. But I couldn't guess it from the name. You look... are you like half Patel, half Bania? I'm just guessing.
Dwarkesh Patel
Yeah, that's exactly right. My dad is Patel and my mom is Bania.
Razib Khan
Okay, that's exactly correct. This was not a conspiracy between us. I didn't know that question was going to be asked. I just looked at him and made an educated guess.
Dwarkesh Patel
That alone should justify your subscription. Okay, I've had this question about the greater male variance theory for a long time. The idea is that men produce more geniuses but also more idiots. I've always wondered why that is the case. There must be some mechanism that increases the variation, giving you higher odds of being a genius but at the cost of higher odds of being an idiot, and it's more activated in men. What is the trade-off?
Razib Khan
I have to BS a little bit about the molecular mechanism. I haven't looked at it in detail in a while. One hypothesis is that we have one X chromosome. With women, there's inactivation of one X chromosome randomly in the cell or tissue region—Barr bodies. Every cell has an X chromosome, and they tend to clump. Barr bodies are X chromosomes that are inactivated, not expressing. One of the X chromosomes has to inactivate, and that's random. Let's say a woman has a major mutation on the X chromosome. She has another copy, but it could be that in that cell, there's a malfunction because the other copy is the one that's inactivated.
If you're a man, there's no choice; it's only one X chromosome. That limits the degree of freedom. If it's a good copy with good stuff going on, that's good. But if it's got bad stuff, you're screwed. The easiest explanation for why men have problems at the low end is that we have a load of deleterious alleles on our X chromosomes that are not masked because we only have one. That's one thing. We are the heterogametic sex. In birds, it's the opposite; females are the heterogametic sex. Developmentally, we all start out female. The female is the template. Men have to go through extra processes.
At the end of life, the opposite happens: women go through menopause, a proactive physiological shutdown, not just a long slow decline like we go through in our reproductive processes. But at the beginning of life, at the end of the first trimester, we go through a testosterone burst. The SRY gene, the sex-determining region, kicks in and we become male, we become masculinized. When you have extra developmental steps, that can mess things up. We also have higher testosterone. Testosterone is antagonistic to immune response. There are more males born than females, probably because the Y chromosome sperm is lighter than X chromosome sperm.
Male sperm probably has an advantage in speed. There are about 104 to 105 males born for every 100 females. But in utero, there's a strong suspicion from people who have done sampling on miscarriage fetuses that males are overrepresented. We start out with a bigger advantage, and we're already culled because of our genetic abnormality. Something on the order of 10 to 50 percent of fetuses miscarry. It's still not clear with the total numbers because it's really hard to track miscarriages early on.
Dwarkesh Patel
And so that explains the downward, the low end. In terms of why there might be more male "geniuses," I think the way you might want to look at it is there's really no reproductive value at the high end. It's just kind of like a freak thing. And if we're less developmentally stable, we can go off target a little bit more. That's the way I think of it. There's no reason you need to have your IQ be... once there's no reason you need to be able to do algebraic topology easily.
Razib Khan
Yeah.
Dwarkesh Patel
Is there some evidence in the genomic literature now with the most recent work that there is some enrichment for schizophrenia and other things with some of these educational attainment genes?
Razib Khan
Yeah, there is some evidence.
Dwarkesh Patel
But is there some reason in the ancestral environment why having a brain capable of algebraic topology would be advantageous? Like, is there something that a human would need to do? And then a separate question: do we have an explanation for why brain size decreased by about 10% or something like that after the... we're just smaller, so our bodies got smaller. When it got warmer, we got smaller, but also agriculture seems to have given us really, really weak bones. We got more fragile, more gracile. We shrunk some with agriculture. And so that natural process of that is smaller brains. I bet you average nutrition probably decreased some in terms of quality as opposed to reliability and consistency. That probably meant that smaller brain sizes are more optimal to survive through the attainment. We know smaller body sizes are for sure.
Razib Khan
We know smaller body sizes are. There's been a lot of negative selection in Southern Europe and in Asia for small body size. Last I checked, this is pretty clear that people in the eastern part of the Indian subcontinent are shorter genetically. Some of it is like East Asian ancestry, but I mean clearly Bengalis are just a short people. If you just meet people from Bangladesh or West Bengal in the West, they're chubby as hell because they get a lot to eat, so it's not genetics, it's environment. I used to, when I was little, people would say, "Oh well, your parents didn't eat a lot of meat." I'm just like, okay.
But now that I know about genetics, nutrition, and class background, I'm like, no. My family, people were obese in my family. They had enough to eat. They didn't suffer from the Bengal famine. And also my family's Muslims, so they eat beef and they got protein. No, they're just short because it's genetics. The Bengali population in Bangladesh, they have cholera resistance, obviously because of the issues with flooding and water. That's different than others. Indians have continental populations. There's some reasons why they're small too. I don't know why Bengalis are small, but that's obviously true.
Dwarkesh Patel
So, sorry, what's the link between cholera and height?
Razib Khan
There's no link. I'm just saying there's been studies in selection. There's selection for resistance to cholera in Bengal. It's one of the canonical examples. The vibrio whatever microbe, there's clearly a strong selection because of the cholera over the last couple of centuries.
Dwarkesh Patel
And then what do you make of the self-domestication hypothesis? The idea that there's a set of genes that happen together, they're associated in many different mammals with domestication: smaller jaws, males and females looking similar, and then less intelligence. There's a cluster of other things. Basically, the idea is that the same thing happened to humans during the agricultural revolution. What do you make of that idea?
Razib Khan
I think it's plausible. I think it hasn't really panned out in terms of the genomics. Let's just put it that way. This hypothesis has been around for a generation and it hasn't really panned out in terms of the genomics. So I guess what I would say is it could be that humans are special in some distinct ways. Because it's been studied in foxes and other organisms extensively, and dogs. There's a spackling pattern, floppy ears, really common patterns across mammals because the same developmental pathways are tuned. We obviously don't have floppy ears and we don't show high coat patterning. So I think it's a great idea. I just don't know for sure how it operationalizes in humans. It's been a generation. We have genomic resources, and it hasn't really... I haven't seen too much advancement in that direction.
Dwarkesh Patel
Gotcha. Okay, so this morning you tweeted, "If everyone who attends a church thinks that the point of church is to bask in the work of the fellow parishioners rather than worshiping God, the church won't last long." And then you followed that up with a tweet that said, in parentheses, "I'm not talking about religion." So I generally don't know what you're referring to in that tweet. I don't know if you meant to keep it unsaid, but I was just kind of curious.
Razib Khan
Yeah, I was being stressed, and I was just having a discussion. I'll tell you what it was. I was having a discussion with a scientist friend of mine. We were talking about collegiality and truth. Sometimes it seems like in science today, not just online, but just in general, the community and comfort is prioritized. And a lot of it's fake. Science is like management consulting, it's up or out. So all this stuff about support and it's just fake. One percent of one percent of incoming graduate students will have a tenured R1 research position. So all this stuff about "we're here to support you" is just to separate the wheat from the chaff.
There's a lot of talk about community and not making people uncomfortable and inclusion and equity. And I'm just like, science is super inequitous. It's not like pediatrics. There are superstar pediatricians, but the average pediatrician makes a difference. In science, you have a few superstars. It's hyper-Pareto principle. It's not the 80/20, it's the 5/95. So it's just a little strange. The whole idea is truth. I've seen things where people are like, "That's just uncomfortable, you can't say that, that makes people..." I'm just like, one, it's a very winnowing profession. They haven't changed that no matter what you say.
You can repeat these mantras, but it doesn't matter. It's a winnowing profession. And the other thing is, if science was here for the truth, if that's not the primary focus, if you're here for quality of life, why are we funding it then?
Dwarkesh Patel
Yeah, it makes sense. It's not only unequal in the sense that there's a log curve where a small minority of scientists make a much larger portion of the total contribution to science. But it's also unequal. I think it was you who said this or wrote about this. Professors, the career of professorship has the highest heredity in the sense that the highest correlation between the parent being a professor and a child being a professor.
Razib Khan
Yeah. I mean, I probably retweeted it. That's obvious. Everybody in science knows this. My dad was a professor, by the way. But anyway, I'm not. But everybody who... People in science, people who go into graduate school in academic science, they reason that "first generation" is a thing because it's so skewed toward professional managerial class people in general. It's very, very class biased. Even people who come from professional managerial backgrounds, if they didn't come from academia, they don't always know everything. I have a friend who came from a very upper, upper middle class background.
He admitted he had to learn some things in terms of what you do to make it in science. I think his dad's a lawyer. It's the same thing in medicine. I have a friend who's in medicine, and I think his parents are engineers. He said that they told him for medical school interviews, it's going to count against you that your parents aren't doctors, because they just assume you don't know as much about how to make it in the profession. There's tacit stuff that gets passed on. I have a friend, he is a research one professor. He does have tenure. He has succeeded. But he comes from a very working class background.
By the time he got to the postdoctoral fellow stage, he was talking to people about their choices they made as undergrads. And he just thought to himself... He's in a top ten institution, not Harvard, but he's doing really well. He literally told me, "I just thought to myself, I never had a chance." He never planned this way. He never optimized his own life because he didn't have that background. On the margin, it makes a big difference.
Dwarkesh Patel
I think this is why there's a lot of virtue signaling from some people. There's a professor, I'm not going to name who it is explicitly, but people who follow academic Twitter probably know who I'm talking about. They work in biomedical science and they do periodic virtue signaling, just standard progressive stuff. But I think their uncle was a Nobel Prize winner and they did research in their uncle's lab when they were in high school. This is a person who got a huge leg up by family. They're smart, okay, but they knew exactly how to succeed in science because they had all the family connections in the world.
I think they overcompensate. That's what everyone assumes privately. There's other people like that. There's one guy online who's super progressive, but a friend of mine told me he's a notorious dick to people in his lab. He's a really bad boss, really mean, really demanding. So obviously he's just covering his butt on social media. Anyway, my tweet was basically alluding to the fact that if you're not there for the right reasons, if everyone's just there to collect a salary, or they don't know what to do with their life, or they like hanging out with this crew and being on the same ideological team...
what is the point of science then? Why are you here? Why aren't you an accountant or a CPA? You're supposed to be here for a higher calling. So the tweet was: the parishioner would be the person involved in the laboratory, the research institution, and God is the truth. If you're not there for the truth, eventually the institution is not going to make it. It's just going to dissolve. At the end of the day, if you don't have passion for research, if you don't have passion for the truth, what's the point?
There's this professor we both know, but obviously I'm not going to say who it is. His kids also want to become professors. The kid just graduated high school, and they had a peer-reviewed published paper while they were in high school because the professor had guided them so they would be in a good position to become a professor themselves. When you consider that kind of advantage versus somebody who just goes to college and thinks, "Oh, this subject seems interesting to me, maybe I should consider a graduate academic career here," there's no comparison. The level of advantage you have if you've been planning it out...
Razib Khan
I don't know what's going on here. I've talked to multiple friends who are at the postdoc level. They talk about how they can see over the last 10 years a massive inflation in publication. Their postdocs now and some of their graduate students have two or three publications coming in, and they didn't have any publications until their third or fourth year of graduate school, and they went to the same university. So what's happening here? In the past, they didn't publish as much, but they did a lot of science. This is one of those issues when you devise a metric to measure something, eventually the metric gets distorted. The metric's getting distorted. There are people who are producing... some of these researchers who have 30 papers a year... you're not really contributing to it.
Dwarkesh Patel
There are full-time bloggers who don't output as many blog posts as you are outputting papers.
Razib Khan
Yeah, yeah.
Dwarkesh Patel
Okay, so within the spheres that I travel, and maybe that you're traveling as well, on EA-adjacent stuff, there's this idea that we are living in a very important time in history. There's a step function, so you have different steps: agriculture, domestication, metallurgy, industrialization, and we're at another step right now. So from all the people I know, you know the most about history, especially ancient history. Do you view history as a series of step functions, each one the newest one more important than the last? Or do you view it as just a gradual exponential curve? What is your view of the long view of history?
Razib Khan
I think it's mostly gradual. We reify it into a step, but I think we might actually be a step now. If the slope is steep enough, it's a step. So I think we might be at a step now. There have been steps in the past, but mostly we reify. The Industrial Revolution, my understanding from economic history, is really more gradual and exponential than a revolution. Agriculture was probably like that as well. Peter Turchin's work with some of his collaborators indicates that the Axial Age was actually more gradual. The coalescing of ideas wasn't like a step within one century around 600 BC or whatever.
So most of the time, I think we tend to simplify it as a step. But right now, we live in an age of miracles that we don't take for granted because we're just scrambling. We have supercomputers in our pockets. They're called phones. We're doing science fictional video stuff. My kids, my oldest is 10, my youngest is 5. They're a little dubious about this idea sometimes. You use the phone for these non-video calls, and they're very confused why people would do that. They're very skeptical that this is what this phone was originally designed for. They see a flip phone and they're very confused with how that could be a phone.
"What is this ancient technology from 2007?" We have a desk full of old phones that we never threw away. My kids saw the phone and they were like, "What is this thing?" And I'm like, "Oh, that's a phone." And they're like, "No, but a phone's square." And they have an old rotary toy phone. They traditionally use it as a hammer. They don't really know what the form factor is. It's totally weird for them. So we are living through a radical change in terms of our social technology or information technology. Most of your viewers probably know Kurzweil: information technology is exponential. So there are some radical changes going on right now, and we need to think about what that means.
VR is going to be a big deal. I said 20 years ago, probably... I know Holden Karnofsky. I think he might be one of the people you're talking about regarding this century. I said this might be the last century of humans in a way that we would recognize, or it might be a century of regression. I think we are in a metastable state right now. I'm looking at you right now and you look like a primate. You are a primate. But you have access to all this technology. When I see myself, I don't see a primate. I see Razib. But if you look at another person, it's really visible that you're an animal of that particular lineage.
You see the way they move. So how long is this going to persist? We obviously evolved during the Pleistocene, even earlier, with a lot of our instincts. But now we have the ability to destroy our civilization. We're not going to exterminate all life on Earth. Probably not even all humans. There's probably going to be people in the Southern Hemisphere that are going to survive. But we could destroy our civilization. Civilizations have been destroyed in the past by overreach. But those civilizations had local collapses, local regressions, and then we got more robust with social technology. For example, you see the Chinese dynastic cycle keeps shrinking every single time in terms of the chaotic interregnum.
That means the previous dynasty's institutional structure is probably more robust to shocks, and it can rewind itself back up relatively easily. The first big unwinding is the Zhou dynasty, that's like 500 years decline, warring states. Then the Qin-Han dynasty, then there's like a 300 year period of collapse. Then there's like a 150 year period of collapse. It just keeps shrinking every single time, showing that cultural or social technology is getting better, information technology is getting better. But now we're global, so even if there's a 50-year collapse, there's a lot of stuff that we're going to lose.
Sam and others have talked about the fact that we can't make rockets the way we could 50 years ago because a lot of those engineers are not... and our military runs on COBOL software that barely anyone can read. It's like cuneiform. We laugh at the Babylonians for using Sumerian liturgy 2,000 years after the last native Sumerian speaker died. But we are going to have a situation soon where there's going to be almost no COBOL programmers, but we have COBOL software bases. People are going to have to train and learn from these manuals, these ancient texts from the 1970s. It's not that difficult, it's feasible.
But the issue is timing, because you might not have enough programmers to service all the code that you have. These are real issues that we have to deal with as primates who've organically developed this technological system and are trying to figure out how to make it work.
Dwarkesh Patel
Yeah, sustain it. I had Sam on just a little while back to discuss exactly these topics, so for the listeners who are interested, definitely check out that episode. Even with something like COBOL or somebody who started computer science, even something that's that legible, you can have implicit knowledge from previous programmers or how does the entire system work. This is literally a written word, that's what a program is, so that's super legible. Compare that to other forms of manufacturing. I know a guy who worked at a fertilizer plant. Maybe I shouldn't say this on air, but he basically said if somebody did something to this fertilizer plant, that's a famine right there.
We've lost the source of nitrogen here. So going back, you made a very interesting comment about we're at a point where we're using our computers to do magic, but the person behind that computer is a primate. Knowing what you know about genetics and the potential malleability of our genetics, do you see the future iteration as us adding on to or modifying or selecting on the same biological substrate? Or do you think it's more feasible that we just move on to entirely virtual, like minds living on computers? Which seems more feasible to you?
Razib Khan
I think in the biological program of redoing ourselves, it's not straightforward in terms of minimal risk. There's going to be a lot of false starts, which is going to be kind of crazy. But I do think people will improve themselves. I think they will edit themselves better over the next century. But I think there's going to be some integration with brain-computer interfaces. Yes, I do. I haven't followed it closely, but I do move in some of the similar circles. I think brain-computer interfaces are going to be a big deal, and they're going to really change the game. The issue there is... I guess I think of gene editing more as Smithian growth, increased efficiencies, because we have the genetic variation now.
We can make him smarter. We have the technology. We're not there yet, but I can see that. We can all understand the basic logic there. John von Neumann existed. The experiment has been done. So we can aspire to create a bunch of von Neumanns. That's great. Now, the issue with human-computer interfaces is that's never been done. That is an innovation. That's technology-driven growth that's increasing the baseline productivity by crazy amounts. That could be the possible quantum leap. That could be a big deal in a good way or a bad way. A lot of your listeners know about all the existential risk and hostile AI and general artificial intelligence.
But perhaps it'll start with us. Perhaps Skynet will be some uploaded crazy kid. Maybe it's going to be a situation like going to the New World. There were attempts to go to the New World. They didn't know the New World was there, but there were people in the Middle Ages who left for the West and they never... the ships just disappeared. They died at sea. So there's going to be people who do things like going to Mars, high mortality rates. Similarly with these human-computer interfaces, there's going to be high death rates. People just disappear into the ether. But the first person that gets in there, it's going to be like Christopher Columbus.
They may be in a very advantageous position. Instead of being a primitive prototype, they might basically have all the land in this cyberspace. They do all the learnings really early on. They iterate, they pivot. And so they can be the god of that universe. I'm just speculating here. I'm trying to say that the possibilities there are pretty extensive, pretty high variance.
Dwarkesh Patel
In the short term, what is the landscape of just gene editing, polygenic selection? What does that look like in the next 10 to 20 years? Is there potential that you could raise your kid's IQ by one or two standard deviations? Or are these going to be marginal improvements? By the time I'm ready to have kids, what will it look like?
Razib Khan
I think with gene editing, the intelligence thing is going to be like 20 years. Let's say 20 years. In the short term, gene editing will probably cure cystic fibrosis. Cure sickle cell. These are Mendelian, positive Mendelian disease with large effect loci. People just have issues. There's always a delivery problem. There's always a problem with off-target effects, which could cause mutations, could cause cancers. But if you're cystic fibrosis, you're going to be dead by 45. You're going to take the risk. So I think that's honestly going to be the first thing. The first thing is going to be transfection or gene editing of adults for Mendelian disease.
That's the next 10 years. It's already happening now. They're already curing people of sickle cell. I think they're working literally right now as we record on cystic fibrosis and ALS, because they're degenerative diseases that kill people in the prime of their lives. 20 years, that's a long time. We have 40-year-old IVF babies now, I think almost 40. So I think 20 years, yes, you will start to see parents editing the genes of their offspring. I think intelligence is difficult because it's a polygenic trait with a lot of different genomic positions. I wonder if they're going to go for other things first and then work to it.
There's that theory that Armand Leroi was talking about, and other people: what you should do is focus on mutations and other things, try to fix those, and see if that inadvertently increases intelligence. Rather than focusing on getting gain-of-function genes, which is like, how do you identify those? Fix all of your copy errors. Because that's a finite number. Compare to the pedigree of the parents, look at the de novo mutations, look at the parents' de novo mutations against the idealized reference, etc. That might be much more feasible.
Dwarkesh Patel
Oh, interesting. I didn't know about that. Do you have an estimate for how many SNPs affect the variation in intelligence between people?
Razib Khan
Let's see, how many SNPs? It's going to be an order of thousands. Just, okay, yeah.
Dwarkesh Patel
Yeah.
Razib Khan
Yeah, yeah, yeah. All right, so just some meta questions to close out on. You distinguish yourself in your career by being somebody who's an expert in history and an expert in genomics and life sciences more generally. Are there other fields where you think knowledge of history would be very useful in setting up a separate niche? Because you have a niche in history and genomics, but it would be hard to imagine, for example, somebody knowing a lot about history and computer science having a special niche, right?
Cultural evolution. That's what Joe Henrich and some of his people wanted to do. I think in cultural evolution, there's going to be a lot of gains because Peter Turchin, Joe Henrich, these people are applying evolutionary principles to historical processes. Having the empirical data set is really important. This is a really new, nascent field. So I think that's going to be the big thing that people should focus on. Peter said, "Oh, get anthropology knowledge." Oh my God. I don't think the short-term knowledge is super important. I think having a deep, deep, deep knowledge about historical arts would probably be pretty useful.
Dwarkesh Patel
Interesting. Joe and his group, they're working on that. They're moving into history. They're doing some serious imperialism that's causing problems.
Razib Khan
Yeah.
Dwarkesh Patel
Causing problems how?
Razib Khan
Historians do not like the turf infringement. That's what I'm saying.
Dwarkesh Patel
I see. Yeah, yeah. Right. So you're one of the top bloggers on Substack and you have this deeply technical blog on the science of genomics and other things. You would think beforehand that your prior would be like, how many people are going to be able to understand this or be interested in this? But in fact, you're one of the top people on Substack. What has the experience of that been like? Has it surprised you, the popularity of your work?
Razib Khan
Honestly, no. Well, I mean, it surprised me—okay, I'm going to be honest—it probably surprised me how many people are willing to pay. But people have been reading me for a long time, so I just kind of professionalized it some. It's been great. It's really helped me figure out what people are interested in terms of what they're willing to pay for. It's giving me some direction, I guess. But I plan to do basically what I've continued to have done in various ways in the past and into the future. Thinking of the startup world the way I would pivot and iterate is what I'm thinking.
Dwarkesh Patel
Final question: do you have any advice for people who want to write about technical topics in a way that's very interesting to a broad audience?
Razib Khan
You have to make it relevant to them somehow. For example, let's say you want to write about signal detection, text-to-speech type stuff. There are things people are super interested in. For example, people are super interested in Jewish genetics. People are super interested in the genetic architecture of skin color. I can talk about the genetic architecture of something else and it wouldn't be as super interesting. So you have to find the domain that they're interested in and then apply your method. If you're interested in... actually, there's a Substack on personality that talks about personality and using machine learning methods to classify personality. Machine learning is technical, but personality is interesting.
Dwarkesh Patel
Interesting. Yeah, that's good advice. All right, Razib. Thanks so much for coming on the podcast. Thanks for your time.
Razib Khan
You too.