Razib Khan's Unsupervised Learning

Today Razib talks to Tim Lee, a previous guest on Unsupervised Learning. Lee hosts Understanding AI. Lee covered tech more generally for a decade for Washington Post, Ars Technica, and Vox.com. He has a master's degree in computer science from Princeton. Lee writes extensively about general AI issues, from Deep Research’s capabilities to the state of large language models. But one of the major areas he has focused on is self-driving cars. With expansion of Waymo to Austin, and this June’s debut of Tesla’s robotaxis, Razib wanted to talk to Lee about the state of the industry.

They discuss the controversies relating to safety and self-driving cars. Is it true, as some research suggests, that Waymo and self-driving cars are safer than human-driven cars? What about the accidents Waymos have been implicated in? Is it true that they were actually due to human error and recklessness, rather than the self-driving cars themselves? Lee also contrasts the different companies’ strategies in the sector, from Waymo to Zoox to Tesla. Razib also asks him about the fact that self-driving cars’ imminent arrival seems to have been overhyped five years ago, with Andrew Yang predicting trucker mass unemployment, to the reality that Waymo has now surpassed Lyft in ride volume in San Francisco. They also discuss the limitations of self-driving cars in terms of their ability to navigate cities and regions where snow might be a major impediment, and why there has been a delay in their expansion to freeway routes.

Subscribe now

Give a gift subscription

Share

This podcast accompanies my post Germans are from Finland, Finns are from Yakutia.

The two preprints at the heart of this post are, Postglacial genomes from foragers across Northern Eurasia reveal prehistoric mobility associated with the spread of the Uralic and Yeniseian languages and Steppe Ancestry in Western Eurasia and the Spread of the Germanic Languages.

Subscribe now

Give a gift subscription

Share

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.

It’s no accident that in political philosophy and theory, we continue to debate to what extent man is egalitarian or hierarchical by nature, and therefore how to structure and organize a society accounting for these biases. As much as socialists might wish to believe otherwise, all complex societies seem to have a top-down structure of power with rulers above the governed. There is a reason that anarchists are perceived to be utopians. But hierarchy still has its limits and natural checks, everywhere from primate societies to human empires. As the late Frans de Waal observed in Chimpanzee Politics, among our great ape relatives a tyrant’s reign is often abruptly and brutally curtailed when a coalition of beta males who chafed at his excesses agree that he has overreached. The violent deaths of many Roman Emperors at the hands of usurpers illustrate the fine line in our species between long-suffering subordinate and fed-up rebel. Stable societies require a balance between equality and hierarchy. This reality is rooted in our very instincts, and born out by the lessons of history. A world without winners and losers is sterile, but brutal winner-take-all competition too often unleashes savage violence with costs to the society’s long-term stability.

Before Peter Turchin became a very well known pundit, celebrated for his prediction of chaos in 2020, he was a quantitative ecologist who turned his focus to historical questions, and in particular to modeling how societies emerged, became complex and then decayed and fell. Historical Dynamics: Why States Rise and Fall came out 22 years ago, but in it Turchin introduces a new way of thinking about history, pioneering the field of cliodynamics. Translating models from ecology into human history, like logistic growth, Turchin demonstrates how to formally describe the rise, apogee and decline of states and societies because of internal variables like social cohesion and the variable fraction of the population that are aspiring elites. Lurking within the differential equations and simulations are insights about the necessity of a balance between hierarchy and egalitarianism. Large polities‌ are hierarchical, but excessive oppression ultimately leads to rebellion and collapse.

The same year Turchin published Historical Dynamics, evolutionary biologist David Sloan Wilson wrote Darwin's Cathedral: Evolution, Religion, and the Nature of Society. Though the book was ostensibly about the evolutionary framework for religion as a social and cultural adaptation, it was also Wilson’s attempt to resurrect a functionalist understanding of culture in anthropology, a tradition that goes back to Emile Durkheim. Similar to Turchin, Wilson in Darwin's Cathedral argues societies should be thought of as social organisms, and that as such, they are more or less adapted to their environments. A classical illustration from the book is Balinese water temples, which nominally serve a religious function, but also coordinate water resources across many farms. Darwin's Cathedral illustrates how human societies and human nature are not best understood by reducing everything to singular impulses or dynamics like “will to power” or “to each according to their need.” Our nature is complex, and the social institutions we develop to maintain peace and establish harmony are similarly multifaceted, with elements that might occasionally work at cross-purposes.

Focused on the organizing unit of fundamental importance in the Western political tradition, Larry Siedentop’s Inventing the Individual: The Origins of Western Liberalism argues that religion is essential. Anyone with a passing familiarity with the Bible, and in particular the New Testament, sees clearly that there is a powerful clarion call toward equality before God within the Christian tradition, and Siedentop develops the thesis that political and legal equality in the West cannot be understood except as an inference derived from the conceptual framework of religious equality. Christianity as adopted by Rome did not usher in an age of egalitarianism; slavery was still practiced and there were still rulers and the ruled. But Christianity smoothed the sharp edges of the ancient world’s brutality for the least of the Empire’s subjects. Siedentop also argues that development of egalitarianism inside the Roman Catholic Church’s hierarchy, where priests from peasant backgrounds could in some cases rise to positions of ultimate power over clerics from the nobility, was essential as a model and prototype for the norm of blind justice and merit irrespective of social background or legal status. Because of the Church’s independence from political power after the fall of Rome it could experiment and evolve in ways that likely would have been impossible under direct governmental control. These legal and institutional innovations, like canon law’s influence on civil law, eventually spread to secular governments, finally becoming normative in early modernity.

Kyle Harper’s ​​From Shame to Sin: The Christian Transformation of Sexual Morality in Late Antiquity also articulates this vision of individualism being the path to egalitarianism. In pre-Christian Rome, Harper notes that sexual abuse of slaves was taken for granted, and not viewed as immoral. Honor and shame governed society; the Emperor Elagabalus was mocked for his homosexuality not because he was engaging in homosexual behavior itself, but because as a high-status individual he took the “passive” role. Harper shows how Christianity changed the moral focus, demoting relational elements from being supreme to morality, instead making all Christians subject to the same ethics before God. The Church and its moral campaigns illustrate the paradox of a hierarchical organization coercively imposing moral norms that resulted in greater egalitarianism, with less sexual abuse of inferiors by superiors, and protections afforded to women. But with these protections came new responsibilities and expectations of sexual purity imposed downward in the social hierarchy.

But even if legal equality has been the overall trend across history, heritable differences in status too have remained constant. At least that is Gregory Clark’s argument in his 2015 The Son Also Rises: Surnames and the History of Social Mobility. He shows that in England, as in other societies like Italy and India, elite families have maintained power and status across centuries and even millennia. Nearly 1,000 years after the Normans’ 1066 arrival to England, those with Norman surnames are still overrepresented among the British elite, in particular in the military. Subsequent work has supported some of Clark’s arguments in the last few years, showing that social status seems to follow the pattern of a heritable trait. Social status of a lineage decays over time, or increases in other cases, but the overall pattern is far more stable than our intuitions about inter-generational opportunity might have us believe.

Who’s a Carthaginian? Genetic Study Revises Ancestry of Rome’s Ancient Nemesis. The original Levantine genetic imprint in western Punic societies seems to have been sharply attenuated, replaced by indigenous Sicilian or North African heritage. This shows the importance of memes, versus genes, in forming an ethnolinguistic identity (in contrast, Greek colonies remained genetically very similar to the motherland).

It Took a Century to Find This Colossal Squid. They finally recorded this squid in its native deep-water environment. Incredible, because they’ve known of its existence for a century.

Will Oakland Finally Reject Progressive Politics? No, it won’t. Barbara Lee won, and Oakland’s dysfunction will continue (meanwhile, Dan Luria across the bay in San Francisco seems to be turning things around).

Revolt of the Dandies. There is a culture war here, though not on Left-Right terms. The sartorial revolution of the 1960’s, with formality’s demotion, is now six decades in the past, and new generations begin returning to older norms.

Alien Poop Means We Are Not Alone. But Let Me Just Adjust This Model Parameter… Erik Hoel announces the paradigm shift: there is now good circumstantial evidence for life on another planet.

Punic people were genetically diverse with almost no Levantine ancestors. Paper that inspired media reaction above in The New York Times, and elsewhere, about genetically distinct Punic societies.

An Eocene Origin of Passerine Birds Estimated Using Bayesian Tip Dating with Fossil Occurrences. This method of analyzing fossils in a model-based framework confirms the intuition that birds diversified (again) after the extinction of the dinosaurs, contradicting molecular results that give a more gradual and deeper diversification.

Widespread recessive effects on common diseases in a cohort of 44,000 British Pakistanis and Bangladeshis with high autozygosity. We know inbreeding is bad, but genetics helps us quantify it harm. Very relevant in Britain, where the NHS is left footing the bill.

Economic Consequences of Kinship: Evidence From U.S. Bans on Cousin Marriage. Basically banning cousin marriage disperses lineages, and sends young people in search of better economic opportunities. Inbreeding is a social ill, not just a genetic one.

A male Denisovan mandible from Pleistocene Taiwan. Finally, a Denisovan from the tropics, where most of the living Denisovan ancestry concentrates today.

Heritability of human lifespan is about 50% when confounding factors are addressed. Heritability is generally assumed to be as low as 10-20% for lifespan, but this preprint claims to have controlled for more factors, isolating the genetic signal with stronger fidelity.

There’s still no free lunch, free subscribers; my most in-depth pieces for this Substack remain beyond the paywall.

Germans are from Finland, Finns are from Yakutia:

Intriguingly, the men who carried I1 were not indigenous to Scandinavia. Clues in their genes indicate that their origins lay to the east, most likely in Finland. And if not Finland itself, somewhere nearby in the eastern Baltic, perhaps Estonia. The new genetic element in Scandinavia diverged somewhat from earlier Corded Ware and Bell Beaker streams deriving from points south because it was notably enriched for Baltic forager ancestry: in other words, EEHG heritage likely from a CCC-descended population. This would align well with haplogroup I1, which is not canonically Indo-European, and had been a vanishingly rare paternal lineage before 2000 BC.

I1 descends from the broader haplogroup I, and it was other branches, especially I2, that overwhelmingly dominated among western Mesolithic forager populations across Europe 10,000 years ago. Even today all the branches of I are quite rare elsewhere. I’s unrivaled ubiquity in Western and Central Europe was broken only by the inrush of Neolithic farmers, who arrived from Anatolia 9,000 years ago, demographically swamping out the foragers. The most likely ur-heimat for these Baltic forager-enriched people to have hailed from is Finland’s lost Kiukainen culture, itself descended from Corded Ware agriculturalists and the predominantly EEHG-ancestry CCC people. The most persuasive scenario today is that after 2000 BC, a Kiukainen splinter group fatefully crossed the Gulf of Bothnia into Scandinavia proper and conquered the indigenous peoples, before going on to do the same in Norway and Denmark. When the dust cleared, the ethno-linguistic configuration still dominant in Scandinavia to this day had crystallized, from the windswept heaths of Jutland to Norway’s fjords and all along the shores of the Baltic.

Lost Green Saharans: ancient DNA unearths a new race from a verdant North African interlude:

Until TKK was genotyped, the ANA were what paleogeneticist David Reich calls a “ghost population.” Like Planet X whose existence we intuit through its gravitational pull on other observable bodies, an ANA imprint was hinted at in Taforalt and many modern Saharan and Mediterranean genomes. But subtle hints take you only so far; the foundations of human evolutionary history require concrete genomes extracted from the remains of those who died millennia ago, those lucky random genetic snapshots from bygone ages. Only when we have an individual human’s genome extracted from ancient DNA does the history of a lost population become so visible to us, its light illuminating ages of prehistory that had been wholly hidden. The passage of time will likely deliver more genomes to flesh out the history of the ANA people. Was TKK part of a relict population unique for its high fraction of ANA, or actually typical of a broader set of peoples across the Sahara who for tens of millennia remained relatively untouched by both Eurasian and sub-Saharan heritage? We don’t know, but someday we will.

More generally, ANA reminds us how much we have yet to learn about the emergence of humans in the recent past. A well distributed sampling of ancient DNA now blankets Eurasia, informatively reaching deep into our past, but the subtropical and tropical zones have only ever been extremely lightly explored. Perhaps a day will come when we run out of genomes, but in the world’s torrid zones, today that is unimaginable: we’ve barely begun to look. Every new discovery brings the possibility that untold ghost populations, physically extinct, but with echoes possibly even discernible in our DNA, will join the human menagerie. Not long ago, we thought we had nailed down the full plotline of humanity’s emergence in Africa 200,000 years ago. But now we behold lost chapters buried deep in the Libyan desert, a genetic trove of unfathomed North African tribes neither black nor Eurasian, but humans on an entirely distinct journey altogether.

I started a new feature for paying subscribers this year, an Unsupervised Learning Journal Club, briskly reviewing a notable paper or preprint. At the end of each edition, I invite subscribers to vote on papers/preprints for the future editions. And indeed since the last Time Well Spent, both the third and fourth topics have been selected by reader vote. I first looked at a 2023 preprint on Albanian genetic origins and then reviewed a 2025 paper exploring modern humans’ deep hybrid origins. That fourth Journal Club came out this past week, so you are still in time to vote for next edition’s paper.

Eternally Illyrian: How Albanians resisted Rome and outlasted a Slavic onslaught:

Though the core of Albanian ancestry, a composite of European farmers and Yamnaya herders, dates to the Iron Age, this paper reiterates the scale of impactful later migrations. The Roman Empire those Illyrian emperors kept afloat changed the region by facilitating migration, and its subsequent withdrawal triggered even more change. First, from the east, whether from populous Anatolia or the venerable Levant, newcomers arrived in Illyria and settled in the western Balkan cities. Then, once imperial authority collapsed in the Balkans after 600 AD, the Slav migration transformed the region, reaching even the remote Albanian highlands. Though the Roman era’s written languages were Latin and Greek, Albanian’s persistence today makes clear that in the Illyrian uplands numerous holdouts consistently resisted assimilation, whether into the Roman world or the Slavic tribes. The Empire’s decline and the abandonment of its formerly glittering cities saw spoken Latin and Greek wither after 600 AD, but the hardy peasants and herders of the deep interior remained much as they always had been, their humble lifestyle equipping them to weather Rome’s collapse. The Illyrians had been there long before both the Romans and the Greeks, and a subset of their descendants, the Albanians, far outlasted those Latin and Greek speakers when civilization gave way to barbarism, thus delivering to the very doorstep of modernity the cadences of Iron-Age dialects and the long obscure culture of Illyria.

Homo with a side of sapiens: the brainy silent partner we co-opted 300,000 years ago:

A plain reading of this result is that modern humans are a synthetic population, and our functions and features are a melange. Most of our biological processes derive from the same lineage that gave rise to Neanderthals and Denisovans. But, a minority of our heritage is a holdover from some very different and alien population that was notably distinct from the Eurasian hominins likely demographically dominant within Africa (population A). That exotic population, labeled B here, seems to have bequeathed our lineage much more of its cognitive function, and perhaps crucially earned us our self-chosen sobriquet Homo sapiens. Modern humans’ uniqueness, and our subsequent rise to prominence, may then owe significantly to a fortuitous admixture event somewhere in Africa 300,000 years ago, where a numerically superior and robust population of hominins, like the giant Nephilim demigods of the Bible, fatefully absorbed a smaller, but more cognitively adept group of humans.

Editions 1 and 2:

Wealth, war and worse: plague’s ubiquity across millennia of human conquest

Where Queens Ruled: ancient DNA confirms legendary Matrilineal Celts were no exception

For free subscribers: a sense of the format from my coverage of two favorite 2024 papers:

The other man: Neanderthal findings test our power of imagination

We were selected: tracing what humans were made for

All my podcasts go ungated two weeks after their Substack release. So I encourage subscribers on the free plan who’d like to automatically get them to subscribe to that podcast stream (Apple, Stitcher, and Spotify). If you want to listen on YouTube, please subscribe.

Here are the podcasts since the last Time Well Spent:

• Bonus monologue: ancient North Africans and the Green Sahara

• Matt Welch: from blog pioneer to podcasting mainstay

• Jacob Shell: academia must diversify or die

• John Sailer: a time of troubles in higher education

• Bonus monologue: man the hybrid monster

• Laura Spinney: rise of the proto-Indo-Europeans

My own domain also has all my links and updates: https://www.razib.com, including links to the few different podcasts I’ve contributed to or run, my total RSS feed, and my more mainstream or print articles when I remember to post them, my Twitter, the occasional guest appearance, etc.

• DM me

• Facebook message me

• Some of my past pieces for Palladium Magazine, The New York Times, Slate, Quillette and Nautilus.

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 or what you hope to read about this year, put them here.

Subscribe now

Give a gift subscription

Subscribe now

Give a gift subscription

Share

Remember how simple our story used to be? Twenty-five years ago, our line was that the fateful lineage leading to us, the anatomically modern populations who dominate the planet today, began with a big bang about 200,000 years ago. A single massive wave like a demographic tsunami rushing concentrically outward swept away all competing branches of humanity, snuffing out their separate flames, mercilessly pruning so many disparate branches from the vast and ancient tree that stretched back over a million years to the roots of our hominin lineage. Today, a generation on, by the light of incessant bombshells of insight from computationally powered modern genomics and the laboratory wizardry of paleogenomics, our simple story is long gone; and every few years seem to produce yet another verified side plot or narrative twist to complicate the story of us. Our terse Hemingway novella insists on becoming a sprawling 19th-century Russian epic, its confoundingly complex plotlines interweaving across continents and epochs.

Perhaps we shouldn’t be surprised. With the relatively small data sets of the late 20th century, it stood to reason that our best models would be simple, too. When raw material is sharply constrained, the edifices produced tend towards minimalism. Today, with the whole genomes of thousands of ancient humans, including from three major lineages, Neanderthals, Denisovans and the African populations that led to all of us, we are no longer constrained by a lack of relevant data. A vast theoretical cathedral is now being erected, no rude preliminary sketch, but a detailed demographic simulacrum of what once was, the dream of eventually working out every detail and nuance more fathomable by the day.

Much of the story to come in the next few decades of human population genetics will unfold in sub-Saharan Africa, as our tools and methods rise to the challenge of illuminating our deeper origins. But though Africa south of the Sahara is the site of most of our species’ evolutionary history, today we’re adjusting to a revelation not out of the cradle of humanity, but from the sands of the continent’s largely barren northern third.

In 2000, the plot was dead simple. A small group of expansionist East African modern humans left the ancestral continent 50,000 years ago, and exterminated all of our cousins, from the Neanderthals in Europe to assorted other hominin populations (these latter at the time unnamed) in East Asia. By 2010, it became clear that the extermination had not been complete; non-Africans carried Neanderthal heritage from an admixture event early in the out-of-Africa migration timeline, likely in West Asia. More surprisingly, we also harbored Denisovan ancestry, traces of an eastern human species that was discovered solely by genomics. And the surprises kept coming. In 2014, geneticists scanning patterns of relatedness across ancient and modern humans realized there had been a branch of non-Africans who never mixed with Neanderthals, and whose heritage many West Asian populations carry. This heritage was also brought to Europe and South Asia during the agricultural revolution’s demographic expansions starting 10,000 years ago. We termed this population “Basal Eurasians,” for their early split from other non-Africans, over 50,000 years ago.

But, like all other non-Africans, Basal Eurasian genes attest that that group too is a product of the “great bottleneck,” a period of thousands of years when the ancestral breeding population of all humans except sub-Saharan Africans collapsed to the scale of just some 1,000 individuals, give or take. It seems likely that Basal Eurasians 50,000 years ago occupied territory in West Asia or Northeast Africa, so they never had a brush with Neanderthals, like their non-Basal cousins, who pushed north and east into the highlands of Iran and the Caucasus. There, our ancestors absorbed a large population of Neanderthals before sweeping outward to the north, northwest and east, taking their Initial Upper Paleolithic (IUP) toolkit all across Eurasia and onward to Australia. The story then, until 2025 ran that the “out of Africa” humans had developed in isolation, and a branch had eventually gone on to conquer the world. This small out-of-Africa population was one of multiple human groups who branched off from Africans south of the Sahara, specifically from East Africans.

Recently, that story has received a significant plot twist. Early modern humanity was a bigger family than we thought. Over 50,000 years ago, the ancestors of today’s humans were divided not just between sub-Saharan Africans and proto-Eurasians, but into at least three broad categories, because now paleogeneticists have uncovered a lost branch of humanity that long occupied northern Africa, separate and distinct from the sub-Saharan populations to their south, and the Eurasians to their east. These were the Ancient North Africans (ANA), and we now have three genomes where this exotic heritage predominates, their 2025 publication opening a new chapter in the quest to retrace the human journey at a level of granularity unimaginable in 2003 when the site was first excavated.

The unwieldy term sub-Saharan African encompasses all the variegated peoples south of the Sahara desert. A simpler label might be black Africans, but populations like the San foragers and Nama pastoralists of southwest Africa have lighter skin than many tropical Asian peoples. An essential aspect that has set sub-Saharan African population structure apart since the emergence of molecular evolution in the 1980’s is that it remains the most genetically diverse on the planet because its peoples alone were spared the great bottleneck that all other known humans endured (or so we thought). This has preserved such wealths of genetic diversity that we find the Khoisan of the far south are genetically more distinct from West Africans than West Africans are from Eurasians (who are, of course, ultimately descendants of East Africans). The cleavages within Africa are deepest and most ancient in the south, but now new light has been shed on the oft neglected north, which in our day is mostly uninhabitable desert.

An April 2nd, 2025 paper in Nature introduces us to an additional population that also skipped the great bottleneck, and also were not sub-Saharan Africans. The paper is Ancient DNA from the Green Sahara reveals ancestral North African lineage. This finding is not a total shock, a 2018 paper reporting seven 15,000-year-old samples from what is today Morocco found only an imperfect fit between them and contemporary populations in the region. Located in Taforalt cave, these individuals’ genomes bore affinities to Near Eastern populations, but gave perplexing signs of genetic variation that didn’t really fit any of our existing models of human evolution or the representative list of populations then in our datasets. The challenge with the Taforalt samples is that it’s hard to understand admixed populations without a good grasp on the putative source populations in the first place. Imagine trying to solve for x in the equation 2 + x = 6 if you know no numbers beyond 2. It just doesn’t compute.

The Nature paper genotypes three individuals from 7,000 years ago in the deep Sahara, in southwest Libya. These new probands were much more geographically and genetically isolated than Taforalt’s Pleistocene Moroccans, and the Libyan genomes offer much more unambiguous evidence of a previously unknown and distinct ancestral population that must have flourished across North Africa for tens of thousands of years prior. The Taforalt individuals were the first clue that our existing understanding of African population history had serious gaps; prehistory north of the Sahara was more complicated than we had thought. Of the three individuals found in the Takarkori rock shelter in the Green Sahara paper, one furnished truly high-quality DNA yielding over 800,000 markers, and reflects 90% descent from a distinct North African ghost lineage that was neither Eurasian nor sub-Saharan African. Named TKK, this individual’s results make clear and indisputable that a lost race of our species once called the Sahara home.

TKK and her lineage flourishing some 7,000 years ago in what is today deep in the heart of the world’s largest desert, are only comprehensible given the Sahara’s climatic fluctuations over the last few hundred thousand years. When TKK lived, the Sahara was not a desert across most of its range. The huge stretch of territory in Africa’s northern third from the Atlantic to the Red Sea was undergoing what is labeled in the paper the “African Humid Period” (AHP), another term for the more colloquial Green Sahara. As is clear in the map above, where inset panel C depicts the arid present and panel B represents the biomes of the wet period, between the end of the Ice Age 11,500 years ago, and 3000 BC, for these crucial millennia, most of the Sahara was not desert. It was instead a mix of woodlands, savanna and scrubland, with only a modest precursor of today’s vast desert along its eastern edge. Humid periods have punctuated the Sahara’s history since it emerged as a climatic and geological entity nearly 10 million years ago. The Eemian interglacial, dating to 110,000 years ago, may have been even more humid than the latest Green Sahara, thereby propelling human populations from the African core northward toward the Mediterranean.

For 2025, I’m trying out a new occasional feature for paying subscribers, the Unsupervised Learning Journal Club: I’ll offer a brisk review and consideration of an interesting paper in human population genomics.

In the spirit of a conventional journal club, at the end of each post, interested subscribers can vote on next papers to review. I’m open both to covering the latest papers/preprints and reflecting back on seminal publications from across these first decades of the genomic era.

If your lab has work we might like or you otherwise want to suggest a paper for me to cover, feel free to respond to this email or comment on this post.

The first three editions are here:.

Wealth, war and worse: plague’s ubiquity across millennia of human conquest

Where Queens Ruled: ancient DNA confirms legendary Matrilineal Celts were no exception

Eternally Illyrian: How Albanians resisted Rome and outlasted a Slavic onslaught

Free subscribers can get a sense of the format from my ungated coverage of two favorite 2024 papers:

The other man: Neanderthal findings test our power of imagination

We were selected: tracing what humans were made for

Today’s paper, the overwhelming preference of last edition’s survey voters is: A structured coalescent model reveals deep ancestral structure shared by all modern humans. It comes out of Aylwyn Scally and Richard Durbin’s groups at Cambridge, and appeared in Nature on 18 March 2025. The first author is Trevor Cousins.

Today’s paper, the overwhelming preference of last edition’s survey voters is: A structured coalescent model reveals deep ancestral structure shared by all modern humans. It comes out of Aylwyn Scally and Richard Durbin’s groups at Cambridge, and appeared in Nature on 18 March 2025. The first author is Trevor Cousins.

One of the most powerful metaphors for understanding evolutionary processes is the phylogenetic tree; a representation of the branching patterns that depict how life has diverged and diversified into many disparate lineages, from a single common ancestor four billion years ago. The idea has medieval origins, and was at the heart of Charles Darwin's argument in On the Origin of Species. Descent with modification along its many branches was how the tree unfurled across evolutionary time.

But a tree metaphor always had limitations, not least because pedigrees have a certain propensity to turn in on themselves; wayward branches fuse together to spawn something new. In Darwin’s age, hybridization across species in the animal world was seen as a relatively rare occurrence, an aberration most famously illustrated with sterile mules, evolutionary dead ends. Yet even Darwin and his contemporaries, avid gardeners, knew that in plants hybridization was not exceptional at all; they were well aware that botany’s phylogenetic tree was more of a tangled bush.

Today, modern genomics and ancient DNA are exposing the reality that even in mammals, tree structures are crisscrossed with connections, a wild lattice of hybridizations. Recently, paleogenomics has raised the possibility that the dire wolf George R. R. Martin’s Game of Thrones made famous, might have been a mixture of two canid lineages as long separated as humans have been from chimpanzees: some six million years. And indeed, that separation between chimpanzee and human lineages itself may not have been as simple as a single divergence date would indicate. Before proto-humans and proto-chimpanzees finally parted ways for all time, they underwent an intense bout of hybridization.

And what was true at the root of the entire bipedal lineage now appears indisputably true later on in human evolution, as we trace the tree structure out along its network of branches. In 2010, paleogeneticists discovered that modern humans still carry both Neanderthal and Denisovan heritage, the latter an ancient East Asian hominin discovered purely through a DNA sequence. Other researchers, using more indirect means leveraging mass volumes of human genomes and the new power to test extremely complex models in a computational framework, have now inferred similar events of archaic admixture into the proto-modern lineage within Africa. What we had sketched as a gracefully branching human family tree turns out to be a messy tangle, with branches twisting together to form new lineages, before again unbundling outward via demographic expansions and replacements that radiated populations out across the entire Old World.

And yet the proportion of Neanderthal DNA in modern populations is about 2% at most, and Denisovan input rises marginally higher solely among New Guineans (otherwise appearing in only trace amounts across most of South, East and Southeast Asia). Regardless of race, most of humanity’s heritage goes back to one population resident in Africa 100,000 years ago. Outside of Africa, all extant populations’ lineages coalesce back to 60,000 years ago, to a small tribe of as few as 1,000 who survived a sharp and prolonged bottleneck before mixing with some Neanderthals and sweeping outward across Eurasia, and into Oceania. But what if these ancestral Africans, from which the vast preponderance of our ancestry comes, were themselves the product of a recent admixture event, and one that left a much more appreciable legacy than the later marginal genetic absorption of mere traces of archaic Neanderthal and Denisovan lineages? Instead of an assimilation, imagine an amalgamation. Perhaps the first modern humans were freaks, a race of hybrid monsters born of what we would today perceive to be abominable unions?

That is what today’s Nature paper, A structured coalescent model reveals deep ancestral structure shared by all modern humans, with first author Trevor Cousins, argues. The paper concludes that our modern human lineage arose out of an admixture event dating to some 300,000 years ago between two hominin species who had split 1.5 million years ago. One lineage, “population A” contributed about 80% of our species’ ancestry. It emerged out of the same lineage that begat Neanderthals and Denisovans. The other, “population B” contributed the remaining 20% of our ancestry. It was only after the mixing between these two human species, evolutionarily as distant as wolves are from coyotes, that the proto-modern human lineage arose, diversified, and went on to conquer Africa, and then the world. So how did population B alter us? The evidence suggests that population B contributed a major software upgrade to be run on hardware mostly from population A, B’s brains hitched to A’s brawn.

Perhaps surprisingly at first blush, Cousins et al. arrived at the audacious conclusion without leaning on ancient DNA. Upon reflection, this stands to reason, on these time scales, what passes for ancient DNA to us is scarcely any closer to 300,000 years ago than we ourselves are. The oldest African DNA we’ve found dates to about 15,000 years ago, so paleogenomics sheds very little light on our species' deeper African history. To get around this, Cousins et al. focused on deeper, more subtle analyses of the vast troves of whole genome sequences from contemporary populations now available to researchers. If 2025’s mountain of data is their gold mine, then a new method, cobraa, is the game-changing extractive device. The researchers extended a method devised in 2011, PSMC (Pairwise Sequentially Markovian Coalescent). Over the last decade and a half, PSMC has equipped researchers to take whole-genome-sequence data, and reconstruct an individual’s demographic past, and more generally, the population sizes and dynamics of their ancestral lineage as reflected in their pedigree fanning out into the past.

The 21st-century computational heavy lifting of PSMC, and now cobraa, rests on the foundation of a spare, elegant 20th-century population-genetic concept, the coalescent. Pioneered in the early 1980’s, the coalescent emerged in a time when we finally had a small, but still paltry amount of DNA data, plus some basic computational resources, but nothing truly powerful yet. In theory, with computers, population geneticists could simulate evolutionary dynamics like selection and drift, observing their impact on allele frequencies forward in time, running experiments in silico. But practically, the branching possibilities were too many for that era’s primitive computational power (the top-of-the-line IBM mainframe computers then had 16 or 32 MB of RAM; today 8 GB of RAM is a modest specification for a desktop computer, so 250-500 times more memory than a top-of-the-line mainframe 45 years ago). Faced with these limitations, some geneticists realized that reversing the arrow of time would make their analyses much more tractable; rather than project numerous evolutionary possibilities forward, they could start with the genetic endpoints, and converge back along the pedigree to a common ancestor. In a coalescent framework, the genealogy has fewer and fewer nodes every generation back, meaning that the analyses become less resource-intensive as they proceed.

The original coalescent focused on a single locus and its genealogy; this was the age of genetics, not genomics. In an effective population of size Ne (Ne being the number of individuals who pass genes to the next generation) the expected number of generations for two distinct gene copies to coalesce is 2Ne generations. By looking at the time depths of the branch lengths to coalescence within a genealogy, you can estimate the Ne that would produce these patterns. The genetic variation is the data you load into the coalescent model, and ancestral population sizes are a parameter whose values you allow the model to tinker with in an effort to reproduce the data’s patterns. The genomic framework extends simple coalescent logic by looking at patterns across thousands of genes; instead of a single genealogy we now test thousands of coalescents. Rather than a single locus, as in the pre-genomic era, PSMC and cobraa scan the entire genome, inching along to examine sliding windows of 100 sequential DNA bases at a time, assessing whether a genomic region within a single individual is heterozygous (two variations of the gene copies) or homozygous (duplicate of a single variant). By comparing an individual’s two complete copies of the human genome, cobraa can estimate patterns of intra-locus coalescence. Since the two gene copies are contributed from two parents, intra-locus coalescence from a single individual reflects distinct parental histories. A model, which includes population history and mutation rates as parameters, can estimate variables like effective population size by iteratively fitting the data better and better.

As an upgrade that goes beyond PSMC’s computational genomic model, one of cobraa’s major extensions comes by considering possible scenarios of structure within the ancestral population. While PSMC was limited to assuming a single population, thereby smoothing out complex histories into a rough average summary, cobraa is powerful enough to contemplate scenarios where different populations experience different histories, before eventually fusing into the single hybrid populations familiar to us. Cousins et al. validated cobraa by testing it against simulated histories (whose ground truth they knew beforehand, having designed the simulations). Then they applied the method to a small sample of human genomes representing numerous disparate populations. Cousins et al. looked at 26 human populations, sampling a single individual from each group in the 1000 Genomes Project, and so constructed 26 population histories (they replicated the finding by looking at similar populations in the Human Genome Diversity Project).

The figure that opens this piece illustrates their model of best fit. Back to about 300,000 years ago, their results align with earlier publications. But further back in time, novel inferences are generated, because a structured ancestral set of populations turns out to better fit the data than a single randomly mating horde, the best PSMC could hope to offer. Here, the fusion of two populations, at a 4:1 ratio, proves a better fit than a balanced and reciprocal admixture. Population B, at a 20% share, can be thought of as mixing into the background of population A, at 80%. But compared to Neanderthal contribution into out-of-Africa humans, this input is more substantial by an order of magnitude. Populations A and B had diverged about 1.2 million years prior, about twice the distance separating our early out-of-Africa population from Neanderthals and Denisovans. Population A, the majority ancestral contributor, is a sister lineage to Neanderthals and Denisovans.

Additionally, population A went through an extreme bottleneck at its founding; all the estimates of our lineage’s divergence from these archaic Eurasian hominins actually derive solely from our population-A heritage. The 20% of our ancestry from population B, is far more divergent. Given this new insight, we may need to revisit our estimated dates of the splits between Neanderthals and modern humans (because we probably actually diverged significantly more recently than assumed). Also, because it is the minority input, population B’s ancestral contribution appears to have been subject to purifying selection; on average, its contributions are clearer and more copiously represented the further away the DNA segment lies from the 1.5% of the genome that codes for proteins. But, there are telling exceptions. In some classes of genes, population B’s contributions are disproportionate. Most importantly, those implicated in the brain.

Today Razib talks to Laura Spinney, Paris-based British author of the forthcoming Proto: How One Ancient Language Went Global. A science journalist, translator and author of both fiction and non-fiction, she has written for Nature, National Geographic, The Economist, New Scientist, and The Guardian. Spinney is the author of two novels, Doctor and The Quick, and a collection of oral history in French from Lausanne entitled Rue Centrale. In 2017, she published Pale Rider, an account of the 1918 flu pandemic. She also translated Swiss writer Charles-Ferdinand Ramuz's novel Derborence into English. Spinney graduated with a Bachelor of Science degree in Natural Sciences from Durham University and did a journalism residency at Berlin’s Planck Institute.

First, Razib asks Spinney how difficult it was to integrate archaeology, linguistics and paleogenetics into her narrative in Proto, which traces the rise and proliferation of Indo-European languages from its ancestral proto-Indo-European. She talks about why this was the time to write a book like this for a general audience, as paleogenetics has revolutionized our understanding of recent prehistory, and in particular the questions around the origin of the Indo-Europeans. Razib and Spinney talk about various scenarios that have been bandied about for decades, for example, the arguments between linguistics and archaeologists whether proto-Indo-European was from the steppe or had an Anatolian homeland, and the exact relationship of the Hittites and their language to other Indo-European branches. They also delve into how genetics has helped shed light on deeper connections between some branches, like Balto-Slavic and Indo-Iranian, or Greek and Armenian. Spinney also addresses how writing a book like Proto involves placing fields like historical linguistics and archaeology with charged political associations in their proper historical context

Subscribe now

Give a gift subscription

Share

Subscribe now

Give a gift subscription

Share

This generation of human paleogenomics has been delivering a steady stream of grand revelations about our species’ origins and evolution, from the reality (and indeed ubiquity) of Neanderthal ancestry that hitchhiked into the present aboard our own genomes, to the discovery of a whole new human species, the Denisovans. Lately, geneticists have moved on to open questions of more recent vintage, shifting from the domains of paleoanthropology and archaeology to recorded history’s very threshold. With the vast volume of ancient human DNA samples now accumulating, massive swaths of Eurasian humanity can be examined along a transect tracing genetic change across tens of thousands of years. Like a waterlogged landscape after an epochal flood finally recedes, our age’s surfeit of data is exposing surprising phenomena that had long lain indiscernible or even wholly unimagined.

Unexpected finds in paleogenomics are driving us to consider startling questions we didn’t even know we needed to ask. Dynamics like a demographic turnover here, or the migration of an unknown people there, force us to reshuffle known archaeological results and theories into new configurations. A 2023 preprint out of David Reich’s lab seems to have come close to pinpointing the origin of the Baltic’s Finnic peoples, while a 2025 preprint from his rival Eske Willerslev’s group may have uncovered the proto-Germanic tribes’ ancestral homeland in the most unexpected locale. Because whereas the Finnic tribes’ destination was the eastern Baltic, that same zone now appears to have been the proto-Germanics’ and their ancestors’ long mysterious origination point. In a general sense, the Finns’ and Estonians’, and their proto-Uralic ancestors’ more than 1,000-year journey from one end of Eurasia to the other is little surprise, just a refinement whose precise details linguists, archaeologists and now geneticists had long quested to pin down. But the very suggestion that what became Finland and Estonia were meanwhile the mysterious homeland of the earliest proto-Germanic-speaking people comes straight out of left field. Disciplines like archaeology have barely had time to come to grips with the ramifications of this possibility, with early scholarly response thus far amounting to little more than stunned silence.

It’s never been any secret that the eastern Baltic’s Finnic-speaking peoples are something of a different stripe of European; not least of their differences is their non-Indo-European languages, which are likely of Asiatic provenance. Nevermind their predominantly Nordic features, in the early 20th century, these linguistic affinities meant racially fixated European anthropologists regarded them with some suspicion; atlases of the period would often classify Finns as a partially Asian ethnicity. On linguistic grounds, this was not without foundation; save for Hungarian and Estonian, Finnish was the westernmost branch of the Uralic language family, whose domains blanketed Nordic Europe’s northernmost reaches, and then stretched over and across the Urals toward central Siberia’s Arctic shores, where scattered bands of hunter-gatherer Samoyeds spoke the family’s easternmost dialects. The central question, though, for early scholars was about direction: did the Asian Samoyeds pick up their language from European Finns who had strayed east, or did the Finns’ language actually descend from that of Siberian interlopers? Were the Finnic peoples Europeanized Asiatics, or were the Samoyed tribes Asianized Europeans who had intruded into Siberia and been swallowed into the peoples of our vastest continent?

As this young century dawned and the Human Genome Project finished its first draft, geneticists began developing new methods to assess and classify populations. First up, phylogenetic techniques developed earlier on small datasets of medical and evolutionary relevance were redeployed to generate family trees of uniparental lineages utilizing the Y chromosome and mitochondrial DNA (mtDNA). Unlike the vast bulk of the genome, these segments of DNA pass purely from father to son (the Y) and mother to daughter (the mtDNA, because while sons indeed inherit their mother’s mtDNA, they pass it no further), yielding tree-shaped genealogies, with branches bifurcating and diversifying over time from common ancestors. Because populations variably undergo periods of rapid growth or collapse, and geographic separation results in distinct breeding populations, the Y and mtDNA patterns in haplotypes track demographic history. These unique genetic signatures can be combined into broader related clusters or haplogroups, to define population-wide lineages.

Using these methods, combining the new genomics and older phylogenetic frameworks, geneticists detected patterns that showed all non-African humans today were broadly related to each other, and that haplogroups tended to exhibit geographic patterns that map satisfyingly onto what anthropology and archaeology had taught. Interestingly, while Finnic mtDNA did not differ from that of their Scandinavian neighbors to the west, Finnic Y chromosomes were markedly distinct. About 60% of Finnish men carried haplogroup N, as compared to 7% of Swedish males, 3% of Norwegian men and 1% of male Danes (while N is basically wholly absent from Western and Southern Europe). Interestingly, haplogroup N, and Finland’s particular sublineage, is also found in populations to the east, from European Russia all the way out to Siberia’s Pacific coast. In Russia’s frigid far north, half of men carry this lineage, while among the Finnic-speaking ethnicities of the Russian Urals, the Udmurts and Mari, its frequency hovers around 30-50%. Among the Samoyed tribes, over 50% of men carry N. Finally, among the northeasternmost Turkic-speaking people in the world: the Yakuts of eastern Siberia, 80-95% of men are N. Though haplogroup N’s ambit is more extensive than the map of Uralic languages today, save for Hungarians, all Uralic-speaking populations harbor N in high numbers. If you are a man who carries N, you may not be Uralic, but if you are a (non-Hungarian) Uralic male, odds are good that you carry N.

But this still didn’t settle the question of directionality, west to east, or east to west? The highest ratios of N were at the two geographic antipodes of its distribution, in Finland and Yakutia, though curiously the Yakuts are Turkic-speaking and so Altaic rather than Uralic. Inspecting the genetic diversity within Uralic branches of N found more variation in the east than the west, and as a rule, more variation usually correlates with a lineage’s deeper history in a region, but correlation does not always tell the whole story. Sometimes greater or lesser genetic diversity just reflects peculiarities of population history, like admixture between ethnicities long separated, as in the New World.

After 2005, genetic tools to extend beyond the narrow purview of sex-based haplogroups began to come online. Geneticists could analyze whole genomes from numerous populations, seeking patterns and informative clusterings among datasets of thousands of individuals across thousands of markers. A 2018 paper focused on Uralic populations found that a particular genetic cluster within the admixture plot often correlated with populations who speak Uralic languages. But overall, this genetic cluster’s highest frequencies appeared among Central Siberians and Arctic populations, in particular Samoyed tribes like the Nenets. Meanwhile, among Europeans, this cluster appeared almost exclusively in Finnic-speakers (or those admixed with Finnic-speakers, like Swedes, Latvians and Lithuanians) and Russians (many northern Russians between the Baltic and the Urals have Finnic ethnic ancestry). This component happens to be maximized among the Samoyed tribes, as is the exclusive contributor to the ~5% East-Asian ancestry observed among Finns today. Though the Finnic-speaking people of the eastern Baltic are overwhelmingly European, they do all exhibit this common Siberian ancestry, whereas Samoyed people from Central Siberian seem to entirely lack European ancestry (at least save for very recent scattered Russian admixture). This makes it much more likely that the Finns derive ancestry from Siberia, albeit diluted, than that the Samoyeds descend from a Finnic people with Baltic origins.

Because northern Eurasian conditions are highly conducive to ancient DNA preservation, paleogeneticists have been able to probe population-genetic diversity across numerous archaeological periods, going far beyond comparisons of modern populations. Early ancient DNA samples from the eastern Baltic show a range of peoples, but the arrival of Siberian heritage and haplogroup N seems to postdate 1000 BC, at least in the coastal regions modern Finns and Estonians inhabit. The Reich lab’s 2023 preprint, Postglacial genomes from foragers across Northern Eurasia reveal prehistoric mobility associated with the spread of the Uralic and Yeniseian languages, marshals a massive number of samples across five millennia to definitively conclude when and where the Uralic people originated.

Today, Razib talks about a new paper, A structured coalescent model reveals deep ancestral structure shared by all modern humans:

Understanding the history of admixture events and population size changes leading to modern humans is central to human evolutionary genetics. Here we introduce a coalescence-based hidden Markov model, cobraa, that explicitly represents an ancestral population split and rejoin, and demonstrate its application on simulated and real data across multiple species. Using cobraa, we present evidence for an extended period of structure in the history of all modern humans, in which two ancestral populations that diverged ~1.5 million years ago came together in an admixture event ~300 thousand years ago, in a ratio of ~80:20%. Immediately after their divergence, we detect a strong bottleneck in the major ancestral population. We inferred regions of the present-day genome derived from each ancestral population, finding that material from the minority correlates strongly with distance to coding sequence, suggesting it was deleterious against the majority background. Moreover, we found a strong correlation between regions of majority ancestry and human–Neanderthal or human–Denisovan divergence, suggesting the majority population was also ancestral to those archaic humans.

Subscribe now

Give a gift subscription

Share

On this episode of the podcast Razib talks to John Sailer. Sailer is currently the director of higher education policy and a senior fellow at the Manhattan Institute. He covers issues of academic freedom, free speech, and ideological capture in higher education. Sailer has written for the Wall Street Journal, the Free Press and Tablet Magazine. Sailer holds a master’s degree in philosophy and education from Columbia University, and a bachelor’s degree in politics, philosophy, and economics from The King’s College. Prior to joining the Manhattan Institute, he was a senior fellow at the National Association of Scholars.

Following on last week’s podcast with Jacob Shell, Razib continues to discuss the rise and fall of woke politics in academia, and the current backlash exploding out of the Trump administration. Sailer discusses his previous work back to 2020 showing how blatant universities became in their discriminatory policies against white males in particular, and how easy it was to demonstrate this dynamic with even the most minimal level of due diligence like freedom of information requests. They also discuss the reality that universities are attempting to adjust to a new landscape with the administration pressuring them to revoke DEI policies, while many faculty are urging that they instead dig in their heels. Higher education is adapting, but Sailer argues that since fundamental values have not changed, some evasion is to be expected.

Subscribe now

Give a gift subscription

Share

On this episode of Unsupervised Learning Razib talks to Jacob Shell. Shell is a professor of geography at Temple University and author of Transportation and Revolt: Pigeons, Mules, Canals, and the Giants of the Monsoon Forest: Living and Working with Elephants. Educated at Columbia and Syracuse universities, Shell is active on social media, where he comments extensively on the politicization of the academy.

The conversation begins with Shell’s piece in Compact Magazine, To Save Academia, Hire Conservatives. The more than 3,000-word essay argues that academia must diversify ideologically to save itself, but also engage in a wider range of scholarship. Shell points out that US academia has become an ideological monoculture, with an overwhelming dominance of left-leaning faculty, especially at elite institutions. This imbalance, driven by extreme partisan ratios in fields like anthropology, leaves universities politically vulnerable and out of step with the broader public. He challenges the common view that this trend is due to self-selection, or the “pipeline problem,” suggesting instead that informal screening mechanisms discourage or exclude conservative scholars. Shell also argues that the grant system encourages conformity and limits academic freedom. More audaciously, he argues that some academics should be singled out by their peers, whether through their institution or professional organizations,t when they engage in politically motivated misrepresentation of their scholarship. Ultimately, Shell insists that academia’s unique role in public life is to observe and understand the world, not to risk co-option as an arm of any political movement.

Give a gift subscription

Subscribe now

Share