David Reich — How one small tribe conquered the world 70,000 years ago

David Reich describes how DNA from present-day people, Neanderthals and Denisovans has changed the study of human history. The Denisovans were identified from DNA recovered from a finger bone found in a Siberian cave. Over the past decade-plus, researchers have combined these samples with traces of archaic DNA in living populations to reconstruct relationships among modern humans and their closest relatives. Reich stresses that the current picture was assembled incrementally as new genomes and mixture events were added.

The standard model places the split between modern humans and the lineage ancestral to Neanderthals and Denisovans about 500,000–750,000 years ago. But mitochondrial DNA and the Y chromosome tell a different story: their shared Neanderthal–modern human ancestors may be only 300,000–400,000 years ago, while Denisovan lineages are more distant. Reich says 3–8% of Neanderthal DNA may have come from a later gene-flow event with modern humans, yet the probability that both the mitochondrial and Y-chromosome patterns arose from that event by chance is low. Explanations invoking separate episodes of natural selection are possible, but he finds them surprising.

Reich and others are considering models with far more modern-human ancestry in Neanderthals—perhaps 30%, 50% or even 70%. In such models, the usual distinction between a Neanderthal–Denisovan sister group and modern humans becomes less clear. He also points to evidence that modern African populations are deeply substructured, with ancestry from multiple lineages. Reich says the standard model has accumulated patches as each new genome and admixture event was discovered; he is agnostic about which alternative is right and argues that the full picture may need to be reconsidered from the ground up.

The location and geography of ancient gene flow remain uncertain. Reich says interaction may have occurred in the Near East or western Eurasia, a region that could function as an ecological corridor between Africa and Eurasia as climates shifted. He argues that, over hundreds of thousands or millions of years, the continents were not necessarily barriers to movement for human populations. At the same time, he cautions that researchers do not know where the ancestral populations lived or how often they exchanged genes.

Dwarkesh highlights how surprising it is that accounts of human origins may need revision, given how central the familiar story is to education. The exchange also connects anthropology to AI forecasting: if larger brains helped humans dominate other human species, perhaps scaling AI models could likewise produce powerful systems. Reich does not endorse that inference here; he emphasizes how much remains unknown about the specific population history behind human expansion.

Reich describes ancient DNA from hunter-gatherer populations in eastern and central Africa, including samples around 15,000 years old. Many groups appear to have lived in small populations with low genetic diversity and limited exchange. He suggests that isolated communities could lose cultural knowledge through disasters or the death of key elders, while groups that reached a critical mass could accumulate tools, strategies and shared information. Across many separate groups, this process of knowledge loss and recovery may have produced occasional breakthroughs.