One of the most dramatic claims about our past is that humanity nearly went extinct around 70’000 years ago. Some genetic studies have been interpreted to suggest that all modern humans descend from as few as 1’280 surviving individuals, a number that sounds more like the population of a small town than a species spread across continents. A slightly less extreme version of this idea sets the number at around 10’000 “breeding” individuals.
The story is compelling: a volcanic catastrophe, a genetic bottleneck, and a narrow escape from extinction. But is it true? When faced with uncertain evidence, we need a way to evaluate which explanations are more likely. This is where Bayesian thinking is useful: start with what we already know about the world, then update our understanding as new evidence appears. Applying this logic to the bottleneck idea, we will weigh the evidence for and against near-extinction and see which scenario fits the data best.
Before we weigh the evidence, we can frame the question as two simple hypotheses.
The Two Hypotheses
70’000 years ago:
- There was a bottleneck, a near extinction of the human race.
- There was no bottleneck and the population never dropped to such low numbers.
What evidence can we use to determine what the world looked like more than 2’000 generations ago? Of course there is no written record. But we do have several ways to investigate what happened, and just as importantly, what did not happen.
The Catastrophe
The most often cited explanation for the bottleneck idea is the eruption of the Toba supervolcano about 74’000 years ago. Toba is located on the island of Sumatra in present-day Indonesia. It was one of the most powerful volcanic events in the last two million years, ejecting enormous amounts of ash and aerosols into the atmosphere. Some researchers have suggested that this event caused a global volcanic winter, leading to a collapse of ecosystems and a severe decline in human populations.
The eruption itself is well-documented in the geological record. Thick ash deposits several meters deep are found across South and Southeast Asia, especially in India. Trace deposits have been detected as far away as the Arabian Peninsula and East Africa, but these layers are thin and localized. In Europe and the Americas, no direct ash layers from Toba have been identified. Ice core data from Greenland and Antarctica show signs of some cooling around the time of the eruption, but the effects appear short-lived rather than catastrophic.
The ash record alone tells us that Toba’s impact was not global in the way a true extinction-level event would be. The deposits are massive and destructive close to the volcano, stretching across South and Southeast Asia with layers several meters thick in parts of India. Beyond this zone, however, the ash thins rapidly. In the Middle East and East Africa, the traces are light and patchy, and in Europe and the Americas no ash deposits from Toba have been identified at all. This already limits the scale of its direct physical impact. If the eruption had truly devastated the entire planet, we would expect a continuous ash layer, or at least detectable volcanic markers, across all continents.
The argument for a global effect relies on the idea that volcanic aerosols from Toba caused a dramatic cooling period, a so-called volcanic winter. Ice cores from Greenland and Antarctica do show signs of a brief temperature drop around this time, but the signal is modest compared to known major climatic events. It suggests a few years of cooling, not decades of collapse. This difference matters: a short, temporary temperature dip may stress ecosystems locally but is unlikely to wipe out an entire species spread across Africa, Asia, and the Middle East. To accept the bottleneck story as it is often told, we would need evidence of a long, global ecological crash following Toba, and we simply do not see that evidence.
Africa is especially revealing. If Toba had caused a near-extinction event, Africa should show signs of sharp population decline, cultural collapse, or abandonment of key regions. Yet archaeological records in East and Southern Africa tell a different story. Stone tool traditions, settlement patterns, and evidence of daily life continue across this period without clear interruption. Humans not only survived there but continued to develop technologies and maintain cultural continuity. This strongly suggests that the impact of Toba, while severe in its immediate surroundings, did not create a global bottleneck.
Taken together, the geological and archaeological evidence weakens the idea that Toba caused a global near-extinction event. The eruption was massive and destructive in parts of Asia, but its effects were neither uniform nor lasting enough to explain the genetic patterns we see today. This leads us to the second piece of evidence often cited for a bottleneck: the relatively low genetic diversity in modern humans compared to other primates. What does this genetic signal actually tell us, and what does it not tell us? From a Bayesian perspective, the lack of global evidence for Toba’s devastation already lowers the probability of a severe bottleneck.
The Genetics
One of the strongest arguments for a human bottleneck is the observation that modern humans have relatively low genetic diversity compared to our closest relatives. Two random chimpanzees, for example, tend to be more genetically different from each other than two random humans. Gorillas and orangutans show even greater variation within their species. At first glance, this seems to suggest that something unusual happened in our evolutionary past, reducing the variety of genetic material in our species.
But genetic diversity is shaped by more than just population size. It is also influenced by how populations are structured, how much they interbreed, and which genetic lineages survive over time. A lower genetic diversity does not automatically mean that our ancestors were reduced to a handful of individuals. It simply tells us that fewer genetic lineages have made it to the present.
Geneticists studying modern humans have found that our current genetic material can be traced back to about 1’280 ancestral sources around 70’000 years ago. This is often presented as evidence of a bottleneck, but what does this number actually tell us? Does it mean there were just over a thousand people alive on Earth at the time? Does it mean there were perhaps around 10’000 people, but only a small fraction of their children survived to pass on their genes? Or is it possible that there were many more people, but most of their lineages simply left no trace in our modern DNA?
How likely is it that such a severe bottleneck really occurred? For the genetic data to represent a true population count of around 1’000 individuals, we would have to assume a catastrophe on a global scale. Something like the Toba eruption, or a plague so devastating that it wiped out 99% of all humans, leaving just a handful of scattered survivors.
But would we expect humanity to recover from such a dramatic collapse? With so few people, the risk of inbreeding would be enormous. Populations this small tend to accumulate harmful mutations and lose resilience within a few generations. For the bottleneck theory to hold, not only would all humans have had to survive the same disaster, they would also have had to rebuild from a genetic base that was alarmingly fragile.
Over thousands of generations, genetic lineages disappear in much the same way that family names do. Imagine a village where some families have many children and others have none. Over time, certain surnames vanish simply because there is no one left to carry them forward. The same happens with genes. Even in a large population, many genetic lines die out naturally because their descendants do not survive or because their traits are diluted and eventually lost.
Not all genetic lines survive by chance alone. Some traits are more likely to spread because they provide an advantage. A gene that improves disease resistance, for example, can become dominant within a few generations, especially in small, interconnected groups. Other traits may vanish simply because they offer no advantage, or because they are linked to harmful effects that reduce survival or reproduction.
Migration and population mixing also play a role. When a small group moves into a new area and grows quickly, its genetic traits can overshadow those of other groups, even if the original population was much larger. Over thousands of years, this process can erase the genetic footprint of many other lineages. In today’s DNA, this can look as though only a small number of ancestors ever existed, even if the real population was far larger.
A clear example of how traits persist or disappear is human skin color. Populations living closer to the equator have, over many generations, developed darker skin. This is not because a child born in Africa becomes darker due to the sun, but because the genetic trait for dark skin provides a survival advantage in regions with intense sunlight. It protects against harmful UV radiation and helps prevent conditions like folate deficiency.
In contrast, populations living in northern latitudes developed lighter skin. In low-light environments, lighter skin allows for more efficient vitamin D production. Over time, these traits became dominant in their respective environments, not because other skin tones were impossible, but because they were less suited to local conditions.
This process of natural selection shows why certain genetic lines survive and others fade, even in large and healthy populations. The genetic narrowing we see in humans today could partly reflect many such advantages being favored over time, rather than a single catastrophic event wiping out diversity.
Another explanation is that some traits are simply dominant, meaning they overshadow other variants even when both are present. Blue eyes, for instance, are recessive. A child must inherit the blue-eye gene from both parents to express it. If one parent has brown eyes, which is a dominant trait, the child will most likely have brown eyes, even though the blue-eye gene may still be present in the background. Over many generations, recessive traits can become rare or even disappear if the dominant variants are more common in the population.
Red hair works in a similar way. It is not less fit or less beneficial, but because the gene for red hair is recessive, it is easily masked by darker hair genes. As populations mix, such traits can fade without any selective pressure, simply because the dominant traits are expressed more frequently.
This illustrates another way genetic lines can appear to narrow over time without any catastrophic event. Even in a large population, the visible or measurable traits can shift toward the dominant ones.
The Problem of a Tiny Starting Population
Even if we accept the bottleneck claim at face value, for example 10'000 survivors, we run into another problem: maintaining genetic health over 70'000 years. A population of 10'000 is not one large family; it would be split into smaller groups, each living in different regions with limited contact. For genetic diversity to remain viable, these groups would need to interbreed regularly. Without this exchange, harmful mutations would accumulate, reducing fertility and survival rates.
Populations this small rarely remain healthy. In the animal kingdom, species with fewer than 1'000 individuals almost always suffer from inbreeding, low fertility, and a higher rate of genetic disorders. Even larger populations, such as cheetahs and some leopard subspecies, show clear genetic weaknesses due to historic population bottlenecks. Cheetahs, for example, have such low genetic diversity that many individuals are almost like clones of each other. We see in modern species that a population this small is almost always on the brink of collapse, which makes the claim of 10'000 humans surviving for tens of thousands of years even harder to accept.
If humans had truly fallen to 10'000 individuals, scattered across continents, we would expect similarly severe genetic uniformity and widespread signs of inbreeding. While humans have low genetic diversity compared to chimpanzees, we do not show the extreme genetic fragility that such a tiny founding population would produce.
Modern conservation biology tells us that populations under 10'000 individuals are considered vulnerable to genetic decline over just a few centuries. Stretching that across 70'000 years is not realistic. This is why species with populations this small, such as endangered animals, require careful management to avoid extinction. If humans had truly been reduced to only 10'000 individuals, we would expect far stronger evidence of inbreeding effects than what we see in our genome today.
Carrying Capacity
When we consider how much land was available 70’000 years ago, a very small global population becomes highly unlikely. Europe was mostly covered by ice or tundra, making it marginal for human survival. Asia may have experienced local devastation from the Toba eruption. But Africa, the main home of Homo sapiens at the time, was largely untouched by these events. Its savannas, forests, and coasts could have supported a substantial population of hunter-gatherers.
Modern studies of hunter-gatherer groups show that even in less productive environments, a density of 0.1 person per square kilometer is sustainable. Modern hunter-gatherer studies suggest this density based on resource availability while fertile or coastal regions can support significantly higher numbers. If we consider only Africa’s habitable zones, we can conservatively estimate a population between 500’000 and 2 million individuals. This is before adding any other groups that may have lived in southern Asia or the Middle East.
This carrying capacity alone makes the idea of a global population shrinking to fewer than 10,000 people extremely unlikely.
Statistics
Another way to look at the question is by tracing population growth backward. The global population reached 8 billion only recently, in 2022. It was 4 billion in 1974 and 2 billion just before the Great Depression. We have reliable census data and population counts back to around 1800, when the population was close to 1 billion. Before 1800, our estimates rely on a mix of tax records, early censuses, and historical documents that survived through time. These give us a good picture of major population centers and the scale of human presence across continents.
For example, around the start of our calendar, in the year 0, the global population is estimated to have been around 200 million, spread across the Roman Empire, China, India, and smaller states and tribal regions. By the year 1500, as agriculture and trade networks expanded, this number had grown to about 450 million. These are not exact figures, but they are consistent across independent historical studies.
When we place this historical arc alongside what we know about population growth, we see a pattern of slow, steady increase that accelerates with technological and social developments. If we follow this trend back into prehistory, before the rise of agriculture, the numbers shrink but they do not vanish. Even using conservative assumptions about growth rates, this backward projection suggests that around 70’000 BCE, the global population would have been close to 1 million. While these numbers become less precise in deep time, they are consistent with archaeological evidence and ecological limits. This estimate aligns with what we know from carrying capacity and archaeological evidence: humanity was never abundant, but we were far from the brink of extinction.
Conclusion
What does all this tell us? Do we have any credible evidence that humanity was truly on the brink of extinction 70’000 years ago? The answer, while less dramatic, is that we do not. The bottleneck theory was proposed as a possible explanation for the lack of genetic variety in modern humans, but it lacks direct evidence from archaeology, population modeling, or the ecological realities of the time. When we weigh these arguments as a Bayesian would, the prior probability of a population in the hundreds of thousands is far higher than that of just a few thousand survivors. Each piece of evidence, archaeological continuity, land capacity, and population modeling updates that probability further against the bottleneck claim.
A much more likely scenario is that the global population was around 1 million, living in scattered groups across Africa and parts of Asia. The reduced genetic variety we see today can be explained by a combination of factors: dominant traits that overshadow others, some degree of inbreeding within small local groups, and the random loss of lineages over thousands of generations, sometimes because a family line simply ends, and sometimes because a camp sets up under a rockslide or another stroke of bad luck.
This picture is less sensational than the idea of 1’000 survivors clinging to life, but it is far more consistent with the evidence we have.
The bottleneck hypothesis is dramatic, but when we weigh it against the evidence, geological, genetic, ecological, and statistical, its probability fades. The more consistent explanation is that humanity never came close to extinction, but that our genetic tree has been shaped by selection, chance, and the slow pruning of time.
Edit: Corrected a mathematical error. Not 200′000 generations ago, some 2500 generations ago. Does however not invalidate the logic