Humanity showed up 10^18 times too early

Robin Hanson, an economist at George Mason University, opens with a technical term he treats the way physicists treat the Big Bang: *grabby aliens*. They expand fast and they change stuff. Fermi asked where the aliens are. Hanson splits the question. Quiet, hard-to-see aliens could be almost anywhere, because they leave no obvious mark. Loud ones, by definition, would be visible. Grabby is his and his co-authors’ particular mathematical model of that louder category, from the paper “If Loud Aliens Explain Human Earliness, Quiet Aliens Are Also Rare.”

Humans already behave this way at small scale. We turn forests into farmland and harbors into cities. In a competitive world, grabbing resources and leaving them as they were is a losing strategy, so expanding civilizations should paint the sky with spheres of altered activity. The model has three parameters, all claimed to be set by data: how fast the spheres expand, and a power-law for when civilizations appear — a constant and an exponent. We have not started expanding yet, but if we do, Hanson says we will do it within something like ten million years, a blink on cosmic clocks. That makes our present date a fair sample of when an advanced civilization shows up.

The exponent comes from *hard steps*: photosynthesis, sexual reproduction, eukaryotic cells, multicellular animals, and whatever else had to happen between the simplest life and a species ready to leave its planet. Each step is a long search through a huge space of possibilities, and all of them had to finish before Earth stopped hosting complex life. Two durations do most of the work: roughly 1.1 billion years left before Earth becomes uninhabitable for complex life, and life appearing about 0.4 billion years after it first became possible. Solving from those windows puts the step count between about 3.9 and 12.5, with a middle estimate of at least six.

A process that needs *N* hard steps has a chance of finishing by time *t* that scales like *t* to the *N*. At *N* around six, the dependence is brutally steep: a little earlier is almost impossible, a little later takes off. Hanson’s analogy is cancer. A cell needs several unlikely mutations, also on the order of six, so any one cell is unlikely to turn malignant in a lifetime — yet a body has so many cells that lifetime risk is around 40 percent, rising roughly as age to the sixth. Same math, different values. He insists on separating prediction from judgment. Mix them and you get wishful thinking; look coldly at what happens if we do nothing, then ask where the levers are.

The remaining parameter is expansion speed, and it is set by a stubborn fact: we look up and see nothing. If grabby civilizations expanded at, say, 1 percent of light speed, a typical newborn civilization would see huge spheres of activity, bigger than the full Moon. We do not. Slow expansion would make the empty sky a puzzle. Fast expansion solves it. At the speed of light you would not see them until they arrived; Hanson’s estimate is that they must be moving at least about a third of light speed. That sounds insane until you remember the alternative is to throw the whole model away.

Why believe the model at all? Because we are crazy early. The universe is almost 14 billion years old, around a star maybe five billion years old, while the average star lasts roughly five trillion years — a thousand times longer. Under the power law, a planet that lasts a thousand times longer is a thousand to the sixth, or 10^18, times more likely to produce advanced life if space stays empty and waiting. “How lucky do you feel?” Hanson asks. Ten to the eighteen is a billion billion. Feeling lucky is not an explanation. One of the assumptions that produced that number has to be wrong, and the assumption he drops is that the universe stays empty.

On this picture the cosmos is already filling. Civilizations appear at random places according to the power law, then expand; his best estimate is roughly one origin per million galaxies, with a meeting in about a billion years if we go out to meet them. When the spheres finish filling the usable volume, it is too late for anything like us to appear. We showed up now because the window is still open and closing. If they can travel faster than light, he says, they would already be everywhere, because faster-than-light travel lets you go backward in spacetime. The empty sky plus our date jointly say they are fast, not omnipresent, and not yet here.

What he likes most is that the model restores a living cosmology. Ancient peoples told origin stories with agents and agendas. Modern cosmology is mostly dead: we are a little knot of activity in a dark, empty sea. Grabby aliens rewrite the plot. The universe starts empty; primitive things like us appear; some stop being quiet and expand; for a few billion years the spheres grow and then meet; then for something like the next hundred billion years they commune, until ordinary cosmology says expansion isolates galaxy clusters around 100 to 150 billion years from now. Each civilization would hear about, even meet, thousands of others. We might hope to join that community. The clock on the wall is one of the few things we can see, and, like the ancient Greeks estimating the size of the Earth from clever geometry, our current date is information if you know how to read it.

Lex Fridman asks whether a universe of alien civilizations would be defined by war. Hanson answers with a narrower claim: it is defined by competition, and how far competition implies war is open. Life on Earth has been competitive for a long time, and among humans that often became war over the last ten thousand years. Many people now hope to suppress business and military competition, to build something close to world government or a world community. They may like that so much that they block interstellar colonization, because colonization would return severe competition. There is no serious way to centrally govern a rapidly expanding sphere. The solid prediction about grabby aliens is that they have already accepted competition, internally and at their borders. Whether the border contest is military is another question.

Even a vast nuclear war, he says, would not kill all humans or all life. Civilization could return within a hundred thousand years. On cosmological scales, one catastrophic “asshole at the party” is a delay, not an ending. What could actually stop expansion is a regime of global governance that limits military and business competition. Fridman is more frightened of centralized power than of war. Hanson plays the other side: for half a century a shared elite community has already produced policy convergence on spectrum, medical experiments, pandemic rules, nuclear power, and aviation, without a single global capitol. Elites meet at places like Davos, gossip, and seek one another’s respect. The war in Ukraine, in this telling, is that community coordinating resources and moral stance — the same passion that animates climate and global-poverty campaigns.

The key choice, Hanson says, is whether global governance makes expansion more likely or less. His answer is less. Inside a solar system a rebellious colony near Neptune can still be smashed with rocks. Once ships leave the system, central control is gone, and a single party that chooses to colonize is enough. People fear competition not only because it produces death, but because it changes who we are and what we value. Look at the last ten thousand years honestly and ancient people are not moderns in costumes. Add brain emulations and dozens of later technologies, and descendants can sit in a vast space of possible minds. *The Age of Em* is his attempt to analyze the next era after foragers, farmers, and industry by using competition as the lever. The world he gets is very different. He thinks it is okay. He admits many readers will not.

Hanson offers two predictions he thinks are robust. First, humans discount the future fast — roughly a factor of two per generation, a side effect of sexual reproduction, since descendants a generation away share only half your genes. That discounting lines up with interest rates. It also explains why ordinary people and market prices shrug at global warming, and why futurists struggle to make anyone care about decades ahead. In the longer run, creatures selected for having many descendants will care about the long run, because otherwise they miss decisions that change descendant counts. On risk, the evolutionary literature says you should be logarithmically averse to shared risks that could zero out all your descendants, and close to risk-neutral on uncorrelated ones. You want a *portfolio of descendants*.

The second prediction is sharper. We reproduce through a hodgepodge of preferences — sleep, hunger, sex, excitement — that happened to produce children in an ancestral environment. In the current environment that package is failing, which is why fertility is falling, a large mistake from evolution’s point of view. Longer-run creatures will abstractly know that what they want is more descendants, and they will think through every strategy that serves that goal. Descendants need not be human babies. They can be AIs, genetically altered copies, or lineages of some other life form. What matters is that they will have thought hard about what counts as a descendant. Hanson also says they will almost entirely be artificial: not necessarily non-organic, but made in factories from CAD-like designs with scale economies far beyond the tiny factories in our cells. Any aliens we meet, he thinks, gave up biological reproduction long ago.

Hanson ran a Twitter series defining “hello aliens”: civilizations that, in the next million years, might recognize humans as intelligent, travel here, and make their presence and abilities known. Followers guessed that more than 80 percent evolved from two-gender species. He is not shocked. Earth has far more two-gender species than one-gender or many-gender ones; males can take higher variance, which weeds out mutations more efficiently. The land-versus-ocean poll is where he thinks people go wrong. Voters favored land. He thinks they discount ocean worlds. The planets that last trillions of years orbit small stars, with flares and tidal locking — problems that barely apply under an ocean. If life can start in the sea, long-lived planets remain in the game. People assume our path is the only path. He wants more credit for dolphins and even *Waterworld*.

On nuclear war with more than ten weapons fired in anger, the popular vote was under 20 percent. Hanson thinks that is too low, because people assume a nuclear war ends civilization forever. He thinks you could rise again in ten thousand or a hundred thousand years, which on cosmic clocks is almost nothing. Mutual assured destruction has clearly had an effect; the question is how strong, for how long. “The night is still young.” Seventy years without a nuclear war is not a completed experiment. War has a thick tail: most damage sits in the few largest events, as with pandemics. Over fifty years he might say nuclear war is unlikely; over five hundred or five thousand, if the underlying risks continue, the cumulative chance gets large.

The cultural change he uses to project those risks forward is a drift back toward forager attitudes as we get rich. Humans spent a million years in small bands that shared food, decided together, and held little property. Farming, ten thousand years ago, required a huge cultural rewrite: slavery, war, kings, inequality, staying put, more disease. Poverty enforced the new norms — a young woman really could starve if she had a child out of wedlock. Wealth made those threats less credible, so people follow inclinations more than official farming rules, except at work. He thinks this drift explains less slavery, more democracy, less religion, lower fertility, more promiscuity, travel, art, leisure, and shorter hours. *Star Trek* and Iain M. Banks’s Culture novels celebrate the same destination. Forager violence, as he reads the record, was mostly sexual jealousy, not organized wars over land. Chimpanzees have tribal wars. Human foragers did not. Farmers brought the chimpanzee pattern back.

Hanson separates prior from likelihood. On priors — which scenarios are even plausible given what we know — he claims expertise. On the sightings themselves he does not: he is not a pilot and does not study atmospheric phenomena. From the amateur chair, the reports look “damn puzzling,” and the casual dismissals he has seen fail. The courtroom analogy: a random person’s prior of being a murderer might be one in a million, yet we hold trials because evidence can overwhelm that. His rough prior that some UFOs are aliens is about one in a thousand, high enough that strong physical evidence could push the claim over 50 percent. He leaves the physical evidence to others.

The scenario he uses to make the data cohere has two parts. First, panspermia siblings. Life on Earth appears early and already complicated, which has always been a puzzle if a few hundred million years of chemistry had to reach a sophisticated cell. Panspermia lets that complexity accumulate on another planet for billions of years, then ride a rock into the stellar nursery where the Sun was born. Many neighboring worlds could have been seeded at once, then drifted apart. A few thousand sibling stars with matching spectra would still be findable in the galaxy. Most would never reach an advanced stage. One might have, hundreds of millions of years before us — enough time to fill the galaxy if they had wanted to. The sky is not full, so they chose not to become grabby.

Not expanding for that long is a fiercely enforced rule. Let one little ship leave with the ability to found a colony and the game is over: the universe becomes grabby from that origin. Hanson compares it to Europe trying to forbid anyone from reaching the New World for a hundred million years. It would take competence, not a slip. An expedition to Earth would then be a rare, intentional exception, and a plausible agenda is to enforce the same rule on us. They have not destroyed us, so some empathy or moral limit is in play. They have likely watched for a long time. Fusion, more than nuclear weapons, would tell them how close we are to expanding. They probably have a button. They would rather not press it.

The last puzzle is the middle path. An advanced visitor could stay fully invisible — dark satellites in orbit — or land on the White House lawn. UFO reports do neither. They hang at the edge of visibility with obviously impressive abilities. Hanson thinks most social animals have status hierarchies, and the visitors would reasonably guess that we do too. The strategy is to take the top of ours, the way humans domesticate dogs by becoming the lead wolf, or kings build bigger palaces and parades. Get too close and you risk hurting us, or revealing some practice we would hate; humans are curious about foreigners until a moral line is crossed, and aliens will cross one. Stay a little mysterious, and clever humans can still infer the agenda from two facts: they are here, and the sky is not full of grabby empires. Many of us already do not want to expand. Impressive, long-lived neighbors saying the same thing could tip the rest.

From UFOs the conversation moves through conspiracy in the ordinary sense — Manhattan Project-scale secrets are possible; looser, longer, less disciplined ones are harder — and into *The Elephant in the Brain*, co-authored with Kevin Simler. The elephant in the room is a social taboo. The elephant in the brain is an introspective one: an important feature of mind we refuse to admit. The two core elephants are selfishness and self-deception. “You don’t make decisions, you justify them to an audience,” Hanson tells Fridman. The voice listening to the podcast likes to think it is president or king of the mind. It is the press secretary.

Humans have norms, and a meta-norm: if you see a violation, tell others, coordinate, and punish. Successful accusation can mean exile, which in the ancestral world meant death. Many norms are about motive — hitting on purpose is a crime, hitting by accident is not — so consciousness spends its time watching behavior and keeping a clean story. You do not know why you really do things. You know the story that will survive an accusation. Most people accept self-deception in the abstract. They will not apply it to school, doctors, or voting. The book walks through ten domains; specialists in those domains, Hanson says, still have not carried the psychology into public policy.

Asked to explain his own contrarian streak, he notes that by definition he cannot tell you what he is deceiving himself about. The practical move is to own the usual motives. He wants to be well thought of. He wants attention and glory from intellectual work. He is less eager to add a footnote inside a local tradition, more willing to hunt neglected questions and surprising updates. The audience in his head is Einstein, Freud, Socrates, and people like Tyler Cowen, not the easiest 60 percent of customers a salesperson would target. That may have cost him ordinary success. It is also, he admits, consistent with wanting to be remembered by a larger, later crowd.

The sharpest everyday application is medicine. In randomized trials that give some people a larger medical budget — more doctors, more drugs, not one branded pill — the extra medicine does not make them healthier. There is no relationship, not even a negative one. Hanson thinks everyone should know this and almost no one does. He has lived in Washington for decades, reading the *Washington Post* and its weekly health section, and never saw it mentioned. It is the kind of ugly fact people would rather not learn, because it undercuts a ritual they feel passionate about.

If you were focused on health, you would care a lot less about medicine. If medicine is how you show that you care, and how others show they care about you, the spending makes sense — unless you find a cheaper substitute. His method for untangling such systems is three steps: do not look at yourself, look at other people’s broad patterns; list theories and pick the one that matches the behavior; then assume it is true of you too. You are poorly placed to judge whether you are the exception. Producers are easy to explain: drug companies and doctors have clear incentives to give customers what they want, and democratic governments to give voters what they want. The puzzle sits with patients and voters.

The excuse “people are just ignorant” works for a new gadget. It works badly for a practice that has been used in roughly the same way across the world for thousands of years. Health-economics papers that claim medicine is effective, he argues, often have more degrees of freedom; the handful of straightforward aggregate randomized experiments are harder to massage. Hidden-motive stories always face a dog-ate-my-homework problem: the cover story is chosen to be locally plausible, so you need a wider range of evidence than a single encounter provides. None of this requires 100 percent of people to be self-deceived. A minority can win by being contrarian. Most people still shower and comb their hair. A few slobs do not kill the hypothesis that people want to look good.

Trust in institutions fell during the pandemic, and Hanson starts from what authority is *for*. It lets you take a less conservative action without looking gullible: you can point to the credential. With no accepted authorities, people grab conflicting local ones or fall back on crude defaults. The gain is shared, context-sensitive advice. The trap is that the best strategy *for the authority* is not maximal information. An ideal source would dump detail and update constantly. An organization that does that does not remain the perceived authority for long. Being seen as the certified voice and being the most informative source are in conflict. That was the point of his blog post on the accuracy of authorities.

Fridman adds the human layer: titles go to people’s heads, chairs want to be remembered as the leader in a crisis, and narrative starts to outrank mission. Hanson, speaking as an economist, would rather blame institutional incentives than individual sin. People play the local game the institution hands them. Information institutions, including academia, are especially far from a cost-effective way to collect and share knowledge. Academics are rewarded for being impressive more than for being interesting or useful. Tenure’s freedom is real — he enjoys it — but he will not celebrate it without asking whether students and research patrons get their money’s worth. Students, in his analysis, are mostly being sorted and labeled at enormous cost. Patrons are mostly buying affiliation with credentialed impressiveness.

His long-standing alternative is prediction markets. If a patron wants the mass of the electron neutrino, subsidize a betting market on that number; people who learn the answer can trade on it. If you want an accomplishment, use prizes, which were once far more common. For the customers who really want to know who is good — students, journalists, patrons — he would create markets on what historians centuries from now will say about each person’s contribution. Looking back two hundred years, we can see which paths were dead ends. Pay those future evaluators, trade the claims now, and rate living researchers on today’s prices instead of publication lists. If the market overweights recent prestige, that is an arbitrage. Academia currently has no such correction: going with prestige is the safe move even when you privately think it is inflated. A sibling design makes research labs for-profit, like hedge funds, earning by trading on what their researchers learn.

Hanson collects *viewquakes*: insights that dramatically change a worldview. In physics, special relativity smashed a perfectly adequate everyday idea of time into spacetime that looks different from different frames. Quantum mechanics was a larger shock still, a representation that looks nothing like particles at points with fields in between. He thinks these results were highly overdetermined. Without Einstein, someone else would have had essentially the same theory within half a century; several formulations of quantum mechanics arrived in the same few years. Crowded fields let you pull a result forward by months. His research strategy is the opposite: find the thing nobody would do for a long time if he did not do it.

In AI, the first viewquake for almost everyone is how complicated human minds are. Outsiders invent a concept and think implementing it yields human-level intelligence. Working through particular problems shows how much the mind is doing that you never notice. Fridman says that shock is early Dunning-Kruger, and the later shock is the reverse: a small thing, like a neural net, can be extremely powerful. Hanson still puts human-level AI at least a decade away, and is mildly confident it is at least three. The useful frame is *chunkiness*. At the scale of the whole economy, almost every innovation is a rounding error; steam engines and shipping containers are the rare exceptions. If AI is “all mental tasks,” it is perhaps half the economy and must be many small pieces. If AGI is a narrow stack of key insights, one chunk can move the world. Brain emulations are chunky in an engineering sense: 25 percent of a working brain is nothing until you cross a threshold. Economically they may still arrive smoothly if the first copy costs a billion dollars and the price then falls.

A curious fact: algorithmic improvement has tracked hardware improvement. His story is that algorithms have a fixed cost plus a marginal cost. Small machines only let you try low-fixed-cost ideas; bigger machines let you pay to search a wide log-normal of heavier algorithms that win on the margin. In economics, the first viewquake is the power of markets — a robust first pass at optimization in a huge space, better than most of the limits people want to impose. A more specialized shock: rational agents who only want the truth would not knowingly disagree. Humans disagree constantly because their priority is not truth. In political science, the main failure is not agreeing on what to do. Most systems could execute if a trusted authority named the answer; we invade Iraq when nobody is positioned to say it is stupid. We could build more trustworthy authorities. We do not. *The Elephant in the Brain* adds that political talk about making the world better is often loyalty signaling, and it works. Tribes are conveniently color-coded. The deeper failure is that we barely explore the vast space of possible institutions. Try them at small scale, he says, and better ones would show up.

Asked for advice to high-school and college students, Hanson first notes that they usually mean success by ordinary standards, which is not how he lived. You live in a rich society, you will live a long time, and whatever career you take, you will have surplus energy for something else. Combining job and passion is nice and often expensive. If you care about a topic, start. You do not need permission. It may take decades, and decades are enough to make enormous progress on almost any interesting thing. He switched careers at 34, with two children aged zero and two, leaving software research for social-science graduate school. Changing your mind in the second half of life is allowed.

What to read: textbooks, then review articles. Until those are in your bones, he is not sure you should be living on blogs, Twitter, and podcasts. Textbooks are humanity’s best compressed summary. If the field is technical, do the problems at the end. Idealistic young people, he finds, are in a rush. Productivity typically rises into the forties or fifties, when income, contacts, and a feel for how the world works all peak. That is when large impact usually happens. Before then you are mostly accumulating. Practice doing hard things, but do not expect to finish them at eighteen.

The advice he would give his younger self is to have more descendants. It mattered more than he realized. Happiness is how you feel right now about right now; meaning is how you feel about your whole life. Many things buy happiness without buying meaning. If you must choose, take meaning. Children are the textbook case: more work, more meaning. Influence on the future has overwhelmingly run through having children, and you are built from thousands of generations that succeeded at it. Even with more technological ways to matter, he still puts genetics at more than half of a person’s long-run influence. Robots can echo a little of that magic. They are not a substitute.

The darkest places in his own life, Hanson says, were rejections, which left him rejection-averse. Fridman wonders why a scarred person would then court intellectual rejection. Hanson sorts the two: the painful rejections were not mostly about his ideas, and intellectuals are still people. On love, his answer is opacity. Some wants are transparent — thirst has a cause, a remedy, and an ending. Love is not like that. You are drawn to it without a clean model of why, and that unclarity is part of what it is. Some emotions cannot be summoned on command. You remember facts of a scene and lose the feeling, then find it again only in a similar state, which is one reason people re-consume the same songs and stories. Language thickens the palette: without a word, a feeling is hard to hold or compare. A community that names it can grab it.

His most painful loss is youth. That earlier person is gone, without fanfare or a funeral, and Hanson does not even remember him well. In a few decades the present person will be gone too. He wears a cryonics tag: if medicine gives up, freeze the head and bet on a future that can emulate the brain in a computer, without needing the whole biological body. He gives himself at least a 5 percent shot, and is sad about the more likely failure. Immortality, strictly, cannot be checked. The workable question is whether, at each moment, you want to keep going. Emulations could archive copies at different ages and later swap a tired self for a younger file. Most of them, he thinks, would live in constructed environments and not experience those rooms as “virtual,” any more than we experience painted drywall as fake because it hides the studs.

On the simulation argument, he wants a number, not a vibe. What share of conscious creatures are having their experiences inside a simulation of some past? That ratio depends on how fast the population of minds grows versus how fast interest in a given past date fades. Google Ngrams, movies, plays, and games all show that attention to a particular year falls faster than population grows. Most entertainment is about the recent past, not Rome, still less Egypt. Counting all creatures ever, he puts the simulated share at probably no more than 10 percent. Grant a small probability; do not treat it as zero, and do not treat it as near-certainty. In a competitive world, leisure is a limited budget, and leisure in which the subject does not even know it is leisure is a thinner slice still.

What is the meaning of life, for a creature like us? We cannot summarize it well. Our motives are an opaque patchwork of thirst, sex, sleep, and attention, so we often cannot say why we are switched on or off. That mess *is* what it means to be a human of this century. Descendants will be different: they will know they want more descendants. Why not keep the mystery on purpose? Because that is not how the world has changed so far. Individuals can plan a career, a city, a partner. World history has mostly been competitive: someone somewhere adopts an advantageous practice, it spreads, and others are forced to follow. That analysis predicts a future. It does not promise we will like it. Grabbiness and explicit descendant-maximizing both come from competition; which arrives first is unclear. Compared with a dead cosmology, Hanson says, at least this version has actors, agendas, and something happening. That is already enough to make the universe interesting. Fridman closes with Ray Bradbury: “We are an impossibility in an impossible universe.”