It was hard times for the bomber pilots that floated over Europe, their planes incinerating cities below, like birds of prey. Even as they turned the once-bustling streets beneath to howling firestorms, death had become a close companion to the crews of the Allied bombers as well. In fact, surviving a tour with the Bomber Command had become a virtual coin flip. While their munitions fell mutely from bomb bays, an upward sleet of fire from smoldering city grids and darkened farmland shot the planes out of the sky like clay pigeons. For recruits encountering the freshly empty bunk beds of dead airmen, morale was sapped before they could even get in the cockpit. Hoping to slow this attrition, Allied officers studied the pattern of bullet holes in returning aircraft for vulnerable parts to reinforce with armor.
It was natural to think that the bombers needed more armor where (it appeared) they were taking the most bullets. But the Hungarian-born mathematician Abraham Wald, and his colleagues at the Statistical Research Group at Columbia University, had a novel, if counterintuitive, prescription. Don’t protect the planes where they were taking the most damage, Wald said. Armor the planes where there were no bullet holes at all.
“You put armor where there are no holes, because the planes that got shot there didn’t return to the home base,” says Anders Sandberg, a senior research fellow at University of Oxford’s Future of Humanity Institute. “They crashed.”
The holes didn’t show where returning planes were likely to get hit, but only what it was possible for later observers to see. This is known as an observer selection effect, and the same sort of bias might apply not only to perforated planes, but to whole worlds as well.
What if, when we looked at our own planet’s past, we saw a similar pattern? After all, there are 100-mile impact craters on our planet’s surface from the past billion years, but no 600-mile craters. But of course, there couldn’t be scars this big. On worlds where such craters exist, there is no one around afterward to ponder them. In a strange way, truly gigantic craters don’t appear on the planet’s surface because we’re here to look for them. Just as the wounds of the returning planes could reflect only the merely survivable, so too for our entire planet’s history. It could be that we’ve been shielded from these existential threats by our very existence.
“Observer selection effects are really the kind of effects where the data you get is going to be dependent, in some sense, on survival, or that you as an observer exist,” Sandberg says. “Now this gets really interesting and scary when we apply it to our own survival.”
It’s something of a miracle that life on our planet has been left to evolve without fatal interruption for billions of years. Such a long unbroken chain of survival, however unlikely, is necessary for bags of mud and water like ourselves to eventually sit up, and just recently, to wonder how we got here. And like the bullet-riddled—but safe—planes, our planet has survived countless near-fatal blows. There have been volcanic apocalypses, body blows from supersonic space rocks the size of Mount Everest, and ice ages that might have frozen the planet almost to the tropics. Had any of these catastrophes been worse, we wouldn’t be here. But they couldn’t have been worse for precisely that reason.
As Sandberg and his coauthors Nick Bostrom and Milan Ćirković write, “The risks associated with catastrophes such as asteroidal/cometary impacts, supervolcanic episodes, and explosions of supernovas/gamma-ray bursts are based on their observed frequencies. As a result, the frequencies of catastrophes that destroy or are otherwise incompatible with the existence of observers are systematically underestimated.”
That is, our forecasts about the future could be blinded by our necessarily lucky past. Not only is it impossible to look back and find truly world-ending impact craters in our planet’s history—stranger still, it would be impossible to find these impacts in the rock record even if they struck planets like ours all the time. Existential hazards, even if they’re extremely likely, might hover just out of frame, concealed by our “anthropic shadow.”
“Maybe the universe is super dangerous and Earth-like planets are destroyed at a very high rate,” Sandberg says. “But if the universe is big enough, then when observers do show up on some very, very rare planets, they’ll look at the record of meteor impacts and disasters and say, ‘The universe looks pretty safe!’ But the problem is, of course, that their existence depends on them being very, very lucky. They’re actually living in an unsafe universe and next Tuesday they might get a very nasty surprise.”
If this is true, it might explain why our radio telescopes have reported only a stark silence from our cosmic neighborhood. Perhaps we’re truly extreme oddballs, held aloft by a near-impossible history—one free from deadly migrating gas giants and solar-system chaos, but also filled with freakishly favorable accidents, like a cataclysmic impact early in our history that created a strange, gigantic moon that stabilized our orbit and allowed complex life to flourish. As the solar system continued to shake out, we somehow ended up with just the right amount of water to lubricate plate tectonics, keeping the climate habitable over hundreds of millions of years and preventing a Venus-style planetary resurfacing catastrophe, but not so much water that we wound up on a lifeless water world.
Earth history teaches us that something as seemingly benign as briefly having a supercontinent can very nearly end the world several times over, illuminating just how fragile the entire project of complex life really is. And not only have we benefited from this astounding series of fortunate events, through it all we’ve somehow never been set back to square one over 4.5 billion years—even while potentially sterilizing comets like Hale-Bopp keep eerily sailing past us. Perhaps this fantastic luck—one that’s necessary to someday produce observers like ourselves—implies that the great Elsewhere is filled, to un-traversable distances, with indifferently swirling gas and lifeless rocks.
The same eerie observer selection effects that could explain our strangely benign cosmic history might have been at work in our very recent past as well. After all, just as we can’t find ourselves on a planet pocked by continent-spanning craters, we similarly can’t find ourselves on a planet that’s been recently destroyed by our own hand either.
We live in a world newly endowed with the capacity to end itself with nuclear weapons. For more than half a century, our world has continually threatened to spill over into an all-out nuclear war. But somehow it never has. If it had, you wouldn’t be reading this article.
Perhaps you could use this seven-decade trial run to estimate how likely the nuclear holocaust really is. Since we’ve gone so long without blowing ourselves up, it stands to reason that the probability of such a catastrophe happening in any given year must be fairly close to zero.
“That sounds really good—except, of course, if there had been a nuclear war you wouldn’t be around doing this calculation,” says Sandberg.
Either nuclear war really is this unlikely or, perhaps, something much stranger is at work. Taking observer selection effects into account, our survival of the nuclear age until now might only point to unseen, towering dangers. Perhaps it’s the case that most civilizations that develop the capacity to kill themselves, quickly do.
“So now you can imagine a world where the probability per year of nuclear war is actually 50 percent. So then the first year, the first half of worlds get nuked. Then the next year half of those survivor worlds get nuked. And so on. So in this very scary scenario—still after 70 years—if you have a big enough universe or many parallel universes, you’re still going to have some observers [left over] who say ‘Hey! It looks like we’re pretty safe!’ And again they will get a very nasty surprise when the nukes start flying.”
The modern world, like the larger cosmos, could be much more dangerous than we’ve experienced so far—precisely because we’re still here. The observed nuclear apocalypses in our past (zero) tells us nothing about their future likelihood.
On September 26, 1983, a Soviet duty officer named Stanislav Petrov found himself paralyzed. He was manning the Soviet Union’s early-warning system for incoming American nuclear missiles when his computer alerted him to the unthinkable: a highly reliable detection of incoming warheads. It was a nuclear first strike and the possible end of civilization. Or it was a computer malfunction.
“The siren howled, but I just sat there for a few seconds, staring at the big, backlit, red screen with the word ‘launch’ on it,” Petrov later told the BBC. “All I had to do was to reach for the phone; to raise the direct line to our top commanders—but I couldn’t move. I felt like I was sitting on a hot frying pan ... Twenty-three minutes later I realized that nothing had happened. If there had been a real strike, then I would already know about it. It was such a relief ... they were lucky it was me on shift that night.”
On November 9, 1979, the U.S. national-security adviser Zbigniew Brzezinski awoke to a military assistant solemnly informing him of another all-out nuclear first strike, this time coming from the Soviet side. As he prepared to call the president to advise a counterattack, Brzezinski decided not to wake up his wife, preferring to let her die peacefully in her sleep with the rest of humanity. With moments to spare Brzezinski learned it was a false alarm. Only a few months later, a 46-cent microchip caused three more false alarms in the country’s early-warning system. In November 1983, the United States and its allies conducted an extremely realistic war-game exercise called Able Archer 83 that nearly provoked the Soviet Union to war, a fact that wasn’t widely appreciated at the time.
On January 17, 1966, a U.S. Air Force bomber carrying four hydrogen bombs crashed off the coast of Spain to no geopolitical effect. On October 27, 1962, when Soviets shot down an American spy plane over Cuba, President John F. Kennedy’s advisors urged him to retaliate. He didn’t. But as his brother Robert later wrote about those harrowing moments, “There was the feeling that the noose was tightening on all of us, on Americans, on mankind, and that the bridges to escape were crumbling.” Two days earlier, with the world on edge and the military at its highest readiness level ever, a bear climbing a fence at an air base in Duluth, Minnesota, set off the wrong alarm at the nearby Volk Field Air National Guard Base, sending pilots scrambling to nuclear-armed jets. A truck frantically speeding down the runway stopped the planes before they could take off.
But somehow none of these odd flirtations with the apocalypse spilled over to the real thing. If they had, we wouldn’t be here. But what if nuclear war was actually quite likely? If so, worlds like ours that avoid doomsday would be statistical oddballs, and require strange and increasingly unlikely lifelines as time went on. But in a big enough universe they’d exist. And they’d be the only ones around to study the history of nuclear diplomacy. Perhaps, then, we’re one of these diminishingly rare, and increasingly bizarre, planets that can only survive its first 72 years with nuclear weapons by witnessing the inexplicable.
When I first spoke to Sandberg three years ago, he seemed to think that this was a possibility—that these close calls of the Cold War might be censoring us from nuclear Armageddon. For every one Stanislav Petrov left confused as to why he didn’t push the button, there might be 99 others who suffered no such equivocation. But on those broken worlds, scattered across the universe, there’s just no one left to condemn him for it.
“We should probably be seeing more miracles,” Sandberg told me in that original conversation. “There would be not just Petrov, but failures of phones and other things that would save us. As time went by we would see more and more of these weird events.”
Since then though, he seems to have changed his mind. Incorporating observer selection effects into his calculations, Sandberg now thinks that the close calls in our past might actually be reassuring. If it’s true that episodes of nuclear brinkmanship tend to go sour and kill everyone, then those who survive the invention of nuclear weapons—like we have—should actually expect to find very few of these close calls in their past. Instead we should expect to look back on strangely peaceful histories, free from hair-raising emergencies like the Cuban missile crisis. That’s because ultra-peaceful worlds would result in far more good outcomes (and far more survivors) than those that continually flirt with the apocalypse. After all, who are you more likely to be: someone who just survived 100 rounds of Russian roulette or someone who never picked up the gun in the first place?
But since we do see these sorts of apparent close calls in our past, maybe they weren’t all that near to the bull’s-eye of extinction to begin with; and what we thought were bullet holes perilously near the engine block were actually harmlessly spattering the fuselage. Perhaps the existence of Petrov means that nuclear brinkmanship, and even nuclear war isn’t quite the existentially dangerous game we thought it was. Perhaps nuclear war doesn’t kill everyone. Or maybe it’s the case that nuclear first strikes don’t inevitably lead to both sides unloading their entire arsenals. If it did, Petrov never would have gotten so close to that red button to begin with.
“My coauthor felt that my conclusion wasn’t quite cool enough,” says Sandberg. “The standards around the office here are kind of crazy sometimes.”
If observer selection effects can affect the errant trajectories of killer comets past, and maybe even what we see in our recent human history, perhaps this strange bias reaches back all the way to the very beginning.
Anthony Aguirre, a theoretical cosmologist at the University of California at Santa Cruz, thinks the fact that the universe has already lasted as long as it has could be the strangest product of observer selection effects of all. Aguirre doubles as a founder of the Future of Life Institute, an organization that—like Sandberg’s—is dedicated to identifying existential threats in our near future, and one that counts Elon Musk and the late Stephen Hawking among its august advisory council.
“So one catastrophe that’s not often talked about ... is vacuum decay,” he said cryptically in his office. As Aguirre would explain, these two words represent the most dangerous and final catastrophe imaginable.
Though the end of the universe is typically thought of as a slow unraveling in the far future—the eternal dissipation into darkness after our brief springtime, leading to a cold, empty epoch that will stretch into forever—the universe could also end violently, Aguirre says, and at any time.
“So there are various fields that permeate space,” he began. “The electromagnetic field is a common one. The electron field, the proton field, the Higgs field—these are all fields that exist everywhere. And when we have empty space what we really mean by that is that there are no excitations of these fields. So an electron is like an excitation of the electron field, and when we take all the electrons away, the field is still there. We say that it’s in its vacuum state. But a vacuum state is not necessarily completely stable. It’s the state that you get if you take away all the electrons, but it might still have some energy associated with it. And that energy level could, in principle, be different, or lower. In fact, there’s no particularly good reason to think that the vacuum that we’re in now is the lowest-possible vacuum state in terms of energy. And there are actually pretty good reasons to think that it’s not.”
He started laughing nervously. I didn’t understand why, but would go on to learn that if the vacuum state spontaneously dropped to some new energy level the laws of physics would go berserk and the universe as we know it would be over.
“It would start out as some sort of point-like event somewhere in the universe that would then expand at the speed of light and just kind of destroy everything in a sphere,” Aguirre said. “So as soon as that bubble passed over us we would then be in some other state of the laws of physics that was totally incompatible with us.”
This outrageous disaster isn’t just some academic flight of fancy, the product of coffee-addled fugues at a chalkboard. Just such a psychedelic transition might have happened in our universe’s infancy, as the laws of physics congealed from the primordial fire of the Big Bang, and the fundamental forces we know today crystallized from more exotic forms. In fact, in 2012, when the world celebrated the discovery of the Higgs field (via the Higgs boson)—the thrilling validation of half a century of theory—some physicists quietly averred that the Higgs appeared to be unstable, and may someday destroy the universe.
“So it could be the Higgs field, but it could be any combination of all the other fields in physics that, in principle, could transition to some other vacuum state.”
If it happened it would be the end of everything. And there almost certainly would be no observers afterward.
“But since we’re still here, there’s a reasonable inference that the timescale for this is billions of years, at least. But there’s no particularly good reason to think that it’s—well,” he stopped himself.
“So, there are different arguments you could make,” he said. “You could say that if it’s already been billions of years it could just as easily be trillions or quadrillions or quintillions of years or whatever, so let’s not worry too much, or ...”
Or it could be like those asteroids that kept missing the planet for billions of years before we got here, securing our eventual appearance but blinding us to future peril. Just as observers never show up on worlds that are quickly destroyed, they also don’t appear in universes that quickly unravel. No matter how common they are.
“It may be that ultimately the reason we’re around for so long,” Aguirre said, “is that we’re around.”
The ghosts of innumerable ill-fated universes began to hover over our conversation like skeletons at the feast—or planes at the bottom of the English Channel.
“So,” I started, “there are billions of universes that were hospitable to life but they just—”
“Yeah, they just didn’t last long enough. And we’re one of the universes that lasted long enough ... It may be that we’re kind of living on the edge. Like, we’re sort of the shortest-lived universe that would allow stuff to arise and start thinking about short-lived universes and so on. So that would be a bad scenario.”
Aguirre started laughing again. Just like Sandberg’s mysteriously absent continent-spanning craters, the end of the universe itself might be looming in the anthropic shadows, held at bay—until now—by our very existence. Perhaps it’s only possible to wake up in a universe that has managed an almost impossible cosmic stay of execution lasting billions of years. In the early days of the Large Hadron Collider, when the megamachine kept running into seemingly endless, and increasingly improbable, financial and technical snags, some researchers—calling these mishaps “anti-miracles”—even half-seriously proposed that the universe was censoring us from this sort of destruction of the world occasioned by a successful run of the collider.
There’s something bracing in the sort of license that cosmology grants to its practitioners to think very strange thoughts.
“Do you want to go through the looking glass?” Aguirre asked me.
“I’d love to,” I said, surprised that we hadn’t already done so.
Quantum mechanics, the remarkably successful and remarkably strange physics of the very small, makes predictions whose accuracy can be verified in the real world to an almost arbitrary number of decimal places. In other words, quantum mechanics provides the best description of how the world works at its most fundamental level. One of the most bewildering experimental results of quantum mechanics, and of the 20th century, is that particles seem to exist in a sort of probabilistic purgatory, existing everywhere at once and nowhere in particular, hazily spinning both clockwise and counterclockwise at the same time—that is, until they are observed. Once measured, these many possibilities collapse into one coherent result and the observer measures some specific value for a particle.
One of the leading interpretations of this quantum weirdness is that all of the possible realities for the particle that were winnowed away in this act of observation actually are realized somewhere in branching-off parallel universes, by observers in parallel universes—parallel universes just as real as the one in which we happen to live. Though the universe may be infinite in distance it may also be infinitely divergent in this sort of ontological zoo. This is called the “many-worlds interpretation” of quantum mechanics. Again, this is not an unpopular or esoteric theory. It is one of the most widely subscribed interpretations of the peculiar world of quantum mechanics among physicists. And if a universe-destroying vacuum-decay catastrophe played out, it would take place in this strange, existential arena.
“What’s interesting about the nucleation of these vacuum-decay bubbles is that it’s a quantum-mechanical event,” Aguirre said about the ultimate catastrophe.
That is, the spontaneous initiation of the end of the universe would be probabilistic—like an extremely high-stakes version of Schrödinger’s hapless cat in a box—splitting reality into versions where everything is obliterated and a luckier version that’s spared. And given that we may be creatures of the “many worlds” multiverse, constantly splitting off into different lives, Aguirre wonders whether we could ever actually experience the quantum apocalypse sweeping through the void when it arrived at our doorstep.
“So, do we notice anything? That’s the question.”
If the observer selection bias applies to our own lives, then perhaps we’re constantly being censored to the end of the universe.
“So suppose the many-worlds interpretation of quantum mechanics is right,” he says. “So, one of the two versions of us ceases to exist, but do we actually notice that? So one of us keeps going on just as if nothing happened. Arguably, from moment to moment, I can’t rule out that five minutes ago the other version of us died. There’s no way for me to say that. So there’s an interesting, troubling question as to whether these things could be happening all the time and we just don’t even notice it.”
If, as Epicurus claimed, where death is, we are not, and when we are, death is not, then perhaps we can only subjectively experience those winnowing branches of reality that aren’t being continuously wiped out by bubbles of vacuum decay. In other words, the universe might be ending every second; we’re just blissfully unaware.
But if this observer selection bias applies to bizarre disasters like vacuum decay it seems as though it should subjectively foreclose any branching many-worlds path that ends in one’s own death. I began to think back to my own existential close calls, like a childhood car crash where the truck bumper of a drunk driver plowed through side window of my mom’s station wagon and missed my head by inches. Am I a biased observer of that accident? After all, I can’t find myself today to be one of the many versions of me that got killed in this childhood crash.
“Doesn’t that lead,” I asked, “To a really bizarre world in which you should never expect to die?”
“So there would be nothing that protects you from slow decrepitude,” he said. “This is a defense against instantaneous, violent death, which you will find was avoided, but anything that sort of maims or injures or mangles you or whatever is still fair game.”
This condition of eternal torment, where one might survive arbitrarily long by subjectively navigating the narrowing tributaries of the many-worlds time lines, staying alive through increasingly—and eventually astronomically—unlikely life paths, is known as quantum immortality, or quantum hell.
“Quantum immortality is one of the scariest ideas imaginable,” says Oxford’s Sandberg. The topic has even become something of a taboo among physicists who think its widespread dissemination might encourage an amateur physicist with the courage of their convictions to try their hand at Russian roulette.
“The guy trying out [Russian roulette] will, from the perspective of most observers, just be dead,” says Sandberg. “Of course, there are a few, very rare observers that see him being very, very lucky. But as he keeps on doing this that set is getting smaller and smaller. But it always includes a version of himself.”
In my discussion with Aguirre I noted that his colleague, the MIT physicist Max Tegmark, had recently remarked that if this sort of subjective immortality was, in fact, granted by the many worlds of quantum mechanics, then I should expect to be a lot older than I am.
“So, if we expect to literally live forever because of quantum hell, can you say, ‘Well, it’s unlikely for me to happen to find myself in the first 41.5 years of my life,’ or whatever? I think that would depend on the measure because—now this is getting totally crazy—but the picture is that there are many, many young versions of you and they’re steadily getting weeded out, so, depending on how you count, you’re pretty likely to be young in that scenario.”
Noting me shift uneasily in my seat, Aguirre attempted to walk back the strangeness and return to the realm of common sense.
“So I think there are probably fundamentally things wrong with all of this,” he said. “I regard it as a reductio ad absurdum, but it’s not entirely clear what’s wrong with it.”
If we’re not, in fact, granted an eternal life by some combination of many worlds and observer selection effects, and the world ends for us with thudding finality, all is not lost—at least, not for conscious observers.
“The universe is certainly really big, and my bet is on infinite,” he said. “Because I think if inflation happened—and I think inflation probably happened—I think it probably implies eternal inflation,” he said, referring to the theory that our universe is only one of an infinite bubble of universes, pinched off from a never-ending roiling act of Creation. “And eternal inflation implies an infinite universe. If there’s an infinite universe then yeah, if we wipe ourselves out there will still be plenty of life in the universe somewhere. And plenty of humans ... and plenty of Peters and plenty of Anthonys sitting in offices.”
Aguirre was being both lighthearted and dead serious. His colleague Max Tegmark estimated that in an infinite universe there was an exact copy of myself 101029 meters away.
“There will be lots of those.”
Aguirre told me he had just finished the book Unbroken, the story of the World War II Air Force lieutenant Louis Zamperini, who miraculously survived the crash of his bullet-riddled bomber, an accident that killed eight of his other 10 crew members. Zamperini then survived a month and a half drifting in the open Pacific Ocean with no water, a period of privation that took the life of another crewmate; and finally two years being tortured by the Imperial Japanese. But survival stories, like Zamperini’s, might tell us less about the triumph and perseverance of the human spirit—though they certainly tell us much about that—than about how many other American soldiers had to die in similar circumstances to produce such a statistically unlikely biography.
“Obviously, most people in those circumstances die,” Aguirre said about Zamperini. “But the one that you read about later is the one who didn’t, so it seems like this startling series of amazing events.”
Maybe we live in a sort of Zamperini universe, owing our existence to a vast looming shadow of unseen, broken worlds. If we can only ever wake up on rare and seemingly miraculous worlds—and it’s a big enough universe—we shouldn’t be surprised to find our past filled with miracles.