Doomsday just isn’t what is used to be. Once the dominion of ancient apologists and their votary, the final destiny of humankind now consumes probability theorists, physicists. and technology luminaries. I’ll give some thoughts on probabilistic aspects of the doomsday argument after a brief comparison of ancient and modern apocalypticism.
The Israelites were enamored by eschatology. “The Lord is going to lay waste the earth and devastate it,” wrote Isaiah, giving few clues about when the wasting would come. The early Christians anticipated and imminent end of days. Matthew 16:27: some of those who are standing here will not taste death until they see the Son of Man coming in His kingdom.
From late antiquity through the middle ages, preoccupation with the Book of Revelation led to conflicting ideas about the finer points of “domesday,” as it was called in Middle English. The first millennium brought a flood of predictions of, well, flood, along with earthquakes, zombies, lakes of fire and more. But a central Christian apocalyptic core was always beneath these varied predictions.
Right up to the enlightenment, punishment awaited the unrepentant in a final judgment that, despite Matthew’s undue haste, was still thought to arrive any day now. Disputes raged over whether the rapture would be precede the tribulation or would follow it, the proponents of each view armed with supporting scripture. Polarization! When Christianity began to lose command of its unruly flock in the 1800’s, Nietzsche wondered just what a society of non-believers would find to flog itself about. If only he could see us now.
Our modern doomsday riches include options that would turn an ancient doomsayer green. Alas, at this eleventh hour we know nature’s annihilatory whims, including global pandemic, supervolcanoes, asteroids, and killer comets. Still in the Acts of God department, more learned handwringers can sweat about earth orbit instability, gamma ray bursts from nearby supernovae, or even a fluctuation in the Higgs field that evaporates the entire universe.
As Stephen Hawking explained bubble nucleation, the Higgs field might be metastable at energies above a certain value, causing a region of false vacuum to undergo catastrophic vacuum decay, causing a bubble of the true vacuum expanding at the speed of light. This might have started eons ago, arriving at your doorstep before you finish this paragraph. Harold Camping, eat your heart out.
Hawking also feared extraterrestrial invasion, a view hard to justify with probabilistic analyses. Glorious as such cataclysms are, they lack any element of contrition. Real apocalypticism needs a guilty party.
Thus anthropogenic climate change reigned for two decades with no creditable competitors. As self-inflicted catastrophes go, it had something for everyone. Almost everyone. Verily, even Pope Francis, in a covenant that astonished adherents, joined – with strong hand and outstretched arm – leftists like Naomi Oreskes, who shares little else with the Vatican, ideologically speaking.
While Global Warming is still revered, some prophets now extend the hand of fellowship to some budding successor fears, still tied to devilries like capitalism and the snare of scientific curiosity. Bioengineered coronaviruses might be invading as we speak. Careless researchers at the Large Hadron Collider could set off a mini black hole that swallows the earth. So some think anyway.
Nanotechnology now gives some prominent intellects the willies too. My favorite in this realm is Gray Goo, a catastrophic chain of events involving molecular nanobots programmed for self-replication. They will devour all life and raw materials at an ever-increasing rate. How they’ll manage this without melting themselves due to the normal exothermic reactions tied to such processes is beyond me. Global Warming activists may become jealous, as the very green Prince Charles himself now diverts a portion of the crown’s royal dread to this upstart alternative apocalypse.
My cataclysm bucks are on full-sized Artificial Intelligence though. I stand with chief worriers Bill Gates, Ray Kurzweil, and Elon Musk. Computer robots will invent and program smarter and more ruthless autonomous computer robots on a rampage against humans seen by the robots as obstacles to their important business of building even smarter robots. Game over.
The Mathematics of Doomsday
The Doomsday Argument is a mathematical proposition arising from the Copernican principle – a trivial application of Bayesian reasoning – wherein we assume that, lacking other info, we should find ourselves, roughly speaking, in the middle of the phenomenon of interest. Copernicus didn’t really hold this view, but 20th century thinkers blamed him for it anyway.
Applying the Copernican principle to human life starts with the knowledge that we’ve been around for 200 hundred thousand years, during which 60 billion of us have lived. Copernicans then justify the belief that half the humans that will have ever lived remain to be born. With an expected peak earth population of 12 billion, we might, using this line of calculating, expect the human race to go extinct in a thousand years or less.
Adding a pinch of statistical rigor, some doomsday theorists calculate a 95% probability that the number of humans to have lived so far is less than 20 times the number that will ever live. Positing individual life expectancy of 100 years and 12 billion occupants, the earth will house humans for no more than 10,000 more years.
That’s the gist of the dominant doomsday argument. Notice that it is purely probabilistic. It applies equally to the Second Coming and to Gray Goo. However, its math and logic are both controversial. Further, I’m not sure why its proponents favor population-based estimates over time-based estimates. That is, it took a lot longer than 10,000 years, the proposed P = .95 extinction term, for the race to arrive at our present population. So why not place the current era in the middle of the duration of the human race, thereby giving us another 200,000 thousand years? That’s quite an improvement on the 10,000 year prediction above.
Even granting that improvement, all the above doomsday logic has some curious bugs. If we’re justified in concluding that we’re midway through our reign on earth, then should we also conclude we’re midway through the existence of agriculture and cities? If so, given that cities and agriculture emerged 10,000 years ago, we’re led to predict a future where cities and agriculture disappear in 10,000 years, followed by 190,000 years of post-agriculture hunter-gatherers. Seems unlikely.
Astute Bayesian reasoners might argue that all of the above logic relies – unjustifiably – on an uninformative prior. But we have prior knowledge suggesting we don’t happen to be at some random point in the life of mankind. Unfortunately, we can’t agree on which direction that skews the outcome. My reading of the evidence leads me to conclude we’re among the first in a long line of civilized people. I don’t share Elon Musk’s pessimism about killer AI. And I find Hawking’s extraterrestrial worries as facile as the anti-GMO rantings of the Union of Concerned Scientists. You might read the evidence differently. Others discount the evidence altogether, and are simply swayed by the fashionable pessimism of the day.
Finally, the above doomsday arguments all assume that we, as observers, are randomly selected from the set of all existing humans, including past, present and future, ever be born, as opposed to being selected from all possible births. That may seem a trivial distinction, but, on close inspection, becomes profound. The former is analogous to Theory 2 in my previous post, The Trouble with Probability. This particular observer effect, first described by Dennis Dieks in 1992, is called the self-sampling assumption by Nick Bostrom. Considering yourself to be randomly selected from all possible births prior to human extinction is the analog of Theory 3 in my last post. It arose from an equally valid assumption about sampling. That assumption, called self-indication by Bostrom, confounds the above doomsday reasoning as it did the hotel problem in the last post.
Th self-indication assumption holds that we should believe that we’re more likely to discover ourselves to be members of larger sets than of smaller sets. As with the hotel room problem discussed last time, self-indication essentially cancels out the self-sampling assumption. We’re more likely to be in a long-lived human race than a short one. In fact, setting aside some secondary effects, we can say that the likelihood of being selected into any set is proportional to the size of the set; and here we are in the only set we know of. Doomsday hasn’t been called off, but it has been postponed indefinitely.