Once we can see them, it’s too late

This month Robin Hanson, the famous and controversy-prone George Mason University economics professor who I’ve known since 2004, was visiting economists here in Austin for a few weeks. So, while my fear of covid considerably exceeds Robin, I met with Robin a few times in the mild Texas winter in an outdoor, socially-distanced way. It took only a few minutes for me to remember why I enjoy talking to Robin so much.

See, while I’d been moping depressed about covid, the vaccine rollout, the insurrection, my inability to focus on work, and a dozen other things, Robin was bubbling with excitement about a brand-new mathematical model he was working on to understand the growth of civilizations across the universe—a model that, Robin said, explained lots of cosmic mysteries in one fell swoop and also made striking predictions. My cloth face mask was unable to protect me from Robin’s infectious enthusiasm.

As I listened, I went through the classic stages of reaction to a new Robin Hanson idea: first, bemusement over the sheer weirdness of what I was being asked to entertain, as well as Robin’s failure to acknowledge that weirdness in any way whatsoever; then, confusion about the unstated steps in his radically-condensed logic; next, the raising by me of numerous objections (each of which, it turned out, Robin had already thought through); finally, the feeling that I must have seen it this way all along, because isn’t it kind of obvious?

Robin has been explaining his model in a sequence of Overcoming Bias posts, and will apparently have a paper out about the model soon. In this post, I’d like to offer my own take on what Robin taught me. Blame for anything I mangle lies with me alone.

To cut to the chase, Robin is trying to explain the famous Fermi Paradox: why, after 60+ years of looking, and despite the periodic excitement around Tabby’s star and ‘Oumuamua and the like, have we not seen a single undisputed sign of an extraterrestrial civilization? Why all this nothing, even though the observable universe is vast, even though (as we now know) organic molecules and planets in Goldilocks zones are everywhere, and even though there have been billions of years for aliens to get a technological head start on us?

Traditional answers to this mystery include: maybe the extraterrestrials quickly annihilate themselves in nuclear wars or environmental cataclysms, just like we soon will; maybe the extraterrestrials don’t want to be found (whether out of self-defense or a cosmic Prime Directive); maybe they spend all their time playing video games. Crucially, though, all answers of that sort founder against the realization that, given a million alien civilizations, each perhaps more different from the others than kangaroos are from squid, it would only take one, spreading across a billion light-years and transforming everything to its liking, for us to have noticed it.

Robin’s answer to the puzzle is as simple as it is terrifying. Such civilizations might well exist, he says, but if so, by the time we noticed one, it would already be nearly too late. Robin proposes, plausibly I think, that if you give a technological civilization 10 million or so years—i.e., an eyeblink on cosmological timescales—then either

1. the civilization wipes itself out, or else
2. it reaches some relatively quiet steady state, or else
3. if it’s serious about spreading widely, then it “maxes out” the technology with which to do so, approaching the limits set by physical law.

In cases 1 or 2, the civilization will be hard for us to detect, unless it happens to be close to us. But what about case 3? There, Robin says, the “civilization” should look from the outside like a sphere expanding at nearly the speed of light, transforming everything in its path.

Now think about it: when could we, on earth, detect such an expanding sphere with our telescopes? Only when the sphere’s thin outer shell reached the earth—perhaps carrying radio signals from the extraterrestrials’ early history, before their rapid expansion started. By that point, though, the expanding sphere itself would be nearly upon us!

What would happen to us once we were inside the sphere? Who knows? The expanding civilization might obliterate us, it might preserve us as zoo animals, it might merge us into its hive-mind, it might do something else that we can’t imagine, but in any case, detecting the civilization would presumably no longer be the relevant concern!

(Of course, one could also wonder what happens when two of these spheres collide: do they fight it out? do they reach some agreement? do they merge? Whatever the answer, though, it doesn’t matter for Robin’s argument.)

On the view described, there’s only a tiny cosmic window in which a SETI program could be expected to succeed: namely, when the thin surface of the first of these expanding bubbles has just hit us, and when it hasn’t yet passed us by. So, given our “selection bias”—meaning, the fact that we haven’t yet been swallowed up by one of the bubbles—it’s no surprise if we don’t now happen to find ourselves in the tiny window!

The proposal above, it turns out, is not original to Robin. Indeed, an Overcoming Bias reader named Daniel X. Varga pointed out to Robin that he (Daniel) shared the same idea right here—in a Shtetl-Optimized comment thread—back in 2008! I must have read Daniel’s comment then, but (embarrassingly) it didn’t make enough of an impression for me to have remembered it. I probably thought the same as you probably thought while reading this post:

“Sure, whatever. This is an amusing speculation that could make for a fun science-fiction story. Alas, like with virtually every story about extraterrestrials, there’s no good reason to favor this over a hundred other stories that a fertile imagination could just as easily spin. Who the hell knows?”

This is where Robin claims to take things further. Robin would say that he takes them further by developing a mathematical model, and fitting the parameters of the model to the known facts of cosmic history. Read Overcoming Bias, or Robin’s forthcoming paper, if you want to know the details of the model. Personally, I confess I’m less interested in those details than I am in the qualitative points, which (unless I’m mistaken) are easy enough to explain in words.

The key realization is this: when we contemplate the Fermi Paradox, we know more than the mere fact that we look and look and we don’t see any aliens. There are other relevant data points to fit, having to do with the one sample of a technological civilization that we do have.

For starters, there’s the fact that life on earth has been evolving for at least ~3.5 billion years—for most of the time the earth has existed—but life has a mere billion more years to go, until the expanding sun boils away the oceans and makes the earth barely habitable. In other words, at least on this planet, we’re already relatively close to the end. Why should that be?

It’s an excellent fit, Robin says, to a model wherein there are a few incredibly difficult, improbable steps along the way to a technological civilization like ours—steps that might include the origin of life, of multicellular life, of consciousness, of language, of something else—and wherein, having achieved some step, evolution basically just does a random search until it either stumbles onto the next step or else runs out of time.

Of course, given that we’re here to talk about it, we necessarily find ourselves on a planet where all the steps necessary for blog-capable life happen to have succeeded. There might be vastly more planets where evolution got stuck on some earlier step.

But here’s the interesting part: conditioned on all the steps having succeeded, we should find ourselves near the end of the useful lifetime of our planet’s star—simply because the more time is available on a given planet, the better the odds there. I.e., look around the universe and you should find that, on most of the planets where evolution achieves all the steps, it nearly runs out the planet’s clock in doing so. Also, as we look back, we should find the hard steps roughly evenly spaced out, with each one having taken a good fraction of the whole available time. All this is an excellent match for what we see.

OK, but it leads to a second puzzle. Life on earth is at least ~3.5 billion years old, while the observable universe is ~13.7 billion years old. Forget for a moment about the oft-stressed enormity of these two timescales and concentrate on their ratio, which is merely ~4. Life on earth stretches a full quarter of the way back in time to the Big Bang. Even as an adolescent, I remember finding that striking, and not at all what I would’ve guessed a priori. It seemed like obviously a clue to something, if I could only figure out what.

The puzzle is compounded once you realize that, even though the sun will boil the oceans in a billion years (and then die in a few billion more), other stars, primarily dwarf stars, will continue shining brightly for trillions more years. Granted, the dwarf stars don’t seem quite as hospitable to life as sun-like stars, but they do seem somewhat hospitable, and there will be lots of them—indeed, more than of sun-like stars. And they’ll last orders of magnitude longer.

To sum up, our temporal position relative to the lifetime of the sun makes it look as though life on earth was just a lucky draw from a gigantic cosmic Poisson process. By contrast, our position relative to the lifetime of all the stars makes it look as though we arrived crazily, freakishly early—not at all what you’d expect under a random model. So what gives?

Robin contends that all of these facts are explained under his bubble scenario. If we’re to have an experience remotely like the human one, he says, then we have to be relatively close to the beginning of time—since hundreds of billions of years from now, the universe will likely be dominated by near-light-speed expanding spheres of intelligence, and a little upstart civilization like ours would no longer stand a chance. I.e., even though our existence is down to some lucky accidents, and even though those same accidents probably recur throughout the cosmos, we shouldn’t yet see any of the other accidents, since if we did see them, it would already be nearly too late for us.

Robin admits that his account leaves a huge question open: namely, why should our experience have been a “merely human,” “pre-bubble” experience at all? If you buy that these expanding bubbles are coming, it seems likely that there will be trillions of times more sentient experiences inside them than outside. So experiences like ours would be rare and anomalous—like finding yourself at the dawn of human history, with Hammurabi et al., and realizing that almost everything that will ever happen is still to the future. So Robin simply accepts as a brute fact that our experience is “earth-like” or “human-like”; he then tries to explain the other observations from that starting point.

Notice that, in Robin’s scenario, the present epoch of the universe is extremely special: it’s when civilizations are just forming, when perhaps a few of them will achieve technological liftoff, but before one or more of the civilizations has remade the whole of creation for its own purposes. Now is the time when the early intelligent beings like us can still look out and see quadrillions of stars shining to no apparent purpose, just wasting all that nuclear fuel in a near-empty cosmos, waiting for someone to come along and put the energy to good use. In that respect, we’re sort of like the Maoris having just landed in New Zealand, or Bill Gates surveying the microcomputer software industry in 1975. We’re ridiculously lucky. The situation is way out of equilibrium. The golden opportunity in front of us can’t possibly last forever.

If we accept the above, then a major question I had was the role of cosmology. In 1998, astronomers discovered that the present cosmological epoch is special for a completely different reason than the one Robin talks about. Namely, right now is when matter and dark energy contribute roughly similarly to the universe’s energy budget, with ~30% the former and ~70% the latter. Billions of years hence, the universe will become more and more dominated by dark energy. Our observable region will get sparser and sparser, as the dark energy pushes the galaxies further and further away from each other and from us, with more and more galaxies receding past the horizon where we could receive signals from them at the speed of light. (Which means, in particular, that if you want to visit a galaxy a few billion light-years from here, you’d better start out while you still can!)

So here’s my question: is it just a coincidence that the time—right now—when the universe is “there for the taking,” potentially poised between competing spacefaring civilizations, is also the time when it’s poised between matter and dark energy? Note that, in 2007, Bousso et al. tried to give a sophisticated anthropic argument for the value of the cosmological constant Λ, which measures the density of dark energy, and hence the eventual size of the observable universe. See here for my blog post on what they did (“The array size of the universe”). Long story short, for reasons that I explain in the post, it turns out to be essential to their anthropic explanation for Λ that civilizations flourish only (or mainly) in the present epoch, rather than trillions of years in the future. If we had to count civilizations that far into the future, then the calculations would favor values of Λ much smaller than what we actually observe. This, of course, seems to dovetail nicely with Robin’s account.

Let me end with some “practical” consequences of Robin’s scenario, supposing as usual that we take it seriously. The most immediate consequence is that the prospects for SETI are dimmer than you might’ve thought before you’d internalized all this. (Even after having interalized it, I’d still like at least an order of magnitude more resources devoted to SETI than what our civilization currently spares. Robin’s assumptions might be wrong!)

But a second consequence is that, if we want human-originated sentience to spread across the universe, then the sooner we get started the better! Just like Bill Gates in 1975, we should expect that there will soon be competitors out there. Indeed, there are likely competitors out there “already” (where “already” means, let’s say, in the rest frame of the cosmic microwave background)—it’s just that the light from them hasn’t yet reached us. So if we want to determine our own cosmic destiny, rather than having post-singularity extraterrestrials determine it for us, then it’s way past time to get our act together as a species. We might have only a few hundred million more years to do so.