onsdag 15 juni 2016

The New York Times claim about extraterrestrials was pulled out of thin air

"Yes, There Have Been Aliens." So reads the spectacular headline of an article about the possible existence of extraterrestrials by astrophysicist Adam Frank in the New York Times this Sunday. If the headline's suggestion - that science has established that extraterrestrials exist or at least have existed - is warranted, then we are faced with one of the greatest scientific breakthroughs ever, with profound consequences for our collective human self-image. But no, no such breakthrough has been made (or is on the radar), and the headline is deeply misleading.

Sometimes, overly enthusiastic newspaper editors play their contributors a trick by assigning a headline that goes far beyond the claims that are made in the article itself. Perhaps that is the case here? Perhaps professor Frank is the innocent victim of just such an editorial prank? Well, no. Frank has only himself to blame, as the headline's claim is made loudly and clearly in the text of his article, where we can read that
    given what we now know about the number and orbital positions of the galaxy’s planets, the degree of pessimism required to doubt the existence, at some point in time, of an advanced extraterrestrial civilization borders on the irrational.
This is a very strong claim, but it is plain false. Frank bases his claim on a recent joint paper by him and astronomer Woodruff Sullivan in the journal Astrobiology, where the claim about the very likely existence of extraterrestrials at some time in the past is derived from an assumption about how a certain probabilistic quantity p relates to a certain threshold c. It turns out, however, that this assumption was simply pulled out of thin air. Consequently, the same verdict follows for the claims in the New York Times article. Let me explain.

The p I'm talking about here is the probability that a randomly chosen potentially life-bearing planet eventually gives rise to intelligent life with a technological civilization on the level of present-day humanity. Here is an essentially correct statement in Frank's New York Times article:
    But what our calculation revealed is that even if this probability is assumed to be extremely low, the odds that we are not the first technological civilization are actually high. Specifically, unless the probability for evolving a civilization on a habitable-zone planet is less than one in 10 billion trillion, then we are not the first. [italics in original]
One in 10 billion trillion is 10-22, which is Frank's choice of threshold c.1 Combining the two qoutes from the article, we see that Frank means to say that doubting that p≥10-22 "borders on the irrational". Well, if Frank is fond of the idea of rational argument, then he should of course have backed up this claim with some strong evidence that p≥10-22, but he offers no such thing, neither in the New York Times article, nor in the Astrobiology paper.

The probability p is one of the three central parameters in the so-called Great Filter formalism, which is a superb framework for addressing the Fermi paradox and the notorious "Are we alone?" question, and which Adam Frank and Woodruff Sullivan would be well-advised to study.2 The other two parameters are N and q, where N is number of potentially life-bearing planets in the universe (a quantity discussed by Frank and Sullivan in the light of recent advances in the observation of exoplanets), and q is the probability that a randomly chosen civilization at the level of present-day humanity goes on to develop an intergalactic technological supercivilization visible to astronomers all over the universe. Of these three parameters, N is the only one whose order of magnitude we currently have a reasonable grasp of: it is a very large number, somewhere in the vicinity of 1022 (give or take an order of magnitude or so). In contrast, we are very much in the dark about the whereabouts of p and of q (other than the fact that they are probabilities, so they must be between 0 and 1). But there is one thing we pretty much know about them combined, namely that the product pq must be tiny, because otherwise Npq would have been a large number, and we would most likely have been able to see signs of a an intergalactic technological civilization out there, which we haven't. And if the product pq is microscopic, than at least one of the factors p and q must be microscopic. But neither of them is by any means obviously microscopic. Whether it is p that is microscopic, or q, or both, is a wide open question as science currently stands. As to p, it might be 0.9, it might be 0.1, it might be 10-10, it might be 10-30, or it might be 10-100 (or something else). None of these values is (notwithstanding Frank's claim that the last two would "border on the irrational") implausible.

There is no shortage of candidate bottlenecks in the evolution of life that might make p microscopic. Biogenesis is an obvious example, and Hanson (1998) lists a few more: the emergence of prokaryotic single-cell life, of eukaryotic single-cell life, of sexual reproduction, of multi-cell life, and of tool-using animals with big brains.3 Any claim that p cannot plausibly be microscopic needs to come with a demonstration that none of these candidate bottlenecks is sufficiently severe and uncircumventable to account for p being microscopic. Preferably, such a demonstration should also be weighed against the available evidence for q not being microscopic (see in particular Armstrong and Sandberg, 2013). Frank and Sullivan offer none of these things.


1) This seems to be a correction compared to the original Frank-Sullivan Astrobiology paper, where c is of the order 10-24. Their calculation of c invloves arbitrarily and confusedly throwing in an extra factor 0.01 in what was probably an attempt to make the estimate scientifically conservative, but whose effect is in fact the opposite.

2) See, e.g., Chapter 9 in my book Here Be Dragons: Science, Technology and the Future of Humanity, or better yet, see Robin Hanson's seminal paper on this topic. Or see my recent paper with Vilhelm Verendel in the International Journal of Astrobiology where we approach the Great Filter from a Bayesian point of view.

3) A tempting reaction, when first confronted with the task of estimating p, is to say something like "Hey, we exist, we evolved, here on Earth, surely that's an indication that p is probably not so small?". I suspect that such reasoning has influenced much discussion about extraterrestrial life and the Fermi paradox over the years, even in cases where it is never spelled out explicitly. However, it is probably not a valid argument, because a low-p and a high-p universe share the feature that everyone in it will find themselves to exist and to have evolved, whence that observation cannot be used to distinguish a low-p universe from a high-p one.

2 kommentarer:

  1. Here's another formulation from Frank's NYT article: "While we do not know if any advanced extraterrestrial civilizations currently exist in our galaxy, we now have enough information to conclude that they almost certainly existed at some point in cosmic history."

    This is just dead wrong.