Hacker News Reader

I read it as saying something a little more subtle, which is that there is a lot of uncertainty in a couple of our critical estimates, and given that uncertainty what we see (no other life in the galaxy) is not that surprising. (That is, if you take the mean and multiply, you get misleadingly-high intuition.)

Of course if we do some more experiments and find that in fact the probability of abiogenesis is better approximated by 10^{-3,-2} that would change the expectations dramatically. This happened recently: we now have much better science around the incidence of earth-ish planets in solar systems, and a better understanding of what seems to contribute to good life-favorable (at least as we understand it) conditions on those planets. So this uncertainty can change with good work.

It's just that at this point it's really quite fair (per the authors) to expect that with more research we might equally well discover that P(abiogenesis) < 10^-100, in which case it would be astounding if we weren't alone in the observable universe.

(Good talk slides: http://www.jodrellbank.manchester.ac.uk/media/eps/jodrell-ba...)

This was helpful. That wasn't the way that I was reading the original paper, but I think I see what you're saying.

There's one remaining subtlety that I don't see in his comment or in almost any of the comments in this thread, and it's the entire novel point of the paper.

If you give any particular term in the Drake Equation a value of some magnitude, say, P(abiogenesis) < 10^-100, then you're still treating it as a point estimate, and this is the flaw at the heart of all arguments about the Fermi Paradox.

What the authors do instead is give each term of the Drake Equation a range of magnitudes. When you multiply those ranges of magnitudes together, you get a ton of different possible outcomes. If you then look at those outcomes together, you find that in over half of them, we are alone in the universe.

This is the "multivariate probability" that a top comment referred to, that seems to be problematic for a lot of people (including me).

The authors point out that with this approach, even if we nail down the range of probabilities for some of the terms, the uncertainties of the remaining terms will still lead towards a > 40% chance of being alone in the universe.

It's a little bit like: in a poker game, I draw five cards. Each of my cards is a 2, 3, 4, 5, or 6. What is the probability that I drew a straight? To calculate that, you'd count the number of different possible card combinations, and then count the number of those combinations that resulted in a straight.

What everyone has been doing previously in this analogy is calculating the odds of drawing each individual card.

In the case of the Drake Equation, using a probability distribution approach to estimating the chances of being alone in the universe leads to a very different result than we expect.

I hope I didn't add to the confusion.

> If you then look at those outcomes together, you find that in over half of them, we are alone in the universe.

Since we do not really know, anything but P=0.5 should imply incorrect prior estimates. That might be more insightful than whether alien life exists or not.

I find the notion that we have enough information to even guess at where life is let alone that we can assume there is none downright laughable to the point of obscenity. We know a lot less about this universe than it takes to make even sensible guesses as to generalizations about the nature of life in the universe.

Our best-effort guesses are literally the best we can do. Pretty much by definition, dismissing them because "we know so little" means taking a view that's worse than that guess.

> literally

excuse me, but no. Our best-effort guesses might be the best we could literally do. But there is no need to be that literal, or literal at all. As long as the uncertainty range includes 0.5, it's OK, but merely a test of how much we know as a fact as opposed to intuitively. And at that, the equation can surely be blown up further. Why would it be optimal?

My thoughts exactly, I always think: perhaps in 10 years we find an alternative to radio based communication which is far superior and then we find the galaxy teaming with communicating life forms... Or not. Who's to say? This whole thing seems to me an exercise in quantifying unknowns. We don't know the size of the categories we don't even know we don't know anything about. This can still be a fun exercise of course and it can give you an idea of the assumptions to prove or disprove but still... the value (beside the entertainment value) is pretty small imho.

Yep. Or maybe we’re all wrong about quantum mechanics or any of a number of theories. There’s a huge portion of the universes mass that we can’t explain. The unknowns are far too large to extrapolate.

Wait, no, I just actually look at the blog and the article. And ultimately, the article is saying more or less the same as the GP but with more detailed math. And blog essentially admits this.

The blog kind of references a situation where you'd have a high variance distribution but doesn't really describe it. There isn't any argument against the claim that the chance of a civilization on a given planet is independent of the chance of life on all the other planets in a given galaxy (modified by star-type but not by whether there's life nearby).

From the original paper:

"In this example, the point estimate for each parameter is 0.1, so the product of point estimates is a probability of 1 in a billion. Given a galaxy of 100 billion stars, the expected number of life-bearing stars would be 100, and the probability of all 100 billion events failing to produce intelligent civilizations can be shown to be vanishingly small: 3.7×10 ^−44

[...]

However, the result is extremely different if, rather than using point estimates, we take account of our uncertainty in the parameters by treating each parameter as if it were uniformly drawn from the interval [0, 0.2]. Monte Carlo simulation shows that this actually produces an empty galaxy 21.45 % of the time."

I take this to mean that collapsing a range into a single point will artificially inflate that point, and instead you need to look at the number of possible outcomes, not the average of all outcomes.

From the related blog post:

"Imagine we knew God flipped a coin. If it came up heads, He made 10 billion alien civilization. If it came up tails, He made none besides Earth. Using our one parameter Drake Equation, we determine that on average there should be 5 billion alien civilizations. Since we see zero, that’s quite the paradox, isn’t it?"

That seems like a consistent reading to me. Maybe I'm wrong - I take it you would draw a different conclusion?

Edit: No, thinking about it more and reading some other posts, I kind of get what you're saying. I'm still not really sure I'd call that a paradox though. It still seems to suggest to me that we may not need to rely on any of the really big theories to describe why we don't see life.

I agree that Fermi isn't a paradox but if the chance of an advanced civilization on a given star is independent from the chance of one on another, then the components of Drake's equation exist and to have the chance be small, the components must be small, for suitable definition of small.

The article and blog's meta-argument involves choosing components over a random distribution of universes and saying "well, there's a good chance you wind-up in a universe with small". But reasoning about random universes is debatable, introducing unnecessary metaphysics but even so, if you wind-up in a universe with a small chance of an advanced civilization, you will wind-up in a universe with Drake's components also small. Which is to say the whole reasoning does nothing but "muddy the waters".

The only way the "randomly chosen universe" argument matters is if we're jumping from universe to universe or if there's a force that inherently keeps us from reasoning about the why of the Drake components - but neither of those conditions are met.

"Trump won the election because we fundamentally don't understand how democracy works in practice", and "Trump won the election because he had a 1 and 5 chance of winning, and sometimes 20% chance events happen."

This is a great point. Still, if you say "we don't understand democracy because we don't understand that 20% chance events happen" then it seems like the points converge.