[ExI] Drake Equation Musings

[Anders Sandberg]

On 2016-05-15 19:47, BillK wrote:
> So your paper intends to show that it is quite likely humans are the 
> only intelligent species in the universe? The universe is ours for the 
> taking! Yippee! Though that could be the definition of maximum hubris. 
> But you are facing very big odds. The universe is quite big, you know. ;) 

"Big" is the interesting term here: what is a big universe? If I tell 
you the universe may have 10^10, 10^20 , 10^50, 10^100, or 10^10^100 
stars, when will you say "whoa, that sounds unlikely?" (assuming we are 
doing this from comfy armchairs without actually rushing off to a 
telescope to check). Presumably you will have some feeling that certain 
numbers are so large that they are a priori unlikely. Conversely, if I 
ask if you think the probability of life on an earthlike planet is about 
1, 0.01, 10^-6, 10^-30 or 10^-10^115, you will likely think some of the 
numbers sound unlikely. But why? How would you justify this from the 
armchair? Much of people's prior estimates have far too narrow ranges of 
likely numbers.

The main point of the paper is that a reasonable range for the life 
probability easily runs from from 1 to 10^-100 given what we know (and 
this does not include the rather uncertain intelligence and tech 
probabilities). There are about 10^25 stars in the observable universe. 
So there is 25 orders of magnitude between the sun and the observable 
universe - but if you choose an exponent for the probability of life 
uniformly between -100 and 0, that means that we have just one chance in 
four to get more life than us in the observable universe.

Now, many of us think we know better. But if we think life has a chance 
in between 10^-6 to 10^-2, we need to give evidence for this range. If 
you think my above log-uniform distribution of probability is stupid, 
you need to have an argument for why something less spread out is 
reasonable. And my argument is not that the universe *must* be empty, 
just that armchair astrobiology, when properly done and taking actual 
uncertainty into account, makes it look like the most likely possibility.


> Even if your calculations allow for a few intelligent species, you 
> still face the problem that none of them can be exponential species 
> that have colonised their galaxy, as we see no signs of them. So the 
> suggestion of nano-tech and resource optimised species could apply to 
> one or millions of species, all undetectable by each other. This is 
> more optimistic than saying that *all* technological civilisations go 
> extinct before expanding through the galaxy.

The fact that we do not see primitive aliens nearby nor 
supercivilizations far away serves to update our initial guesses in such 
a way that a past great filter (low probability of life, complex life, 
or intelligence) is more likely than a future great filter (low 
probability of communication ability or longevity). We do not know the 
likeliehood of undetectable civilizations, but our framework can 
actually model the effect on our conclusions for different likeliehoods, 
and generally the effect is that unless you claim they are extremely 
(0.9999... for some large number of decimals) certain the empty universe 
observation still moves the filter guess significantly.

It is a funny paper: it looks like astrobiology, but it is really about 
uncertainty.

-- 
Anders Sandberg
Future of Humanity Institute
Oxford Martin School
Oxford University