Thursday, 4 February 2016

Luke Barnes (Probably) Uncloaks Even More

It's not every day that you see a cosmologist unload on a historian with "righteous" indignation, but that is what has been happening over at Letters to Nature (initially in January but also stretching into February).  What a fuss about Bayesian probability! – I wonder, what was the likelihood of that?

I see this as further evidence that Barnes is (probably) uncloaking some more because he has drawn attention to the fact that, two years ago, he was arguing with Richard Carrier for an argument in favour of NID, or "non-terrestrial intelligent design".  He didn't do this unknowingly.  Here's an extract from one of his comments:

One more question. I won’t make any claims. I’ll even make it multiple choice. Just give me your main argument against fine-tuning in probability notation.


o = intelligent observers exist

f = a finely tuned universe exists

b = background information.

NID = a non-terrestrial intelligent designer exists.

Here’s the question.

1. In probability notation, what follows about the posterior probability of NID from the fact that p(f | o) = 1?

Multiple choice:

a) p(NID | f.o.b) / p(~NID | f.o.b) = 1

b) p(NID | f.o.b) / p(~NID | f.o.b) = p(NID | b) / p(~NID | b)
(Does this follow from footnote 29?)

c) p(NID | f.o.b) / p(~NID | f.o.b) = p(NID | o.b) / p(~NID | o.b)
(if o is part of b, is this option equivalent to the previous one?)

d) p(NID | f.o.b) / p(~NID | f.o.b) is independent of p(o | ~NID) – the probability that a life permitting universe would exist by chance. Thus, even if p(o | ~NID) << 1, this fact is irrelevant to the posterior probability of NID.

e) Some of the above. Please specify.

f) None of the above. Please answer in probability notation.

So, he (Barnes) is fully cognisant of the fact that his fine-tuning argument is essentially an intelligent design argument.

It's interesting that he should be so keen to faff around with Bayesian probability when his speciality is apparently star formation (or some such).  These sorts of arguments have primarily been the preserve of philosophers, and quite often apologists (and, as a consequence, anti-apologists).


On the topic of Barnes' faffing about with Bayes, he wrote a piece called 10 Nice things about Bayes' Theorem.  I too have dabbled in Bayes' Theorem, primarily because of WLC's abuse of it, which means that it relates to the resurrection thing, rather than the fine-tuning thing.

Barnes had a go at Carrier, in a comment from two years ago, in which he wrote:

* “all prior probabilities are the posterior probabilities of previous equations”.

Learn to use probability terminology correctly, please. Don’t talk about the probability of an equation.

Interestingly, in his 10 nice things piece, Barnes wrote:

Today’s posterior becomes tomorrow’s prior. Hence, Bayesian updating.

This is basically exactly what Carrier was saying, although in the eyes of a hostile witness, he was apparently talking about Pr(equation) rather than the probability that falls out of an equation like Pr(A|B)=Pr(B|A).Pr(A)/P(B) – the probability associated with that equation, perhaps, if not "of" as in common parlance.


Perhaps Barnes just doesn't like to call Pr(A|B)=Pr(B|A).Pr(A)/P(B) an equation which is still strange, because it acts precisely like an equation, with multiplication and division, and in some cases there are additions and subtractions in Bayesian calculations – just like in real equations.  Still, it does sound like he was grasping desperately at straws.

I wasn't that keen to slog through Barnes' most recent attacks on Carrier, but after noting this little own-goal on his part, I'm now a little more intrigued.


Note that Carrier has since replied, here.

No comments:

Post a Comment

Feel free to comment, but play nicely!

Sadly, the unremitting attention of a spambot means you may have to verify your humanity.