Defence of the Fine Tuning argument

One common objection to Robin Collin's fine tuning argument is that we are not justified in assuming a uniform probability distribution of universes. And so even if the set of life permitting universes is but an unfathomably small subset of all possible universes, for all we know the dice are loaded and it's not in fact all that unlikely that a life permitting universe would come to exist naturally. A thought experiment may go along with this point:

When flipping a coin there are three events that could occur; the coin lands heads up, it lands tails up, or it lands perfectly balanced on its side. If we were to simply assume a uniform distribution, it would follow that the likelihood of a coin landing on its side is 1/3; orders of magnitude greater than it really is. We know this, of course, only by our past experiences of coin flipping. But with the fine tuning argument we have experience with only one universe, so we shouldn't hastily assume that all other possible universes are more or less as likely to have existed as our own.

But there's a problem with this thought experiment; we have to make assumptions about the probability distribution to even draw conclusions from the trial. For example, suppose we flip a coin six times and get:

H, H, T, H, T, T

From the six flips we got three heads and three tails. And so we would want to say the trial suggests that Pr(H) = Pr(T) = 0.5, while Pr(S) = 0. But wait, we've assumed a uniform distribution to do so. Unless Pr(HHTHTT) is equal to the probability of any other possible combination of six coin flips, your assessment is going to be skewed. In other words if the event HHTHTT was actually an anomaly (maybe the coin is weighted), then we're not justified in drawing any conclusions about the probability of H, T, or S.

Okay, so maybe we want to perform an experiment to see if the coin is fair; to see if our previous experiment should be considered reliable. And so we flip a coin 30 times, getting;

A: HTHTHH

B: THTHHT

C: HTTHTH

D: THHTTT

E: THTHTH

In trial A heads was flipped 4/6 times, in B 3/6 times, in C 3/6 times, in D 2/6 times, in E 3/6 times. Average that up and we get 1/2. Cool, our experiment seems to be confirming what we thought we already knew. But wait a second... How do we know Pr(A) = Pr(B) = Pr(C) = Pr(D) = Pr(E) = any other possible outcome? Uh oh..

You should see the point; no matter what we will always end up assuming a uniform probability distribution where we haven't yet been able to run trials. So the fact that we only have experience with one universe shouldn't give us warrant to suppose the probability distribution of possible universes wildly favours those which permit life. Obviously the first trial was sufficient to judge the probability of H, T, and S even though we hadn't yet determined whether or not the coin was weighted, and the second trial was sufficient to judge that the coin wasn't waited even though we hadn't ruled out any mystical forces that could be further affecting the experiment. Because that's just implausible: that some unknown factor would be throwing the experiment off so dramatically. And yet that seems to be exactly what the original objection amounts to. That, for all we know, some unknown factor is throwing off the probability distribution so much that, while the range of life permitting universes is but an unfathomably small sliver of all possible universes, it's still not very unlikely that one would exist. This sort of skepticism is just untenable, and doesn't offer a viable objection to Robin Collins' fine tuning argument.