It is assumed in the premise of the argument, but there is evidence of it:
http://en.wikipedia.org/wiki/Fine-tuned_universe
The premise of the fine-tuned universe assertion is that a small change in several of the approximately 26 dimensionless fundamental physical constants would make the universe radically different: if, for example, the strong nuclear force were 2% stronger than it is (i.e. if the coupling constant representing its strength were 2% larger), diprotons would be stable and hydrogen would fuse into them instead of deuterium and helium. This would drastically alter the physics of stars, and presumably prevent the universe from developing life as it is currently observed on the earth.
-Bri
The only way I can make any sense of fine-tuning in Bayesian terms is this:
Using the wiki example: let's assume the strong nuclear force could have been 50% stronger or weaker than it is, and that +/- 2% is the range of values that "fine-tune" for life. Then the odds that the value would be so fine-tuned randomly are 2/50 = .04. Further, let's assume that the strong nuclear force is typical of the 26 constants. Then the probability that all of them would be fine-tuned is (.04)
26 = 4.5 x 10
-37.
Now what is this probability? Is this the chance there is life in the universe? No, this is the chance the universe will be fine-tuned, given our assumptions. For P(H|E) = P(E|H)P(H)/P(E), if we want to make E = evidence of life, then we should make H = hypothesis that the universe is fine-tuned. Then we can reasonably argue P(E)
< P(H); since we have no idea how difficult it is for life to arise, except for our assumption that it certainly requires fine-tuning if there is to be a chance at all.
Continuing, as we're otherwise ignorant about life arising, we make P(E) = 1/2(P(H)) = 2.25 x 10
-37. P(E|H) then is the probability of life given fine-tuning, which we just answered in assigning P(E) a value to correspond to our ignorance: P(E|H) = .5. Running the numbers:
P(H|E) = P(E|H)P(H)/P(E) = (.5)(4.5 x 10
-37)/2.25 x 10
-37 = 1;
which when translated back into English yields the rather unstupendous result that if we assume the universe is finely-tuned, and admit we are ignorant about how difficult it is for life to arise even with fine tuning, then finding life in the universe means it is certain the universe is finely-tuned! In other words, if we assume fine-tuning, then just by being alive [in order to assume fine-tuning] we prove fine-tuning! Wow! What a marvelous proof! But wait, as if that weren't miracle enough, for the true
coup de grace de dieu, if we go one HUGE eschatological leap further and assume fine-tuning can only be the work of god, then we can state confidently that
if we assume god, then by being alive to assume god... we prove god! Fantastic! Wonderful! Who ever would have believed such a feat was possible with mere logic and statistics. Call the papers, the UN, the Vatican, call my mother-in-law!
Thus it seems all the Bayesian F-T argument ever reflects is the bias of our prior and posterior assumptions: one giant fudgesicle factory, and god the dairy queen.