Posted: Jan 06, 2015 9:58 pm
Thommo wrote:You can look at it that way, informally.
Think of probability like science - it's only meaningful as a model of something. We need some assumptions, which people are a bit lax at spelling out.
Formally what you need is some kind of random variable on a sample space. The argument you gave is flawed in the same way that actual fine tuning arguments are flawed - it makes assumptions about the sample space which we have no reason to make.
The fine tuning argument artificially assumes that the constants found in physical formulas (or their ratios), relating to gravity, electromagnetism can vary from what they are and that they do so with fixed probability over some interval. We don't know that they can vary, we don't know that there is anything in reality that represents the mathematical process of imagining universes being generated at random by some unknown mechanism with varying constants, although if there actually is such a process that would probably indicate a "multiverse" anyway, which the argument just ignores.
Similarly your argument is assuming that natural laws "could" generate non life sustaining universes, it makes similar assumptions about the sample space, then makes an ad hoc declaration comparing it to an assumed sample space of universes that "could" be made by a god. You will end up with weird (and untrue) conclusions because of the ad hoc nature - for example that it's "evidence" against the hypothesis that god created the universe, but not against the hypothesis that god created the universe with the intent of it sustaining natural laws (since this also would only ever 100% result in universes that sustain life, but appear to have natural laws).
In both cases this is because there is no "fitting" of the sample space to reality. There's no physical process to match the "probability" when there needs to be. It's ok to model a dice with a uniform probability on {1,2,3,4,5,6} because we can test it. What do we do if the model doesn't match that dice? We toss out the model and use a different probability distribution to reflect that the dice is "unfair", "biased" or "weighted". It's basically meaningless to talk about probability without modeling or fitting.
This is different from what I wrote above, you're basically taking the view that the question of fine tuning is meaningless since we don't even know whether the laws and constants could be any different from what they are.
Okay, I agree, but then the argument isn't actually logically invalid in the sense that the conclusion doesn't follow from the premises, it's just that there's an unstated foundational assumption (that they could be different) we don't know whether is true. But then it merely requires us to state that premise openly as an assumption, like:
P1: We assume the laws and constants of nature could be different.
P2: ...
?