Does modern physics state that air molecules can randomly whizz around in such a way to cause a voice emanating from the clouds that is clear and coherent?
Is this possible or not? Is this extremely improbable or is it possible?
The motivation of the question comes from Dawkins where he talks about how a human statue could theoretically move in a way that doesn’t contradict the laws of physics. He just says it is incomprehensibly improbable.
Which is correct?
The thing about thermodynamics, is it's difficult to get the scale of improbability. Say you want to work out how often in a 1 litre bottle of air, it will all sponteneously gather on one side, leaving a vacuum on the other. On average, that won't happen once in the age of the universe. And that's very simple, compared to say forming clear letters in clouds, nevermind a booming voice with something interesting to say from the clouds. Discussed with more detail here: How improbable does an event have to be before we can say it didn't happen by chance?
Another thing that is hard on our intuition, is that however unlikely individually, incredibly improbable events do sometimes happen. For them to come out a very particular way at a particular time, vanishingly unlikely. But, it seems from the lack of evidence for life in our galaxy, that the emergence of life requires a series of unlikely steps to begin, then favourable conditions in order to continue and develop. For it to happen elsewhere and produce even quite similar outcomes, is very unlikely (although convergent evolution and similar ecological niche demands will likely impose some constraints). But for it to happen somewhere in the universe, a virtual certainty.
The confounding factor is, if the Many Worlds Interpretation of quantum mechanics is correct, because then however unlikely there is some timeline branch where all possible outcomes happen - though if we cannot observe or interact with them in any possible way, you can argue they are metaphysical, rather than scientifically real in the normal sense. Quantum Immortality is a nice example of the implications of this, that within the average universe there is some timeline where we get to live forever, so we might expect our own personal experiences to violate expected averages, if MWI is true. That is, Schroedingers Cat always experiences the universe where it didn't die. If so, our subjective experience of living a statistically ordinary life, can be argued as evidence against the MWI.
The real question though, is what import we give to unlikely events. If a booming voice gives whole sentences from the sky, we could dismiss it by saying however unlikely it's possible that it's a result of an atmospheric phenomenon, or we could completely rearrange our lives to fit whatever it said. Both of these extremes are pretty irrational, and Bayesian inference can be used to show if that happens, the likelihood is it's niether fluke nor a cosmic mind that made the universe but doesn't like leaving any clearer evidence it exists.
David Hume made the broader case against inferring some unusual outcome means the laws of physics have been suspended:
"No testimony is sufficient to establish a miracle unless it is of such a kind that its falsehood would be more miraculous than the fact that it tries to establish. And even in that case there is a mutual destruction of arguments, and the stronger one only gives us an assurance suitable to the force that remains to it after the force needed to cancel the other has been subtracted."
-David Hume, from the chapter 'On Miracles', in An Enquiry Concerning Human Understanding
If we keep rolling 6s on a die, or getting heads on a coin flip, as the number of times increases, the unlikelihood of it being by chance rises exponentially. So it is far more reasonable to conclude the die or the coin are biased, the more unlikely the sequence becomes.
Fundamental physics is almost all reversible, events could happen in either direction with no preference. There seems to be a link between the thermodynamic Arrow of Time, the (geometrical) spreading out of information, and memory. It may be that events are happening forwards and backwards in time, but our creating memories involves spreading information into our brains, so we can only lay down memories of the forward direction of time. Discussed here: How does entropy explain consciousness and the forward direction of time? That increase of entropy 'directs' what changes are likely, could rather than being funamental instead only be a subjective necessity.
If time is emergent in this way, which approaches like Loop Quantum Gravity seem to imply, there could be a kind of reinforcing of likely events even in MWI, because the unlikely low-entropy timeline wouldn't have to happen just once, but would have to stay stable with a kind of bouncing backwards and fowards.
The Anthropic Principle and Fine Tuning are another example of the reasoning behind Quantum Immortality, where an unlikely event is a necessary precursor to experience. So even in multiverse with emergent time a kind of 'fishing' for unlikely possibilities that enable complexity is possible. Discussed in more detail here: How can nature without self-awareness and intelligence create living beings with self-awareness and intelligence?
If you go to the PhysicsSE and search for the word "possible," you get over 3,000 results. In quantum physics, there is also a distinction between not just actual and possible but actual and virtual excitations of fields.
Moreover, there is a (family of) modal interpretation(s) of quantum physics:
Fletcher[19] is an overview of nomological modality; see also Szabó, et. al. [20].
The "upshot": so if you are asking about the concept of possibility roughly in general, and how this concept is applied in physics, then in some sense physics doesn't entail that "everything is possible." But what is this "everything," then? We might say "everything that is not self-contradictory," but we will then want to parse the difference between "possible independent on the actual past" and "possible, given whatever the past has turned out to have been." And that latter will rule out possibilities just in case we press hard enough on the "necessity of the past" parameter. In abstracto, though, one might imagine that there is a dedicated temporal quasi-particle field, so that there are virtual temporal quasi-particles that can tunnel not only through space but also time, perhaps pastwards, so as to alter the past. Our imagining, here, will then be possible concretely if the past is not necessary in the required manner.
But so if one is asking, "Does physics allow that any x is possible just so long as x has a nonzero chance of happening?" then all such x's will be possible, assuming that nonzero probability that... implies possibility that..., or then that probability logic is embedded into modal logic sufficiently.
