Physicallity and Undetectibility

[Angrboda]

I've always had an issue with miracles, which is similar to other problems in assigning probability to supernatural explanations. With things like the resurrection or any of the stories of miracle men from Appollonius on there is always the problem that the reasoning toward the most probable explanation must include all alternate explanations, including fraud, invention, transmissional distortion, mistake, misinterpretation, and so on. These are all events that obey natural law, yet most often the probability of them happening can be assumed to be manyfold higher than non-natural explanations (there's a question of the base-rate of non-natural phenomenon in figuring prior probability, but for simplicity's sake, that discussion is omitted). So even if the resurrection actually occurred, the probability of these other explanations is higher, and it is irrational to believe in an explanation of considerably lower probability when an explanation of greater probability is available.

This enters into the explanation of miracles in several ways. By most accepted definitions, a miracle has to satisfy two conditions: a) it is unexpected according to natural law, or even contrary to it, b) it is caused by an agent (in this case, a god). People tend to focus on the first part and neglect the second part, that it's not a miracle unless it is caused by God. Even if I take away all the alternative explanations mentioned above which are attributable to human deceit or error, that still leaves the explanation that the event has a natural explanation, but we just don't happen to know what that explanation is. The history of science is replete with events that lacked a natural explanation at one time, yet later were explained; so the prior probability of an occurrence of a natural event without an explanation is reasonably high. In this sense, and because we can't directly track a miracle back to an, essentially non-existent agent (from the view of physics), any actual miracles that occur will, rationally speaking, always fail the test, because they occur in the "epistemic shadow" of more reasonable explanations, and are pushed further into darkness by an inaccessible causal history.