Thanks for the welcome. I understand that Hard Rationality is often not readily applicable and you are tempted to make do with what you have, such as in forensics.
The issue I have with this approach is that “the most effective means available” is not really effective, unless the issue is either clear-cut (shoelaces are either tied or untied) or not very important (what’s the worst that can happen if wrongly reflected light makes you mistakenly believe that your shoelaces are tied?).
My concern (summarized in the last paragraph of my original comment) is that people naturally and subconsciously gravitate toward collecting evidence (comparatively easy) instead of building testable models (hard). This issue probably deserves a separate thread, unless it has already been discussed, in which case I’d appreciate a link.
Presumably no definitive (i.e. testable/falsifiable) evidence either way is originally available, so the whole argument appears to be rather anti-rationalist.
A better approach would be to set personal prejudices and idle musings aside and to attempt to construct a model that can be tested (for example: what Sylvanian-made tech would be capable of diverting asteroids? would it leave residue at the impact site? were there sudden and unexplained expenditures at the time the tech would have been designed/activated? what other artificial and/or natural events could have affected the strike timing/location?)
If one remains content with upshifting/downshifting probabilities, s/he is distracted from the only task that makes sense: building a testable model and testing it.