I'm glad on not on the birdsite to read the hot takes, accusations, rejoinders, recriminations, and general nonsense around the DOE's lab leak announcement.
Because people are asking, a few brief thoughts.
First, there is obviously a political element to this. Announcing "I think X with low confidence" is not your usual strategy where communicating new science is concerned.
Just a reminder that the intelligence assessments used to assess Saddam Hussein's potential use of WMDs 20 years ago were also "low confidence".
FWIW here are the definitions of likelihood and confidence that the intelligence agencies apparently use. Note that "most likely" isn't even defined. And, of course, with the large confidence intervals the best answer would have been "we don't know".
@ct_bergstrom That's interesting - I never heard of imprecise probability.
Also interesting after my quick search (and ironic in light of the Iraq WMD assessment) is that the "Ellsberg paradox" played a big part in the development of the theory.
@ct_bergstrom @twitskeptic that’s just a primer on the terms of art. What you’re talking about could very easily be part of how any confidence level could have been used as part of that assessment, but there are lots of reasons a declassified or down-classed report wouldn’t go into the specifics.
@twitskeptic @ct_bergstrom So it is not a "slam dunk" conclusion as George Tenet would say?
@twitskeptic Good point.
It's interesting to think about this mode of assessment in light of so-called imprecise probability (https://en.wikipedia.org/wiki/Imprecise_probability).
DOD and others were instrumental in the work about eliciting probabilities via proper scoring rules etc; I wonder these efforts were extended to imprecise probability and, if so, how we ended up with an instrument such as the one you've linked.