Great question, that I don't know the answer to. The answer to that question is part of the uncertainty that attenuates the signal I described as "weak". I think the reason I like the approach I'm describing is that you can model uncertainty at every level of knowledge: you don't need to become an epidemiologist to reason about the facts (with correspondingly higher error bars than a domain expert would have), _and_ you don't need to blindly trust that experts in every domain are somehow magically optimal when it comes to reasoning under uncertainty.
To put it another way, that's the kind of information I would spend the time digging into once knowing the answer would push my model across a decision threshold. I look forward to seeing the follow-up research, which is usually the quickest way to fill in gaps in the "primitives".
To put it another way, that's the kind of information I would spend the time digging into once knowing the answer would push my model across a decision threshold. I look forward to seeing the follow-up research, which is usually the quickest way to fill in gaps in the "primitives".