We’ve read a zillion profiles. Almost all troubled young men. Indistinguishable. There are many, many people who are troubled or go through a tough time. There is no science, surveillance, early warning, AI magic that will reliably identify who’ll pick up the gun to kill. None.

I remember the initial buzz around the fraudalent Surgisphere HCQ papers that made it into The Lancet and NEJM (eventually retracted). Many with data/number/scientific sense *immediately* raised eyebrows, despite also thinking HCQ hype was false and misguided. Others rushed.

There's now a fairly robust post-peer review process, but takes a bit longer when the paper/study is coming from public health authorities and/or is in prestige journals. Running with papers quickly without diving into that process isn't a good idea. Not that anyone will listen.

Wait a bit and let the post-peer review happen and ask around widely is a good idea for anyone not in a position to themselves look at the methods, specifications, statistics, etc., deeply and haven't also read many other papers on the topic. (Not that anyone will listen).

I've seen a lot of reluctance to speak up by senior people who absolutely notice the issues—careers, grants, etc. but also peers and a real fear of being tarred as a minimizer. I think lack of that kind of coherent, senior voices has torpedoed credibility, and our response.

@BenMazer  I get that. Problem is this makes a mockery of public health messaging. Option one, we believe this but ignoring implications. Option two, we don't believe it (because we certainly aren't acting like we do). Option three, we never think about what our findings imply, as policy.

That’s good they’re ring vaccinating.

Always read @kashhill . Many institutions use automated surveillance to decide if a student is cheating on tests as part of remote learning. Often, this means a clunky, unaccountable algorithm barely overseen by low-paid workers abroad has great power.

Flags half-mast. And the reason was not any kind of *well-regulated* militia.

Also, the epicurve is the ultimate reality review. Whatever numbers are being claimed eventually have to show up somewhere— and the bigger the claim, the less likely that bias or lack of surveillance is hiding *all of it.* More numbersense less numberwang would help greatly.