Regardless, it's a really slick method that could be used effectively. Just might require some deep thinking to avoid biases.

This bias could be small, particularly for lower R0 epidemics in larger populations. And there should be ways to ameliorate it at a cost of tracking more events. Unfortunately, any solution will allow reseeding the epidemic thus enabling the CI to cross 0.

Am I missing something? From my reading, counterfactual susceptibles generated by removing transmisison will behave dynamically like a recovered individual. They can not be re-infected in the original trajectory--effectively depleting susceptibles in the counterfactual and biasing estimates.

Got a number in mind?

A new pre-print by @lshtm.bsky.social colleagues surveyed 13,000 participants, and found it's about 9 per day in the current era.

It also looks at how social mixing varies with age, ethnicity and social economic status. Link to paper:

something, something, niche construction, something, something

Agreed, "longevity" was the wrong word for me to use, because AI is here to stay in some iteration.

I think AI will be truly transformational regardless of what's under the hood. Specifically the magnitude of societal adoption will be transformational, even if the model isn't.

I think AI's longevity hinges on our perception of how transformative it is, not its actual utility.

Clearly, AI does some things very well and some things very poorly. People differ in whether they see the successes as the edge cases or the failures as edge cases.

I was always impressed by how cleanly the empirical results matched dynamical systems theory in this work from Jeff Gore's lab about critical slowing down before population collapse: www.science.org/doi/10.1126/...

Solid. I now realize I might've just misinterpreted the plot because the blue line does not span the entire domain. Since it met the red perfectly visually I thought it was plotted over at later time points (which in retrospect wouldn't make any sense).

You're creating, and I'm just destroying :-\. Appreciate it all, especially your book

Looks like it's after 365 days too

Seems odd blue and red lines merge. Any sense what's up?

Not the most visually appealing, but I’ve found dot plots (eg, via mummer/nucmer) adequate

Check out our preprint: www.biorxiv.org/content/10.1...

Caveat, the results in the paper are underpowered because a lack of funding to sequence more. The figures are fine though

26 households too, so that variance is probably up there, but not shown

This is certainly interesting… and in Utah. Shifting demographics between groups likely explains the curious nonlinearity.