The instinct-log item from last entry is no longer a footnote. The Wuhan cluster has a name (a novel coronavirus, “2019-nCoV” provisionally), confirmed human-to-human transmission, a first US case (a traveler in Washington State, confirmed Tuesday), and, as of yesterday, the unprecedented sight of China locking down Wuhan, a city of eleven million, plus neighboring cities. Airports are screening; the WHO is convening; the epidemiology Twitter accounts I started following two weeks ago (the archive’s instinct pays for itself again) are doing R0 estimation with error bars that should alarm more people than they’re alarming. I’m not a doomer by temperament, seven years of this blog is mostly me telling people the sky isn’t falling, but I know what exponential growth looks like on a dashboard before the on-call believes it, and I know that “contained” is a claim that requires instrumentation the world visibly lacks. Filed with maximum uncertainty and minimum comfort: watch this one like it’s a sev-2 that hasn’t paged yet.
Ordinary January continued regardless, as it does: platform team week three shipped our first paved road (standardized service scaffolding, new service bootstrap time from three days to forty minutes, the canary lesson industrialized); Davos discussed “stakeholder capitalism” with a straight face; and the Champions League draw sets up a Real Madrid vs Manchester City Round of 16 clash, Pep’s passing machine versus Zidane’s knockout-stage magic, the kind of matchup the group chat’s win-probability discipline was built for.
TIL: R0 and dispersion (k). Average spread rate matters less than variance in spread (superspreading events dominate outcomes). Epidemics are hot-partition problems. I wish I found that less clarifying than I do.