Read on Twitter
1/ @RapportSA published a front page article this Sunday saying the lockdown could extend for 4months. It& #39;s based on a report by BCG, that& #39;s already outdated (published 25 March). BCG explicitly state it& #39;s a "beta" model that& #39;s not meant for publication or public dissemination
2/ I did some digging & found the report. In the interest of just showing how badly Rapport stretched to use this report, even though they never should& #39;ve (see pic), I& #39;ll just post slides from South Africa & brief intro on their model.
3/ Their timing estimate is primarily based on what China did along with country specific factors. They then calculate either 8-12 weeks after peak new cases based on effectiveness. But how do they calculate peak cases?
4/ They apparently use a standard epidemic curve (can& #39;t really comment on this, & #39;cause hey not an epidemiologist). Looks okay to me.
5/ But check their slide for South Africa. It& #39;s already heavily outdated. According to this we should& #39;ve have 1000+ new cases per day by now. So obviously you can& #39;t use this data anymore.
6/ Everything is moving at a rapid rate & reality doesn& #39;t quite line up with models. That& #39;s why you adapt & keep modeling. I don& #39;t blame BCG at all here. The lockdown timing seems arbitrary to me though. So mileage may vary if you think that every lockdown should follow China& #39;s
https://abs.twimg.com/emoji/v2/... draggable="false" alt="🤷‍♂️" title="Man shrugging" aria-label="Emoji: Man shrugging">
7/ But a quick glance at these slides would& #39;ve shown that doing any reporting on this is basically useless & frankly in my mind irresponsible. EVEN when BCG explicitly stated it& #39;s a beta model. Again, read the CAVEATS
8/ Who knows how returning to normal looks like. It& #39;s likely we are stuck with restrictions for a while. Fear mongering & clickbait sells. But just use common sense.
For now, keep staying home so we can keep proving the models wrong.
For now, keep staying home so we can keep proving the models wrong.
