3 ms·
As Naval said, https://twitter.com/naval/status/1252185901087592448 https://twitter.com/naval/status/1252185901087592448 > Recent serology studies may be wrong
by walterbell 7y ago
As Naval said, https://twitter.com/naval/status/1252185901087592448 https://twitter.com/naval/status/1252185901087592448
> Recent serology studies may be wrong. But if wrong, they‘re all wrong _in the same direction_.
Finnish: 20-50x undercount
Scottish: 27-55x
Stanford: 50-85x
Italian: 30x
Mass: 17x
Germany: 0.37% CFR
Denmark: 0.21%
> If true, COVID19 has much more spread and much lower fatality.
So we have consistent data showing that the risk is much lower. Could we update our now-incorrect models that were used to justify contact tracing and Palantir social network analysis?
If we don't trust these expensive studies that all point in the same direction, why should we trust Bluetooth? If we don't trust the tests used in the above studies, how can we use those exact tests to forcibly quarantine not only the people who "test positive" but their entire Palantir contact chain?
It would be helpful if we can pick one mathematical dystopia and stick with it. The math-rhetoric dissonance is deafening.
- zucker42 7y agoThe Stanford study does not tell us 50-85x. The results they provided "have no known statistical provenance as of now". https://twitter.com/wfithian/status/1252692365967384576 https://twitter.com/wfithian/status/1252692365967384576
- walterbell 7y agoAs Naval said, if the studies are wrong, they are all wrong in the same direction. A good use of contact tracing would be to point Palantir at the social graphs on each "side" defending or attacking specific virus studies, looking for correlations with industry and other ties. Would be a fun project for social network analysts.