Best of Twitter - Week of January 4, 2021
The findings of this study are wild: "Almost every trial" pubbed in 5 high-impact med journals over a 6-week period had engaged in outcome switching. The authors sent correction letters to the journals & most were rejected trialsjournal.biomedcentral.com/articles/10.11… #bioethics #researchethics
This matters b/c whether you use 1947 as a start date or 1973 (the supposed date of divergence) there's no decoupling of pay and productivity. I use the NFC sector rather than broader corp. sector because there's no hours worked series I can find for the former.
My default position is that lot of noticeable changes or peculiarities in long-run historical series are about the series themselves and not the real world.
Donald Schneider @DonFSchneider
^ this is very underappreciated i think. esp comparing things over time, they're measured by different people with different ad hoc choices, individual knowledge, unknown unknowns factors to make lots of comparisons of these numbers in 1970 vs 2020 totally meaningless imo. maybe if not meaningless per se, but introducing such error bars as to make e.g. comparing 2% vs 3% growth in GDP 50 years apart effectively meaningless
once we encounter an argument that seems overwhelming, we lose the interest in the conversation
reddit.com/r/slatestarcod…
Heh, this is *exactly* why I ended up in Japan.
“No future for engineering in the US. The WSJ said so and they would know.”
Lee Edwards @terronk
So turns out maths + philosophy twitter have totally missed a thing where a guy with a billion Youtube followers was accused of cheating a Minecraft speedrun with a subtle luck modification mod and now all the kids are trying to figure out what "frequentism" and "Bayesianism" are
^ Did Dream Fake His Speedruns - Official Moderator Analysis
8.5 years ago I was training restricted boltzmann machines in Matlab on CPU on my machine below the desk.
Quarantined section
This is, sadly, a common but truly mistaken point of view. Not all vaccine candidates work; not all of them have worked in this pandemic. Some of the very similar post-2003 SARS vaccine candidates actually made things worse.
Clinical trials are *essential*.
John Cochrane @JohnHCochrane