Chemistry Nobel laureate uninvited from biodesign conference
twitter.com
twitter.com
Those trying to silence discussion(s) are really the ones who should be uninvited (the committee members).
He's not being silenced; he can say whatever he wants, but the conference isn't obligated to lend him creditability by making him a speaker.
I have a few friends working on startups/solutions to this (non-sensationalist reporting, fact-checking apps, etc.), but I personally think the cat's already out of the bag.
(I actually don't even agree with his ideas about covid)
Reading Giorgio Agamben, one can see that he was completely devastated by the sight of absolute cowardice and obedience (I don't mean particular social and governmental issues, I mean lack of mental self-sufficiency), and his theoretical work realized in such a way.
Carl Sagan, Cosmos
https://twitter.com/MLevitt_NP2013/status/128703673856573849...?
I wonder if anyone can find a prediction of his that turned out correct. He's been saying it will be over in a month, every month since January.
"""A computer-savvy scientist, Dr. Levitt relies on a Macintosh laptop with VMWare virtualization running a Windows OS, where he stores 200 gigabytes of data, including 40 gigabytes of over 300,000 emails, and of course relies on X1 to make sense of it all"""
Michael had other interesting anecdotes for me, like when he advised Sergey Brin to charge for Google search (Sergey was taking some course of his when launching Google).
It's not surprising that the community has lost faith in him.
https://www.stanforddaily.com/2020/08/02/qa-michael-levitt-o...
> TSD: What do you predict to happen in the coming months regarding COVID-19 in the U.S.?
> ML: I think it’s going to end up with less deaths than what I thought. In March, I thought there would be a total of 220,000, but there will be less than that. Right now it is around 155,000 in the U.S., and I am expecting it to end up under 170,000 or maybe 175,000.
---
The article is dated August 2, and so these confident predictions are not looking too good at all. I'm not cherry-picking, there's a lot more in the article, like:
> I think that when we come to look back, we’re going to say that wasn’t such a terrible disease.
This just doesn't seem to be a person who has a good perspective on public health or public policy.
---
And finally, ML's remarks about the Swedish death rate are not correct, as pointed out editorially in the article.
You should familiarize yourself with power law processes:
Edit: In other words, when his prediction is given with an implied error of +/- 5000, that speaks to a level of confidence that as you point out, anybody familiar with power law processes would not dare to give.
Digging, found a paper from 2007 shitting on it.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1906780/
Pubmed reference. Abstract predicts 200,000 cases total.
https://pubmed.ncbi.nlm.nih.gov/2308183/
I'm left with the opinion that Michael Levitt is basically a stupid person.
Perhaps the proper term is fachidiot.
Since you brought it up: my PhD was in applied probability, and I work with heavy-tailed natural data every weekday.
*
As my comment noted, ML literally says 170K or maybe 175K, and flat-out dismisses 220K. The latter is an interesting figure, because we recently surpassed it.
We're now well out of the 170-175K envelope and still rising at about 5K deaths per week, about 20K deaths per month [1]. Note that our weekly death toll is the implied accuracy of ML's prediction. This is a failure.
If ML thought there was a power law behind the mortality from an in-process event, he should have said "170K deaths or maybe 1.7M". But he didn't say that, did he?
*
It's interesting to see an interview with an excellent science communicator. They will say what we know, and not a bit more than that. They will choose their words carefully. They will distinguish between speculation and accepted truth.
I live in southern CA, and Lucy Jones (formerly of the USGS) is such a person, for earthquakes. These people are a treasure, because this stuff is hard and it's important to get it right if you're going to make public statements.
[1] https://www.cdc.gov/coronavirus/2019-ncov/covid-data/forecas...
He presented a bunch of authoritative looking numbers and graphs, and says "if we draw a curve through these, we see this trend". And it turned out to be massively wrong, but he never issued a mea culpa. Instead, he says "Going off of the only numbers that I already had, it basically quite clearly said that inside Hubei, there would be a few thousand deaths, and outside of Hubei in China there would only be a few hundred deaths. And that was enough to give people more confidence in understanding the virus, though I don’t think it really prepared the rest of the world for what was going to end up happening. "
Yes, there is a reason that actual epidemiologists did NOT come to that conclusion: due to myriad factors, data about disease is messy, and you need to account for that. I never see him acknowledge that fact; I just see more "I connected these dots and extrapolated the curve predictions", even though that hasn't worked for him yet:
> I think it’s going to end up with less deaths than what I thought. In March, I thought there would be a total of 220,000, but there will be less than that. Right now it is around 155,000 in the U.S., and I am expecting it to end up under 170,000 or maybe 175,000.
If you want to fault this guy for something it's trusting the data out of China at all and even considering it appropriate to do analysis with those numbers.
Again, 170k estimated vs 220k actual deaths is not a huge difference when dealing with a highly multiplicative event like a pandemic.