Quoting: "I’m concerned that people interpret your conjectures as facts.". Harris the author is a Phd candidate in biology at Stanford who is calling out Levitt, a Nobel Prize winning structural biologist at Stanford, for his evidently non-scientifically based popular COVID advice "in keeping with your hostility to epidemiology". Harris maintains Levitt is doing damage, real damage to public health, to his reputation, and the reputation of Stanford. I am just some random, but I was persuaded by her reasoned and pointed argument. Her article took some guts to write. I hope it is constructively received. I would be curious to understand what is driving Levitt.
The author of this article presented documented evidence of Dr. Levitt's incorrect and speculative statements which he has not bothered to clarify. I am persuaded by the article because it presents the evidence clearly and backs it up with citations of that evidence.
I was persuaded by the evidence cited by Harris in her editorial. I am categorically interested in Levitt's reasoning. Based on this article, his beliefs are miscallibrated, and that seems odd to me.
A broader question is what should happen when a scientist voices a non scientific opinion? A famous scientist is more likely to get my attention than some ordinary scientist or some pundit. Maybe I am a sophisticated consumer, or may be not. Should I believe this person just because he or she is a scientist? Should the scientist issue a disclaimer - "this is my non scientific opinion"? I think a point Harris is making is that if one is going to opine while wearing the mantle of scientific authority - Nobel, Stanford prof - your arguments should be robust. Harris met that standard for me.
It is very easy - all interviews, opinions, public speeches by all scientists are under "non scientific opinion". Only scientific articles on reputable scientific journals are scientific opinion. All other media is not, because even if you link your speech to specific article and you speak with journalist you use dumbed down terminology. During process of converting scientific article into layman terms there is loss of information. Sometimes it is crucial information or it becomes source of misinterpretation.
Well it is very easy to check consensus view of science - it is published scientific articles. Why it is consensus view? Because it is peer reviewed and different journals have different reputation ranks. In scientific debate you always link the source - scientific article. If he cannot link sources, means there is no need to understand either “what is driving Levitt”, either “what is Levitt’s reasoning?”
She's loose with facts, herself: consider the four links provided at the end of the sentence "I was personally distressed to watch you tell the governor of Florida, where my high-risk family members live, that COVID-19 does not present a significant threat in terms of overwhelming hospitals. This is verifiably false." That's a pretty big, specific claim. What is the evidence she is providing?
Every word in "this is verifiably false" links to a different article, only one of which provides even weak evidence that one might characterize as supporting the claim:
* Link one is to an article about San Francisco from December 10, claiming that hospitals there will run out of beds in 17 days. Florida is not mentioned (incidentally, hospital capacity in SF has increased slightly since the article was written [1]).
* Link two is to a an article about Miami-Dade from December 4, where they quote CEOs from two hospital chains, who say that they have enough beds to handle the current caseload, but "may" not have enough staff.
* Link three is about cases nationwide. It does not mention Florida.
* Link four is about a White House report from November. The article explicitly says: "The November 29 document showed a minimal change in total hospitalization from the week prior". The "evidence" here is that the document predicts a "post-Thanksgiving surge".
To my knowledge, no hospital system in Florida has been "overwhelmed" by any conventional definition of the word [2]. Florida hospitals currently appear to be well below the peak seen in July [3]. The predicted "post-thanksgiving surge" does not appear to have materialized [4].
Reasonable people can certainly disagree and debate the current level of strain on hospitals in the US, and the appropriate trade-offs in our response to these strains. However, it's not fair or intellectually honest to attack another person for weaknesses in past speculation when you are yourself engaging in speculative, possibly misleading forms of arguments.
According to the New York Times' "How Full Are I.C.U.s Near You?", AdventHealth Ocala in Florida is currently at 97% capacity with 56 Covid-19 patients, and the average I.C.U. in Florida is at 82% capacity.
It's possible to find specific hospitals that are at/near capacity at nearly any time of year. Right now, hospitals across Florida appear to be at about 80% capacity, with individual hospitals ranging from 0% to 100% available:
This is consistent with the CDC link my previous post. Of course, it's not possible to tell how this compares to any normal year, because we don't have that data.
To be fair to her, she did not claim that hospitals in Florida specifically were at risk from being overwhelmed. Instead, she takes issue with Levitt telling the governor of Florida that COVID generally does not present a significant threat in terms of overwhelming hospitals, which is in fact verifiably false; COVID has overwhelmed hospitals all over the world multiple times. I write this from LA, where our hospitals are currently overwhelmed, and likely to get more so in the next three weeks as people return from holiday travel.
Yes, I considered this. The sentence is phrased somewhat ambiguously, which is why I said she's being loose with the facts. But the links she provides neither back up the claim (all are speculative), and 3/4 are not specific to Florida, so I think it's a fair critique given the forceful nature of the statement, and the fact that she's calling out Levitt for his own past speculation.
The link to Levitt's supposed comments to the governor is broken, so we can't verify what he actually said in detail. However, if you follow the other links, you can see that she is not above taking comments out of context when it serves her purpose. In the comment about about 10 deaths in Israel, it's fairly clear that Levitt is not making a bold claim about the distant future -- he's saying that given the case counts at the time, it would be surprising to see more than ten deaths, not predicting ten deaths in Israel, ever.
Mallory Harris is critiquing Dr. Levit because he is not agreeing with the scientific consensus. She is also holding him accountable for re-interpretations of his twitter stream. If this is true, then anyone with any authority or rank should stop using twitter or say anything at all; there is always a chance that someone else can twist your words and use your authority to back their case.
Successful public figures typically write with their particular audience in mind, and often hire others to help them do this effectively, and in a way their audience understands and does not misinterpret. There are entire career disciplines that focus almost exclusively on this skill.
Maybe the character limits inherent in platforms like twitter cause many of the problems we're seeing right now, because people literally don't have the words to explain themselves.
Twitter is another form of public speech, like interviews, public lectures, etc. Scientists always were accountable for re-interpretations of their lectures, public speeches. Why it should change on twitter?
Really? Tell me how scientists have been held responsible for reinterpretations. It's easy to find many examples of scientists who can't be blamed for how their discoveries are used. For instance, Watson and Crick would never have reported DNA because they knew it would be used by supremicists and other racists to justify their thoughts.
Mallory Harris is critiquing Dr. Levitt because he has made scientifically incorrect statements and when presented with the contrary evidence has not corrected his misstatements. The article documents these misstatements, implores him to correct the record, and he does not. These misstatements are clearly presented and cited.
It would be problematic for a medical doctor to disseminate medically incorrect information to the public and not correct or clarify it. Similarly, it is problematic for a scientist to disseminate scientifically incorrect information and not correct or clarify it. It would be a simple thing for him to simply tweet, "On x date I tweeted y. The evidence now suggests that this statement was incorrect." The science on this continues to evolve, acknowledging that sometimes we are wrong is not a stain on one's reputation. It is the mark of a scientist who is pursuing evidence rather than a narrative. Dr. Levitt is pursuing narrative over evidence, and his actions have consequences.
I have been following Dr. Levitt since the beginning of the pandemic, and I have now count loss of how many of his predictions turned out to be wrong. Also he never fleshes out how he is doing his calculations. At this point, any reasonable person with some self-reflection would just shut up.
There's plenty of "wrong" to go around. Nick Kristoff at the NYT predicted we might have more than 1m deaths by the end of the summer. In the US. But he's not shutting up.
And gosh, Neil Ferguson, the famous epidemiologist in England keeps coming up with predictions based on the thinnest amount of data. Yet he's not shutting up.
I think everyone who is devoting any time to shutting down or shutting up someone else would be better served spending the time generating better data. Censorship is a terrible impulse.
Given that the 1M prediction was with no intervention, and interventions happened all across the country (and are approaching the prediction outlined in that article), are you repeating this misinformation out of ignorance or malice?
Of course, denying that interventions played a big role in preventing this from getting even more out of control would be be table-stakes with you: https://news.ycombinator.com/item?id=23116782
> Nick Kristoff at the NYT predicted we might have more than 1m deaths by the end of the summer.
This would be a considerably better prediction than the idea that asymptomatic spread can't be serious factor; it'd be merely off by a factor of 6 whereas I'm honestly not sure how anyone with a high school biology education could dismiss the possibility that a viral infection might be contagious and asymptomatic.
And I don't know when Kristoff's prediction was made, but if it was before June, given a NYC-centric perspective it's a credible projection absent changes of behavior and treatment advances, both of which we got in modest degrees (probably around a factor of 6).
Plenty of wrong to go around? Sure. Some parties seem to be carrying more of the plenty.
The magnitudes of these errors are wildly different.
The U.S. actually had about 200,000 deaths at the end of the summer. A million was an overestimate, but this was on March 20. Kristof/Ferguson's error was 5x.
On the other hand, on the same day, Levitt said that he would be surprised if Israel had more than 10 COVID deaths total. Currently Israel has 3307 COVID deaths, an error of 330x.
He has made a whole series of wildly wrong predictions, all in the same direction.
It's fairly clear from the linked article that he's saying given the low case count at the time (~400) he's not expecting more than ten deaths:
"Speaking to Kan's Reshet Bet radio on Wednesday, Levitt, who lives part-time in Tel Aviv, said that on a global scale, the number of cases in Israel is very small."
Regardless of what you think of Levitt, he's clearly not making predictions about the indefinite future.
This is different. This is a scientist condemning another scientist for standing on the pulpit and proselytizing to a mass of people. The cirtique by Mallory is that Levitt is using his credentials to bypass the scientific process and undermines the work of many. This is, in my opinion, something we should all condemn.
I find it fascinating that people always object to others' speech not because they might be misled by it, but because they are concerned that others might. It's patronizing and arrogant.
Mallory is not saying that his words are directly swaying people but that his words are being used as political tool to undermine the scientific consensus.
Once the political climate around epidemiology settles down and allows for hard questions to be asked (and answered), there are some questions about a perceived loss of credibility that I would like to see be addressed.
When a lay person first encounters an epidemic, the first questions are something like:
1. How does it spread?
2. How likely am I to catch it if I’m exposed to it?
3. How can I prevent the spread? How effective is it?
4. What should I consider if I have to risk an exposure?
I think about HIV and these questions. With HIV, each question has an extremely clear answer, often with statistics attached (eg: condoms prevent transmission with 99% effectiveness), is clearly communicated and understood. For non-experts, the mental model developed by doing even a little bit of reading is sufficiently accurate to make reasonable guesses on how to behave to reduce risks with HIV.
As a lay person who wants to understand how to reduce risks around covid-19, the epidemiology community has not been helpful. Even now, it’s hard to find straightforward answers from peer reviewed publications for questions like “Does covid spread through touch?” in a way that lets me model risks in my head. “Use hand sanitizers just to be safe” isn’t good enough from a field dedicated to figure out how infections spread and at what rates.
It’s December 2020. A year since Covid-19 appeared on the world stage. What I want is a table that lists vectors of transmission, effectiveness of transmission along that vector.
Something like
| Vector. | Effectiveness |
——————————————————————————————————————
| Breathing 10ppm | |
| air for 5 mins. | |
| Breathing 100ppm | |
| air for 5 mins. | |
...
Throw in some graphs, confidence intervals and what variables to adjust for (temperature, time of exposure etc)
With the right data, it should be blindingly obvious that masks work or that 5 covid positive people breathing out viral particles inside a Costco can increase the viral
concentration inside to dangerous levels in 40 mins (I’m
making up all the numbers here).
Instead, we have public announcements that 23 mutations
in a new strain result in a 71% more transmissible virus
with implications that some expert in the field without
any data or details on how the 71% number was derived. Do the new mutations cause an increase in viral particles being exhaled by an infected person? How much increase? Are the
viral particles somehow smaller and able to penetrate masks
more effectively or survive longer in the air and thus travel
greater distances downwind? What measurements have you made, what variables did you control for, what scrutiny has your
observations seen?
When LIGO announced the first observations of gravity waves, everyone believed that the people behind the announcement had done due diligence before making the announcement. The physics community holds itself to a very high code of conduct, and this lends them credibility. Even mistakes like the faster-than-light neutrinos are forgiven because of the credibility the physicists maintain.
The absence of that credibility has been a disaster. I’m not going to forget the announcements that masks don’t work and shouldn’t be used earlier this year. That’s an irreparable damage to the credibility of epidemiology - and when covid is over, I would like to see a real post-mortem with actionable steps to fix the code of conduct and credibility.
For a comparable case, Ian Plimer, an AGW denialist geologist, of repute in his own field, but now notorious for making completely outrageous statements to his side of belief in the climate science.
He's off-piste, outside his competency, arguing outside of the peer review process.
Why do senior scientists do this? Is this some form of relevance deprivation syndrome?
37 comments
[ 2.9 ms ] story [ 98.0 ms ] threadEDIT: adding a link to an interview w/ Levitt from 8/20: https://www.stanforddaily.com/2020/08/02/qa-michael-levitt-o...
That’s a startling outcome to me. Is it because she claims to hold the consensus view? Because it was well written?
How do you conclude with being curious to understand “what is driving Levitt” rather than “what is Levitt’s reasoning?”
A broader question is what should happen when a scientist voices a non scientific opinion? A famous scientist is more likely to get my attention than some ordinary scientist or some pundit. Maybe I am a sophisticated consumer, or may be not. Should I believe this person just because he or she is a scientist? Should the scientist issue a disclaimer - "this is my non scientific opinion"? I think a point Harris is making is that if one is going to opine while wearing the mantle of scientific authority - Nobel, Stanford prof - your arguments should be robust. Harris met that standard for me.
Every word in "this is verifiably false" links to a different article, only one of which provides even weak evidence that one might characterize as supporting the claim:
* Link one is to an article about San Francisco from December 10, claiming that hospitals there will run out of beds in 17 days. Florida is not mentioned (incidentally, hospital capacity in SF has increased slightly since the article was written [1]).
* Link two is to a an article about Miami-Dade from December 4, where they quote CEOs from two hospital chains, who say that they have enough beds to handle the current caseload, but "may" not have enough staff.
* Link three is about cases nationwide. It does not mention Florida.
* Link four is about a White House report from November. The article explicitly says: "The November 29 document showed a minimal change in total hospitalization from the week prior". The "evidence" here is that the document predicts a "post-Thanksgiving surge".
To my knowledge, no hospital system in Florida has been "overwhelmed" by any conventional definition of the word [2]. Florida hospitals currently appear to be well below the peak seen in July [3]. The predicted "post-thanksgiving surge" does not appear to have materialized [4].
Reasonable people can certainly disagree and debate the current level of strain on hospitals in the US, and the appropriate trade-offs in our response to these strains. However, it's not fair or intellectually honest to attack another person for weaknesses in past speculation when you are yourself engaging in speculative, possibly misleading forms of arguments.
[1] https://data.sfgov.org/stories/s/Hospital-Capacity/qtdt-yqr2...
[2] https://www.cdc.gov/nhsn/covid19/report-patient-impact.html
[3] https://www.orlandosentinel.com/coronavirus/os-ne-coronaviru...
[4] https://covid-19.direct/state/FL
https://www.nytimes.com/interactive/2020/us/covid-hospitals-...
https://bi.ahca.myflorida.com/t/ABICC/views/Public/HospitalB...
This is consistent with the CDC link my previous post. Of course, it's not possible to tell how this compares to any normal year, because we don't have that data.
The link to Levitt's supposed comments to the governor is broken, so we can't verify what he actually said in detail. However, if you follow the other links, you can see that she is not above taking comments out of context when it serves her purpose. In the comment about about 10 deaths in Israel, it's fairly clear that Levitt is not making a bold claim about the distant future -- he's saying that given the case counts at the time, it would be surprising to see more than ten deaths, not predicting ten deaths in Israel, ever.
https://twitter.com/MLevitt_NP2013/status/130176837049748275...
What if he had ended this tweet with "But: I will take any FDA approved vaccine." ?
Except - he was at exactly 280 characters, so he couldn't have added this without removing other text, or doing a multi-tweet message.
https://academic.oup.com/joc/article-abstract/69/4/345/55470... says "We show that doubling the permissible length of a tweet led to less uncivil, more polite, and more constructive discussions online."
Maybe the character limits inherent in platforms like twitter cause many of the problems we're seeing right now, because people literally don't have the words to explain themselves.
It would be problematic for a medical doctor to disseminate medically incorrect information to the public and not correct or clarify it. Similarly, it is problematic for a scientist to disseminate scientifically incorrect information and not correct or clarify it. It would be a simple thing for him to simply tweet, "On x date I tweeted y. The evidence now suggests that this statement was incorrect." The science on this continues to evolve, acknowledging that sometimes we are wrong is not a stain on one's reputation. It is the mark of a scientist who is pursuing evidence rather than a narrative. Dr. Levitt is pursuing narrative over evidence, and his actions have consequences.
And gosh, Neil Ferguson, the famous epidemiologist in England keeps coming up with predictions based on the thinnest amount of data. Yet he's not shutting up.
I think everyone who is devoting any time to shutting down or shutting up someone else would be better served spending the time generating better data. Censorship is a terrible impulse.
Given that the 1M prediction was with no intervention, and interventions happened all across the country (and are approaching the prediction outlined in that article), are you repeating this misinformation out of ignorance or malice?
Of course, denying that interventions played a big role in preventing this from getting even more out of control would be be table-stakes with you: https://news.ycombinator.com/item?id=23116782
This would be a considerably better prediction than the idea that asymptomatic spread can't be serious factor; it'd be merely off by a factor of 6 whereas I'm honestly not sure how anyone with a high school biology education could dismiss the possibility that a viral infection might be contagious and asymptomatic.
And I don't know when Kristoff's prediction was made, but if it was before June, given a NYC-centric perspective it's a credible projection absent changes of behavior and treatment advances, both of which we got in modest degrees (probably around a factor of 6).
Plenty of wrong to go around? Sure. Some parties seem to be carrying more of the plenty.
The U.S. actually had about 200,000 deaths at the end of the summer. A million was an overestimate, but this was on March 20. Kristof/Ferguson's error was 5x.
On the other hand, on the same day, Levitt said that he would be surprised if Israel had more than 10 COVID deaths total. Currently Israel has 3307 COVID deaths, an error of 330x.
He has made a whole series of wildly wrong predictions, all in the same direction.
"Speaking to Kan's Reshet Bet radio on Wednesday, Levitt, who lives part-time in Tel Aviv, said that on a global scale, the number of cases in Israel is very small."
Regardless of what you think of Levitt, he's clearly not making predictions about the indefinite future.
There’s rarely anything that can be said nowadays without some activist finding a reason to take offense.
It’s funny when it’s the activist themselves that get cannibalized by their own Puritanical brethren.
The scientific process isn't owned by anyone. People are free to speak words; those who say otherwise are speech police.
The obsession of certain people with "the harms of words" is illiberal and authoritarian.
It’s in vogue to complain about the bad people that say bad things.
You are authoritarian insofar as you're appealing to the authority of science, only if you think it seems reasonable.
If there's anything struggle session-like about this, it's you contorting the author's respectful opinion into some kind of public flogging.
When a lay person first encounters an epidemic, the first questions are something like: 1. How does it spread? 2. How likely am I to catch it if I’m exposed to it? 3. How can I prevent the spread? How effective is it? 4. What should I consider if I have to risk an exposure?
I think about HIV and these questions. With HIV, each question has an extremely clear answer, often with statistics attached (eg: condoms prevent transmission with 99% effectiveness), is clearly communicated and understood. For non-experts, the mental model developed by doing even a little bit of reading is sufficiently accurate to make reasonable guesses on how to behave to reduce risks with HIV.
As a lay person who wants to understand how to reduce risks around covid-19, the epidemiology community has not been helpful. Even now, it’s hard to find straightforward answers from peer reviewed publications for questions like “Does covid spread through touch?” in a way that lets me model risks in my head. “Use hand sanitizers just to be safe” isn’t good enough from a field dedicated to figure out how infections spread and at what rates.
It’s December 2020. A year since Covid-19 appeared on the world stage. What I want is a table that lists vectors of transmission, effectiveness of transmission along that vector.
Something like
Throw in some graphs, confidence intervals and what variables to adjust for (temperature, time of exposure etc)With the right data, it should be blindingly obvious that masks work or that 5 covid positive people breathing out viral particles inside a Costco can increase the viral concentration inside to dangerous levels in 40 mins (I’m making up all the numbers here).
Instead, we have public announcements that 23 mutations in a new strain result in a 71% more transmissible virus with implications that some expert in the field without any data or details on how the 71% number was derived. Do the new mutations cause an increase in viral particles being exhaled by an infected person? How much increase? Are the viral particles somehow smaller and able to penetrate masks more effectively or survive longer in the air and thus travel greater distances downwind? What measurements have you made, what variables did you control for, what scrutiny has your observations seen?
When LIGO announced the first observations of gravity waves, everyone believed that the people behind the announcement had done due diligence before making the announcement. The physics community holds itself to a very high code of conduct, and this lends them credibility. Even mistakes like the faster-than-light neutrinos are forgiven because of the credibility the physicists maintain.
The absence of that credibility has been a disaster. I’m not going to forget the announcements that masks don’t work and shouldn’t be used earlier this year. That’s an irreparable damage to the credibility of epidemiology - and when covid is over, I would like to see a real post-mortem with actionable steps to fix the code of conduct and credibility.
He's off-piste, outside his competency, arguing outside of the peer review process.
Why do senior scientists do this? Is this some form of relevance deprivation syndrome?