What was my mistake evaluating risk in this situation?

post by Andrew Vlahos (andrew-vlahos) · 2021-08-03T02:54:35.887Z · LW · GW · 6 comments

This is a question post.

Contents

  Answers
    30 dawangy
    22 lsusr
    18 JBlack
    11 Kelly Smith
    10 Radu Floricica
    9 Dagon
    4 ChristianKl
    2 Unnamed
    2 AllAmericanBreakfast
    0 Stuart Anderson
None
6 comments

At the start of the pandemic, I heard that Covid was likely to be a problem. I didn't pay much attention, because I remembered the Swine Flu "pandemic", and Covid was getting less attention than Swine Flu did at the time. I also knew that the media has a strong incentive to cause hype over things that aren't really dangerous trends. I am in an age range very unlikely to die from disease, and claims of a lung plague causing lasting brain problems sounded absurd.

I figured that if a disease about as deadly as the regular flu caused so much panic, and Covid caused less and mainly harmed people already at respiratory risk, I didn't need to worry about it.

I eventually updated my beliefs, but I don't see how I could have avoided making my mistake. How could I have done better?


Now that a bunch of answers are in, I will say what I think the mistakes were.

First of all, a clarification. I heard about it sometime in early February (I don't remember specifically since I wasn't paying much attention), and I re-examined the data and updated my belief in whether it was a problem at the beginning of March, when I noticed that even people who normally don't worry about things like this started considering a lockdown. Remember: I knew it was spreading, I thought that it wouldn't be much worse than regular diseases.

The mistake wasn't "I should have listened to the trustworthy sources instead of the untrustworthy ones". I didn't know much about which sources were the most reliable, and what I did know wouldn't have helped. For example, I know that where people put their own money is much more reliable than what they tell other people to do (why prediction markets are so much more accurate than other things). However, the US stock market reached an all time high on February 12, indicating that people whose whole job is paying attention to this stuff weren't worried at all.

The mistakes I did do were these:

  1. I had been too overconfident in civilization's ability to withstand plague. Disease has been pretty much eradicated, and even if a new one appears there are over-the-counter cough medicine and fever reducers and things.
Approximately how I thought.

My mistake was not re-evaluating this belief based on things that almost became big. I should have updated based on Ebola and SARS but didn't.

2. Personal experience said it was even less dangerous than the flu. My elderly great-aunt tested positive in late February, but was almost asymptomatic. Also, my parents and I got a bad cold at the end of January with the same symptoms (dry cough, fever, got better then a few days later it became bad again), and although we didn't think to test it at the time, once we heard of covid we thought that was probably it. Our symptoms definitely weren't worth freaking out about. I probably should have payed more attention to other people's reports of worse symptoms, although my prior of "lung virus causes permanent brain damage" was much less than "placebo effect and/or typical media terrible statistics", and although genetic variation is real it didn't seem likely to have  "shut down the country" level severity. I'm not quite sure how this mistake could have been fixed.

3. I should have checked a source about China, not just from places that don't have much covid yet. I'm not fluent in Chinese, but I really should have at least checked how strictly they were quarantining and compared it to what they did during SARS. I didn't really look into it at all.

4. There was another problem, but this would take longer to explain and would need its own post.

Answers

answer by dawangy · 2021-08-05T06:07:40.193Z · LW(p) · GW(p)

I also believe I'm qualified to speak on this issue as I was very worried about this from a very early time (even bought masks in late January). Others have brought up some good points here but I would say that the key issue you had is failure to pay attention to details. It may be true that the news likes to cause unnecessary hype over things for attention, but it is the details of the situation that really set this one apart from others.

  1. After acknowledging human to human transmission, early reports suggested that the virus was capable of spreading through respiratory routes without the host displaying any symptoms of illness. This is an extremely distinct and alarming feature and is the main reason I started flipping out.
  2. You may not have had ready access to this information, but people on Chinese social media were posting videos of packed hospitals in Wuhan. A back of the envelope estimation of the scale of the pandemic using estimates of hospital capacity in the region meant that the epidemic was far larger than reported at the time.
  3. At some point in February we found several cases of COVID in the United States that could not be traced to any other known case of COVID. Again, if you pay attention to the details and implications this means more than if you just read the headline passively.
comment by lsusr · 2021-08-05T07:01:22.833Z · LW(p) · GW(p)

This is a good analysis.

I would say that the key issue you had is failure to pay attention to details.

I think it deserves attention how dawangy thought through the material details while disregarding how much the media did (or didn't) hype the virus. Material details matter. Hype doesn't.

answer by lsusr · 2021-08-05T05:39:16.437Z · LW(p) · GW(p)

When China [LW · GW] quarantines an entire city, you should be afraid of what's in that city.

I think I'm qualified to answer this question because I was paying attention to COVID-19 since the Wuhan quarantine, many months before the pandemic spread to the West. The day person-to-person spread was confirmed in the United States, I sprung [LW · GW] into action [LW · GW].

You made two mistakes.

Our ancestors must rarely have encountered cases of exponential growth, because our intuitions are no guide here

Startup = Growth by Paul Graham

Here are the antidotes.

  • Ignore the news. Focus on long-term knowledge [LW · GW] instead.
  • Create personal fire alarms for events with the potential for sustained exponential growth. More generally, you should prepare far in advance to protect yourself from anything the authorities are unprepared for (especially disasters unprecedented in living memory).
comment by Stuart Anderson (stuart-anderson) · 2021-08-05T14:58:54.809Z · LW(p) · GW(p)

_

Replies from: lsusr, habryka4
comment by lsusr · 2021-08-05T21:39:46.871Z · LW(p) · GW(p)

What do you mean?

Replies from: stuart-anderson
comment by habryka (habryka4) · 2021-08-07T20:42:02.703Z · LW(p) · GW(p)

Mod note:  This has nothing to do with the topic of this post, and veers into random political discussion. Comments like this are not welcome on LessWrong and can lead to a ban.

Replies from: Viliam
comment by Viliam · 2021-08-10T21:44:09.005Z · LW(p) · GW(p)

Oh, I thought the original comment was literally "_", because the reply "What do you mean?" also makes sense in that context.

answer by JBlack · 2021-08-05T04:54:35.663Z · LW(p) · GW(p)

It's quite possible that you didn't make a mistake. There are a few hints at mistakes in the post, but that's presumably just a very short approximation of your actual reasoning.

I'd consider it problematic to treat degree of media panic as a decent proxy for actual risk. It's almost cliché by now to compare reporting on terrorism to that on heart disease, but still a valid pointer toward severe failures of this proxy.

I suspect you may also be looking back on the 2009 H1N1 reporting and condensing about a year or so of reporting into one "media panic" blob. Comparing newspaper article records at similar times from the first known infections, it looks like COVID-19 reporting actually took off much faster and stronger in tone than that for 2009 H1N1.

I think a third potential problem, not at the time but right now, is a hindsight bias. Now that we know that the event happened, it's tempting to update too far in favour of it having been likely beforehand. There are hundreds of new diseases and variants infecting people every year. During the first few weeks to months, their pandemic potential is often completely unknown. Paying little day-to-day attention to each one is rational for the general public, though not for epidemiologists or public health officials.

As a general probability, by the time a disease of this sort spreads to your own area (and therefore you should be changing your behaviour), there is almost always a lot more known and published about it. There is a low probability that any given person is going to be one of the unlucky first hundred thousand infected out of 8 billion people on the planet. By that time it's usually pretty well known whether it's likely to be serious enough to take action.

So perhaps your estimation wasn't really as miscalibrated as you think.

answer by Kelly Smith · 2021-08-05T06:47:45.848Z · LW(p) · GW(p)

I think the answer here is pretty cut and dry. 

"Media has a strong incentive to cause hype over things that aren't really dangerous trends."

This was your mistake. If  you didn't believe your sources were trustworthy, then why didn't you find sources you could trust? 

There were plenty of trustworthy sources of information early in the pandemic. A lot of credible people were talking about Covid a month before the mainstream media started prioritizing their coverage. 

comment by Neel Nanda (neel-nanda-1) · 2021-08-05T09:39:44.952Z · LW(p) · GW(p)
Media has a strong incentive to cause hype over things that aren't really dangerous trends

I think the inference here was 'media has a strong incentive to cause hype over stuff that doesn't matter, so surely they have an even stronger incentive to cause hype over stuff that is actually dangerous'. Empirically, this was wrong, but I'm confused about why!

Replies from: kelly-smith
comment by Kelly Smith (kelly-smith) · 2021-08-06T09:03:00.153Z · LW(p) · GW(p)

I don't think that's an accurate statement to begin with. The media has a strong incentive to cause hype over anything that drives ratings, so using mainstream media coverage to forecast the severity of a once in a lifetime event such as a viral pandemic just seems like a bad idea in my opinion. There are much better ways. 

comment by Stuart Anderson (stuart-anderson) · 2021-08-05T13:14:02.140Z · LW(p) · GW(p)

-

Replies from: kelly-smith
comment by Kelly Smith (kelly-smith) · 2021-08-06T08:50:19.471Z · LW(p) · GW(p)

I don't know what you're even implying about the Fauci email drop. If you dig even a little into that story it's clear there was nothing there and it was the typical media tactic where they dig through thousands of emails and quote tiny snippets out of context to people who don't know better and will take the bait. If stories like these are popping up in the media you consume, that means you're consuming entertainment media, not actual journalism. Even if Fauci contradicted himself or made a mistake, who cares? He is one of the foremost public health experts in the world and his credentials are impeccable. If you listen to experts, you're going to be on the right side of the vast majority of issues. 

...which is why I don't agree that trustworthy sources "no longer exist". For people who are sufficiently motivated and good at doing research, it's never been easier to find accurate, trustworthy information. Why would you bother going to the mainstream media for news on a virus when you can get that information directly from the world's top physicians and research scientists? They're all on Twitter and many of them have blogs or podcasts. 

When you're getting your Covid information from an epidemiologist at Johns Hopkins or the Harvard School of Medicine, you aren't worried that it's poorly sourced, second hand information revolving around some political narrative. 

Replies from: ChristianKl
comment by ChristianKl · 2021-08-06T09:28:20.410Z · LW(p) · GW(p)

If you dig even a little into that story it's clear there was nothing there and it was the typical media tactic where they dig through thousands of emails and quote tiny snippets out of context to people who don't know better and will take the bait. 

That's quite false. The emails are a lot more incriminating in context then without the context. Take the fact that Fauci emailed about the Baric&Shi paper from 2015 with "gain of function" in the file name. Taking alone that's already bad. If you however read it in context, he did that hours after getting the email that told him that the virus being of natural origin is not in line with evolutionary theory. 

If you want to see the emails in context, I did the work at https://www.lesswrong.com/posts/wQLXNjMKXdXXdK8kL/fauci-s-emails-and-the-lab-leak-hypothesis

Even if Fauci contradicted himself or made a mistake, who cares? 

He made a mistake. He circumvented US government policy to fund gain of function research under his watch. We are now at the HOUSE FOREIGN AFFAIRS COMMITTEEREPORT MINORITY STAFF concluding that the preponderance of evidence indicated that it was a lab leak due of a virus created via gain of function research.

Then he engaged in the coverup that you see in the Fauci emails. Calling it a mistake doesn't really do it justice given that there was clear malicious intent.  

Replies from: gjm
comment by gjm · 2021-08-07T13:31:10.232Z · LW(p) · GW(p)

It may be worth being clear what "House Foreign Affairs Committee Minority Staff" means. (It wasn't immediately obvious to me on reading this.)

The US House of Representatives has a Committee on Foreign Affairs. At present it has 27 Democratic members and 25 Republican members. The Committee Minority Staff means just the 25 Republican members of that committee.

So this is a strongly partisan group of people, and this is an explicitly political undertaking.

Exactly how much this means we should mistrust them (or, symmetrically, a group of Democrats acting similarly) depends on exactly how they got started. E.g., imagine that in a meeting of the whole committee someone said "We should look into the origins of Covid-19" and all the Democrats refused, so the Republicans decided to go it alone; while that would strongly suggest political polarization is at work, it would still be consistent with this being a group of people defined by their wish to investigate the issue, and whatever they produce might be our best chance of getting at the truth.

On the other hand, if what happened is that just the Republicans on the committee got together and decided that they would investigate the topic and write a report -- well, that looks much more like a politically-motivated thing which will only by coincidence tell us anything useful, and whose bottom line was already written before they began.

So what's the actual history? According to this latest report, the "Committee Minority Staff" decided in March 2020 to put together a report on the origins of Covid-19. They published their "House Foreign Affairs Committee Minority Staff Final Report on The Origins of the COVID-19 Global Pandemic, Including the Roles of the Chinese Communist Party and the World Health Organization" in September 2020. Curiously, I can find on the internet little sign of any actual publication of that document. I did find what appears to be a copy of it, though. The latest thing is, formally, an addendum to the earlier report.

I had a look through the Congressional Record for March 2020, and (so far as I can tell this is all there is) the videos of House Committee on Foreign Affairs meetings. I didn't find anything in either discussing the (prospective or already decided) formation of a group to investigate the origins of Covid-19. (There's one video of a committee meeting about Covid-19, on 2020-02-27; nothing else prior to 2020-03-16. There was one tangential reference to the possible origins of the disease, but nothing remotely to do with any sort of US investigation of it. The Congressional Record turns up a few people complaining about the CCP being dishonest, but again nothing about a US investigation of any kind.) I don't really know how these things work, so I don't know how much evidence that is that there wasn't an attempt to form a bipartisan group of this sort; it seems like it's some evidence, though.

(Maybe it needs saying explicitly that there are reasons why Republican politicians might have wanted to do this, with something like the reports they've produced as an already-decided goal. For the earlier report: The Trump administration was keen to emphasize the Chinese-ness of Covid-19, and to blame China for its spread, from the very beginning; their political opponents responded by calling that racism; so making China look bad = making Republicans look good and Democrats bad. More generally, the Trump administration was very keen to be seen as tough on China, perhaps as a counterweight to accusations that they were soft on Russia. The Trump administration had been sharply critical of the WHO's handling of Covid-19, and in April 2020 they announced that they were cutting off US funding to the WHO. Of course their political opponents condemned all that. So making the WHO look bad = making Republicans look good and Democrats bad. For the later addendum: for whatever reason the lab leak hypothesis was highly politically polarized, so supporting it = making Republicans look good and Democrats bad. The tension between Mr Trump and Mr Fauci is well documented, and of course now Fauci is working for a Democratic administration; so attacking Fauci = making Republicans look good and Democrats bad. To be clear, none of this is evidence in favour of China, the WHO, the animal-origin hypothesis, or Mr Fauci. What it is is reason to be suspicious about an all-Republican-politicians group that just happens to reach conclusions unfavourable to those; it would likewise be cause for suspicion if a group composed entirely of Democratic politicians reached the opposite conclusions.)

So what this (tentatively) looks like to me is a largely political thing, which is only tangentially interested in getting at the actual truth and whose bottom line was mostly written before it started. In order for its conclusions to be good evidence of anything for me, I would want reason to believe that their investigation was more honest than most partisan-from-the-outset investigations are. Do you know of reason to believe that? Or of other reasons why we should put any trust in their conclusions?

Replies from: ChristianKl
comment by ChristianKl · 2021-08-07T14:42:01.073Z · LW(p) · GW(p)

On the other hand, if what happened is that just the Republicans on the committee got together and decided that they would investigate the topic and write a report -- well, that looks much more like a politically-motivated thing which will only by coincidence tell us anything useful, and whose bottom line was already written before they began.

Any congressional report is political. Even if the report would only reflect a clear consensus of the Republicans for the lab leak, that in itself is evidence of political pressure building up. 

If there's any committee started where the Republicans can invite witnesses they for example can invite Hugh Auchincloss on the stand and asking him why he thought that funding the Baric&Shi would have been expected to go through the P3 framework.

Then the follow up question of whether he thinks that Fauci thought at the time that the PDF that Fauci sent him with the title "Baric, Shi et al - Nature medicine - SARS Gain of function" described any gain of function experiments. 

I don't see any way how Auchincloss could answer those questions in a way that wouldn't be clearly look very bad for Fauci. 

"Fauci's college thought Fauci thought the Baric&Shi paper is gain of function research while Fauci tells congress under oath that it isn't" would be a story that's simple enough for everybody to understand.

I don't really know how these things work, so I don't know how much evidence that is that there wasn't an attempt to form a bipartisan group of this sort; it seems like it's some evidence, though.

A bipartisan group would actually need buy-in from the democracts. The decision to task the US intelligence community which writing their report likely removed the possibility to push for a congressional commission on the topic because democrats would just say "lets wait for the US intelligence community". Tasking the intelligence community worked as a bipartisan way to get a report.

In order for its conclusions to be good evidence of anything for me, I would want reason to believe that their investigation was more honest than most partisan-from-the-outset investigations are. Do you know of reason to believe that? Or of other reasons why we should put any trust in their conclusions?

How about actually reading their arguments and evaluating the arguments based on their merits? I know that's a strange idea, but maybe it's worthwhile here. The report does contain some new information such as that which comes from US satellite surveilance.

It adds to the evidence about the Chinese fight against the virus before it was officially announced and thus puts the allegations into a form where China not answering them is more incriminating. 

Replies from: gjm
comment by gjm · 2021-08-07T23:26:49.703Z · LW(p) · GW(p)

Yes, any congressional report is political, but some are more purely political than others, which is why I pointed out that this one is produced by people all on one side of the aisle (which makes it likely to be more than averagely political) and considered how we might tell whether it's likely to be better than most things produced by people all on one side of the aisle (which it doesn't).

Your previous comment didn't look to me as if it was commenting on the level of political pressure, and to whatever extent it was it was therefore nonresponsive to its parent, which was commenting on whether in fact Dr Fauci did bad things, not on whether there is political pressure to admit/deny/insist/etc. that Dr Fauci did bad things.

Imaginary scenarios in which someone gives testimony before Congress that's damaging to Dr Fauci may be fun to imagine, but imaginary scenarios don't provide real evidence.

I can read the arguments in the report and evaluate them, but I can't so easily tell what arguments they aren't presenting because they don't lead the way they want, and I can't tell what evidence they aren't presenting because it points the wrong way, and I don't know how trustworthy the people they quote are, and if the report makes some claim about something that happened in China in 2019, or about alleged genetic evidence that SARS-CoV-2 is likely the result of engineering, then I can't readily tell whether that claim is true or false. So, alas, "just read the report and evaluate it for yourself" is a much less effective strategy than one might hope.

In any case, your great-grandparent comment didn't say "this report makes a strong argument, namely X". It just offered the report's conclusions as evidence:

He made a mistake. He circumvented US government policy to fund gain of function research under his watch. We are now at the HOUSE FOREIGN AFFAIRS COMMITTEEREPORT MINORITY STAFF concluding that the preponderance of evidence indicated that it was a lab leak due of a virus created via gain of function research.

I'm trying to understand whether there's actually any reason to think that this report's conclusions are any evidence of anything, or even that they're likely enough to be evidence that it's worth putting in the (substantial) effort it would take to tell whether the investigation and reporting are competent and honest, given that my prior for such things is that they are extremely unlikely to be so. I would expect a group composed entirely of Republican politicians to say Covid-19 was probably the result of a lab-leak of an artificial virus, and that China covered it up, and so forth, pretty much regardless of whether any of that is true, and pretty much regardless of the state of the evidence.

(I had a quick look at the report. It's 83 pages long and does indeed appeal to all sorts of alleged evidence I am not readily able to evaluate; my initial impression of its logic is not good; its authors seem to be excessively ready to jump from suggestive hand-waving to "so it is reasonable to conclude that ...". But I would expect it to be like that even if the conclusions are correct, because it is written by a partisan group of politicians and such groups are rarely very honest even when the claims they are making are true.)

Replies from: ChristianKl
comment by ChristianKl · 2021-08-08T07:48:07.815Z · LW(p) · GW(p)

If the House Republicans would have a clear position that the lab leak theory is true, why wouldn't the prewritten conclusion be "It's clear beyond reasonable doubt that the lab leak theory is true?" The fact that the settled on the preponderance of evidence standard suggests to me that there wasn't a clear prewritten conclusion. The claim is also more specific then just the lab leak happening. It's specific about the timing of early September. It wouldn't need that specifity for the political purpose.

I think you have a model where the House Republicans only care about what the media says that's inaccurate. Accross party lines there are powerful people in DC that don't want the lab leak theory out in the open. That's why the Trump administration never managed to publish a report like the one that the HOUSE FOREIGN AFFAIRS COMMITTEE REPORT MINORITY STAFF published. There was just to much opposition in his administration to do so even if Trump wanted it.

my initial impression of its logic is not good

What isn't good about the logic? If we take the unpublished database. Why unpublish the database at that point and then say it was unpublished during the pandemic because of cyber attacks?

Western money went into the creation of that database. It was created to help us fight pandemics and the Chinese took it down and didn't give access to it during the pandemic thereby making it harder to fight the pandemic.

I don't see what you mean with handwaving. What kind of evidence wouldn't be handwaving? I have the impression that you are partisan about the issue and there's no evidence that would convince you.

Replies from: gjm
comment by gjm · 2021-08-08T11:27:48.573Z · LW(p) · GW(p)

I am unconvinced by both of the arguments in your first paragraph. (1) To whatever extent the bottom line was pre-written, the expected result is that their conclusion would be as confident as they could convince themselves the evidence justifies and that others might believe. Everyone knows that a lot of what might be crucial evidence is unavailable; they couldn't e.g. credibly say "we know for sure that it was a lab leak". But they do say it was a lab leak, not just that it probably was. (2) For most readers, being more specific makes a claim more plausible rather than less. (As Gilbert and Sullivan had one of their characters say: "Merely corroborative detail, intended to give artistic verisimilitude to an otherwise bald and unconvincing narrative".) So neither of the things you describe seems to me unexpected if this is politically motivated and had a basically-prewritten bottom line.

I don't at all think that the House Republicans only care about what the media says, and I have no idea what I said to give that impression. The rest of your second paragraph sounds to me like pure conspiracy-mongering.

The specific example of handwaving and poor logic I was looking at when I wrote what I did about those is on page 26. "Therefore it is reasonable to conclude, based on the WIV's extensive sample library and history of genetically manipulating coronaviruses, that in early September, one or more researchers became infected with SARS-CoV-2 in the lab and carried it out into the city." The material in the previous pages doesn't constitute anything remotely like a justification of this claim, so far as I can tell.

For instance, they present graphs showing traffic around some Wuhan hospitals and Baidu searches for "cough" and "diarrhea" (presumably actually for Chinese equivalents of those, but they don't give details). The pattern of searches for "cough" and for "diarrhea" is very different, which to me suggests that they can't both be signals of the spread of Covid-19 (and in the time period they're talking about, unless I'm confused, there is no increase in searches for the latter; but there are increases before and a big spike after; I can't help suspecting that they are hoping to confuse). When writing about these graphs they refer to "the hospitals that show a rise in traffic with patients complaining of COVID-19 symptoms", suggesting that they have evidence that people at those hospitals were complaining of COVID-19 symptoms, but they've presented no such evidence.

This sort of thing is typical of writing whose goal is to persuade rather than to find the truth. It looks superficially like reasoning but when you look more closely all the actual logical steps are missing or broken. They show some graphs involving hospitals and COVID-19 symptoms, and then a paragraph later they are talking about "hospitals ... with patients complaining of COVID-19 symptoms", and if you aren't reading carefully you might think they have actual evidence of that. They show evidence that people at the WIV worked on coronaviruses, that they have some history of poor safety procedures, and that an online database of WIV samples was taken offline in 2019-09 -- all of which, I completely agree, is highly suggestive -- and jump from that to "it is reasonable to conclude that in early September one or more WIV researchers became infected with SARS-CoV-2". Sure, that's one possible story consistent with the evidence, but you need more than that to say something's "reasonable to conclude", and it looks to me as if they haven't even tried to justify their reasoning. Handwave, handwave, confident conclusion.

Your impression of me is symmetrically matched by my impression of you (on this specific issue). I hope at least one of us is wrong.

answer by Radu Floricica · 2021-08-05T04:43:47.231Z · LW(p) · GW(p)

The key moment was the realization that can spread easily.

We have much worse diseased by effect, like aids and ebola. Key thing is - we can limit the spread with moderate effort, even without vaccines. Once you could put together the belief that this one will not be contained, meant living in a different world and should have required adapting. At least internal.

answer by Dagon · 2021-08-05T18:02:32.178Z · LW(p) · GW(p)

Did you make a mistake?  What behavior do you wish you'd undertaken, and what is the prediction you failed to make which would have made it happen?

I started working from home a few weeks before my peers, and I got KN95 masks early.  And I had serious conversations to convince my folks to learn about grocery delivery and stop going to stores by early March.  But otherwise, mostly followed the general recommendations.

answer by ChristianKl · 2021-08-05T08:35:22.212Z · LW(p) · GW(p)

You extrapolated two times from a single example. That's generally bad when dealing with risky situation. Just because no catastrophe happened in one example from which you are extrapolating doesn't mean that there's no problem.

Part of the reference class for Covid was SARS. It's very unclear why the Flu which is a far more distant virus should be a better comparison then SARS. SARS causing long-term CFS was very worrying at the beginning of the pandemic. Fortunately, long COVID isn't as common and strong as CFS was for SARS but it was clear that long-term effects were something to worry about from the beginning.

The fact that pandemics were a big risk was a belief that a lot of people in our community held before COVID-19, to the extend that it was seen as a bigger GCR then unfriendly AI by many people in our census. Treating swine flu as evidence that pandemics aren't a huge problem seems to me generally bad reasoning about tail risk.  

answer by Unnamed · 2021-08-06T00:54:43.282Z · LW(p) · GW(p)
  1. It sounds like you were relying pretty heavily on the amount of alarm in the media as one of your main indicators of how much to worry (while using an interpretive filter). What you took from the swine flu example is that the media tends to be too alarmist, but you also/instead could've concluded that the media is not very good at risk assessment (and maybe isn't even trying that hard to do good risk assessments). The line of reasoning that this new virus is probably less dangerous than swine flu because it's less media hype depends on the assumption that the level of alarm in the media is strongly correlated with the level of danger (with a systematic bias towards exaggerated alarm); I think the correlation between media alarm and danger is not that strong in which case this argument doesn't go through. So, the media isn't that functional as an alarm, and you need some other approach for figuring out if there's a big problem. Really bad pandemics are possible, and the amount of alarm in the media isn't that strong an indicator of whether a new virus is likely to turn into a bad pandemic, so how could I tell if one is coming? You maybe could still use the media as an initial alert: the media is alarmed about this thing, and it is the sort of thing that has the potential to be really bad, so I'll take that as my cue to put effort into understanding what's going on (via some other approach that doesn't rely on the media). Or, you could try to be plugged into other information environments which are more reliable, such that you'd trust them more if they raised an alarm. I benefited from hearing things like this [EA · GW] and this, and similar things by word-of-mouth & ephemeral Facebook posts.
  2. It helps to think in terms of probabilities & expected values. Scott wrote about this in some depth in A Failure, But Not of Prediction. For example: If swine flu turned out to be unimportant after the media hyped it up, that gives reason to think that the probability that the next media-hyped virus will be really bad is more like 25% than 75%. But it doesn't give much reason to think that the probability is more like 1% than 25% - not enough data for that. And if you see the probability that a novel coronavirus will turn into a really bad pandemic as being as high as 25%, then it's worth investigating & preparing for.
  3. It sounds like a big part of your lack of concern is that you thought the illness wasn't that serious, so that even if the virus became widespread it wouldn't affect you much. My memory is that this is different from the reasoning of most people who weren't very concerned, as it was more common to think that the virus wouldn't become widespread. So (a) this mismatch seems like a clue that something might be up, and worth looking into. And (b) I think there were reasons to think that the virus would be a big deal if it became widespread, e.g. the lives of people in Wuhan had changed in pretty drastic ways as a result of the virus.
answer by DirectedEvolution (AllAmericanBreakfast) · 2021-08-05T19:36:53.480Z · LW(p) · GW(p)

Back in April 2020, I made an alarm bell for the next pandemic [LW(p) · GW(p)]. It attempts to capture in checklist form some of the qualities of COVID that, in retrospect, could have signaled that this was no Swine Flu. I think there are basically two ways we could have avoided this mistake:

  1. Understanding that we should be worried not only about certainties, but about risks. We were right to worry about the downside potential of Swine Flu, and we were just lucky that time. Likewise, we'd still be right to worry about the downside potential of COVID, even though Swine Flu didn't turn out to be as bad as we feared.
  2. Looking at the underlying mechanisms that distinguish a pandemic potential from a non-pandemic potential illness, as I tried to capture in the alarm bell.
answer by Stuart Anderson · 2021-08-05T14:42:47.526Z · LW(p) · GW(p)

-

6 comments

Comments sorted by top scores.

comment by mingyuan · 2021-08-05T04:09:10.024Z · LW(p) · GW(p)

I'm curious at what point you believed you didn't need to worry about it and at what point you updated. I think it's hard to answer this question without knowing that (although I don't personally know how to answer it anyway).

comment by lsusr · 2021-08-05T07:06:24.136Z · LW(p) · GW(p)

I want to commend Andrew Vlahos for identifying a place his beliefs need to be updated and seeking out improvement. This is what Less Wrong is all about.

comment by Pattern · 2021-08-21T05:58:44.684Z · LW(p) · GW(p)
How could I have done better?

Well, you don't know, what you don't know. Would you have found it less absurd if you had known more about polio? Other lung diseases? Other diseases that caused permanent brain damage?

comment by habryka (habryka4) · 2021-08-07T20:43:39.816Z · LW(p) · GW(p)

Note: I accidently moved this to drafts, and when I republished it it seems to have gone into "Latest" again. Sorry

Mod note: I moved the publication date back to its original date and removed this notice.

comment by player_03 · 2021-08-05T05:38:15.490Z · LW(p) · GW(p)

Back when you were underestimating Covid, how much did you hear from epidemiologists? Either directly or filtered through media coverage?

I was going to give an answer about how "taking the outside view" should work, but I realized I needed this information first.

Replies from: lsusr
comment by lsusr · 2021-08-05T07:04:19.169Z · LW(p) · GW(p)

In the early days of COVID, I listened to the This Week in Virology podcast (which is hosted by virologists) because I wanted to hear from experts. I was surprised by how they weren't much better than the mainstream media at predicting what would happen. This could be because they were virologists and not epidemiologists.