For experts who study coronaviruses, a grim vindication | Hacker News Reader

For experts who study coronaviruses, a grim vindication

undark.org

114 pointssohkamyung101 comments

Retrospectively you can justify anything. But the coronavirus outbreak happening now is not proof of the idea that the optimal thing to do in 2010 was to invest in coronavirus research- that could have still been the correct decision based off the probabilities of an outbreak. Just about every virus has a virologist convinced it's the next big one. In recent times next big ones have been the flu (kind of expected), HIV (basically out of nowhere), COVID (we could have done better, but if you'd asked me to bet I'd have bet on avian influenza/H5N1 since that's already dangerously close to becoming an outbreak and has been on radars for a while). Ig a super virulent hantavirus emerged, are those that studied those viruses noe "vindicates" as to pandemic threat? They infect humans and cause small outbreaks too.

See also Black Swans. If you call out 200 possible Black Swan events, and 1 comes true, people will call you a genius and ignore the 199 that didn't happen.

Pandemics are not black swans though. They occur at predictable intervals. Taleb explicitly says they’re not black swans in his work.

The lockdowns though and their scale were totally black swans with no precedent

There were widespread lockdowns during the 1918 pandemic and quarantines are a practice about as old and storied as urbanized human settlement. They’re not black swans any more than the pandemics themselves. Taleb has also talked about this quite a lot recently.

The voluntary shutdown of almost all major economies was definitely unexpected. Nobody alive today has ever witnessed anything like this, on this scale, before. You can debate the academic definition, but it's definitely a black swan in my book.

We’re not talking about academic definitions, but explicitly what the guy who invented the concept has said and how he’s used that understanding.

Again, in his book, he says that pandemics are not black swans. Taleb used this knowledge to advise a fund that saw massive returns betting that all of this would happen:

https://www.bloombergquint.com/business/taleb-advised-univer...

Yes, the pandemic itself is not the "black swan." Pandemics have happened before. But none of them caused a reaction like this. None of them caused voluntary shutdowns of all the world's major economies at the same time.

It's predictable if you think about it for a few minutes.

1. It's hard to keep outbreaks localized, thanks to widely available air travel -- that's pretty new compared to 1918, but extremely common knowledge.

2. The standard procedure, for centuries now, has been lockdown / quarantine.

1+2=widespread lockdowns

It's the way the whole western world responded in lockstep. And then did it again in solidarity with BLM. Marshall MacLuhan in the 1960s predicted that with the rise of "peer to peer electronic media" we would all be marching to the beat of the tribal drum. Here we are.

There were widespread closures of public accommodations, but there were no efforts made in 1918 to keep people from going outside or stop them from visiting friends and family.

No precedent within living memory is not the same as no precedent.

Aside from 1918, the quarantine of entire cities and counties has been a common tactic to control plagues from bubonic to yellow fever. There are even reports from the Black Death of some Italian cities delivering food to peasants to encourage them to stay home.

Not to mention the draconian quarantine of merchant vessels entering/leaving ports.

30+ days mandatory quarantine wasn't unusual.

The word Quarantine traces it's roots back to quarantino (40 days) and those laws where passed all over the place from (what we know) the 14th Century.

With the expectation that if the ship is infected, everyone on it will die, and no medical help will be given. Pretty brutal by modern standards.

It's not like they had a lot of medical help to provide in those days anyway.

Leeches.

Headache to much blood, leeches.

Stomach ache, too much blood to the stomach, leeches.

No precedent since the days of medicine being about the four humors of the body may as well be no precedent at all given how different the world is.

That just isn’t true. Aside from the fact that 1918 wasn’t exactly the area of “four humors”, one can’t decide that “unprecedented” falls within a convenient range for your argument.

Besides, if your argument is that precedent doesn’t exist because medical technology was different, consider that we’re dealing with exactly the problem that quarantine was designed to fix: an infectious disease that we neither have immunity nor treatment for. For this exact scenario, the precedent is both strong and relevant.

Here's Taleb on Black Swan and corona: https://www.youtube.com/watch?v=Tb2pXXUSzmI

As a counterpoint, if the black swans are big enough, then it is possible that a 1 / 200 track record is still useful.

How so? Predictions of black swan disasters are only useful if the prediction leads to investment into preventative measures, right?

He's talking about cases in which preventative measures could be taken regarding those 200 predictions (eg. 200 asteroids with high probability of crashing into earth, maybe we could track them, identify one coming towards us, and blow it up)

Even though, they were 200, the fact that one of them was identified correctly, saved human civilization as a whole.

The point being, 1 in 200 is not a bad success ratio per se

It's a bad success ratio if the 200 predictions didn't have good explanations for why they were made, because that means they are essentially 200 arbitrary predictions. In that case, the fact that one prediction came true is irrelevant.

To put it another way: if you're making arbitrary predictions without explanations, then you increase the likelihood of a prediction coming true by increasing the number of predictions you make. But clearly we shouldn't value predictors simply based on the number of predictions they make.

As a practical matter, if somebody has a 1/200 chance of being right when sounding the alarm, by the time something actually happens I think nearly everybody will have written them off as the 'boy who cries wolf.'

I'm into paranormal stuffs and really intrigued by psychics. But i haven't found a single claim from any psychic that predicted 2020. I think they are now below El Chupra Cabra in my totem pole

Diminishing returns is a thing, tossing another billion dollars at Cancer research is unlikely to make much of a difference. There are only so many diseases with a proven track record of significant human risk.

SARS was a huge sign that something like SARS-CoV-2 (coronavirus) was a significant risk. But, give it 10 years and I suspect research funding will once again be scarce until the next outbreak. It’s not about this disease so much as a reasonable response to a range of diseases like Ebola which easily and frequently make the jump to humans.

Cancer is common enough and bad enough that even a small difference matters. For comparison, even in the countries with the worst outbreaks, cancer is expected to kill more people than SARS-CoV-2 this year.

Cancer isn't one disease, though. We lump hundreds of conditions and causes under that descriptor. Counting cancer totals that way is more like counting SARS-CoV-2 and influenza and colds all together.

Epidemiologists have long argued that we over index on diseases like Ebola, who are too lethal to pose a pandemic risk. There have been 1,590 confirmed Ebola fatalities since 1976. Scary if you live where the outbreaks happen, but surprisingly small for how much we freak out about it.

Should we spend some resources on Ebola? Of course. But those resources should be spent in line with the risks as we best understand them. By 2004 we already had evidence that coronaviruses were much more dangerous than Ebola, with the 2003/2004 SARS outbreak killing half as many people (774) as Ebola has in nearly 50 years. Ignoring COVID-19, SARS and MERS (both coronaviruses) have killed 1,640 people, exceeding Ebola’s grim toll.

You seem to be very low on your Ebola numbers[0]. Is that based on a super-strict definition of "confirmed"?

[0] https://www.bbc.com/news/world-africa-28755033

Pulled from Wikipedia, which might've been a mistake.

> surprisingly small for how much we freak out about it.

It might be small _because_ we freak out about it. Kind of like New Zealand's COVID-19 numbers (or other countries with aggressive prevention / mitigation methods).

In comparing diseases we probably want to compare their transmission and mortality rates and long-term effects. If we compare their infection or death count then we're comparing the disease + the response together which muddies the waters a bit.

I'd like to add to to your statement.

The entire quote which this article branches on reads:

> “The presence of a large reservoir of SARS-CoV-like viruses in horseshoe bats, together with the culture of eating exotic animals in southern China, is a time bomb.”

I'd say that the second part of that statement carries more importance then the exact type of virus and the specific research on that type.

I'm no virologist, but if the barrage of news from the last few months have taught me anything, it's that the actions of humans are a big factor in increasing the probability of any type of outbreak. Sure, researching specific families and types of viruses gives us valuable insights. But in and of themselves, tying that knowledge into a larger interdisciplinary framework and translating that then into socio-economic policies, safety guidelines, legal frameworks,... is really important.

Going back to the quote above, this Vox piece does a great job explaining that side of the equation:

https://www.youtube.com/watch?v=TPpoJGYlW54

What's the baseline for similar predictions from credible sources? In retrospect, that's spot on, but we can't evaluate past prioritization without knowing how many other equally credible dire predictions were made that haven't panned out.

In other words, you think someone in China ate a bat and made the world sick?

My understanding of the extant research is that SARS-CoV-2 jumped from bats to pangolins to humans in wet markets in China that sold "farmed" wildlife.

In short, yes. Wet, wildlife markets in China were the source of this outbreak.

> "Retrospectively you can justify anything. But the coronavirus outbreak happening now is not proof of the idea that the optimal thing to do in 2010 was to invest in coronavirus research..."

you know what we can justify prospectively? any and all scientific research, including corona research and every past and future ig nobel nominee, over funneling billions of dollars into the pockets of the already wealthy, as our politicoeconomic system is currently set up to do.

that's not to belittle your point that you can't a priori know what research will "pay off", but rather to buttress the idea that we should be pouring many magnitudes more money into research so we're that much less likely to miss potentially valuable science in the face of the vast unknown.

Great response. Increasing investment in the "Science Index Fund" prepares us to handle risks like COVID-19 better without requiring the "active management / forecasting" ability to know what the probabilities of a COVID-19 outbreak would have been.

However, an indexing strategy still requires an estimate of the value of coronavirus research. Suppose $5T USD was funneled into all research over the last 10 years. How much of that would virology/epidemiology have gotten, and would it have been enough?

Should be all of it, considering this outbreak alone cost the global economy more than $5 trillion

This ignores the a priori knowledge problem. Sure, with perfect knowlege that there would be an outbreak in 2020, this investment might have made sense.

Without knowing the future, you can't make this statement.

i think it's really tricky because we aren't really smart when it comes to dealing with the future and/or counterfactuals.

risk averse investors (?) may want a lot to go towards prevention of catastrophic but perhaps lower probability things (e.g. pandemic, asteroid impact within 1000 years), others may want towards high probability things (climate change related), yet others may want to just solve nuclear fusion so that all of the above become way easier, etc. imagine if $5T concentrated solely in energy research got us some breakthrough leading to cheap/infinite energy, for example.

prediction market? i'm not really sure.

yes, the allocation problem is thornier, and valuing science properly more so, but it's pretty obviously working badly in our current politicoeconomic system. we can't both hoard wealth and power and expect good allocation. concentrating societal decision-making into fewer and fewer hands inevitably leads to bad and self-serving decisions.

with that said, once $5T (or whatever) is allocated to science, it's a matter for the decision markets in and around science to sort out allocation within. that mechanism has the best chance of coming up with the best available (but unlikely perfect) answer.

> Suppose $5T USD was funneled into all research over the last 10 years.

Hard to answer exactly, but we can get a decent answer. A quick quack (search)[0] shows that in 2013 the US spent $435bn in research (We'll round to an even $500bn) funding. So to answer your question:

> How much of that would virology/epidemiology have gotten, and would it have been enough?

Exactly as much as they already got and in exactly the same way it has.

But this answers your question too specifically because it is just all funding. Personally I'm in the view that if we're not in an active war we should be focusing more money into R&D. Science has a ROI of 1 year - a few hundred years, with 20 years probably being around the mean. A 10-30 year investment is not something I want my industry to focus around (or something I can expect industry to even spend money on because their incentives are not aligned for that). But it IS something I want my country to be focused on, because even 50 years for a country is not that long. I also want that knowledge in the public domain so businesses can form around these new things. Essentially the gov raises the tide and all boats benefit, even the people that are making new boats. I want my government to be funding basic scientific research because often there is no immediate benefit from this but the long term benefits are huge. Institutions like NASA and CERN don't spin off technologies because there is something unique about them, they spin off technologies because they are pushing the edges of technological capabilities and have to invent new components along the way, things they didn't know they needed. This isn't something I can expect a company smaller than FAANG to invest in.

[0] https://www.nsf.gov/statistics/2015/nsf15330/nsf15330.pdf

[1] Further breakdown: https://www.nsf.gov/statistics/2017/nsf17316/

Ideally you should build defenses against a broad range of virus families that could become dangerous. Including hantaviruses and coronaviruses. Also, there are synergies as demonstrated by remdesivir: antivirals developed for one virus family might help with other families as well.

Many emergent viruses are under close surveillance and animal vaccines are under development. Sometimes that alone makes economic sense already, e.g. when you want to protect expensive race horses from dying.

The danger of new viruses will only increase in coming years as we 1. travel far more than we used to, taking viruses with us and 2. go into the environments previously inhabited by wild animals like bats (which are a host for a large number of viurses).

> Retrospectively you can justify anything. But the coronavirus outbreak happening now is not proof of the idea that the optimal thing to do in 2010 was to invest in coronavirus research- that could have still been the correct decision based off the probabilities of an outbreak.

Right. Yes, a paper on coronaviruses described this as a "ticking time bomb". The interesting question is how many other papers around the same time used similar language for other diseases. Without that, you don't have any evidence that people made the wrong decision then based on what they knew at the time.

It's also possible that some of those other papers were also right, and those diseases just haven't hit yet.

Coronaviruses were considered a high threat because there are at least two of them that are highly infectious in humans (they cause a large percentage of “common colds”), and they’re prevalent in species that are high risk of transferring to humans.

I would have bet on an ebola variant. It's extremely deadly and about as transmissible as you can get for something that generally doesn't spread through coughing or sneezing. Fortunately it doesn't have the long incubation/asymptomatic period of Covid-19, which turned out to be the key to its success.

It would have to be a significant variant since Ebola is too deadly to spread quickly, it kills the host too fast.

That's what makes covid nightmare fuel, a virus that can be entirely asymptomatic in one person but hit like a train on another that has a long incubation/shedding period.

The comments here are weirdly anti-science. There've been plenty of discussions about coronaviruses being a potential threat over the years. Given the proximity of bats to major population centres in China, I don't see how anybody could say "Well, it's so unlikely that research is unnecessary". All I see here is modern economic systems and scientific institutions failing because there isn't an immediate benefit, yet again, despite there being clear signs of a threat.

Others have pointed out that other threats may have seemed just as likely, and that hindsight may be essential to pick out this one.

Now that we have hindsight about that, maybe the lesson is that it's worth it to find the money to research the hell out of every top-20 or top-200 pandemic threats. Spray-and-pray.

Another pandemic will come. With luck, it will be another century from now. With bad luck, it will be next year. We must be better prepared.

People are very apt to justify or condemn a decision based off of the result and not based off of the data available at the time the decision was made. Only one of these helps us make better decisions going forward.

It almost certainly went like this:

    Scientist: "Boss, it's only time for a disaster to happen"

    Policymaker: "Will it happen now?"

    Scientist: "No, but it will happen"

    Policymaker: "OK, then, remind me when it happens"

Scientist: "So we'll do nothing until it does?"

Policymaker: "No, but rest assured we'll blame you when it does happen"

Or one of these: "we did not know", "we did not have enough data to make a decision", "thanks to all the heroes at the frontlines", "we are doing a great job", "it's the previous guy's fault".

Actually I think focusing on understanding the human immune system would be just as important. Know the immune system and how it really works and you can prepare therapeutic defenses against new viruses because you understand what they do to the body and counteract their effect or diminish it.

I would like to see a Manhattan project level funded research program on the human immune system.

I think the fear of a terror virus was so great that a virus that didn't have such a bad quality was mentally turned into a virus that has this bad quality.

Does nobody question the premise that Sars-Cov-19 is actually a big threat? At least the current incarnation seems rather tame. The panic over it however is rather real.

I'd argue that a virus that can kill 500,000 (probably way more) people in a matter of months is a pretty big threat, and thats with most world economies shut down and actively trying to contain it. Imagining the damage that it could do without those containment strategies in place world wide, is why I'm convinced it is.

The flu kills way more than that and Corona will kill more than that.

You can see what happens without containment in Sweden. Given that countries with a bad health system probably need more measures.

The real difference is that we are informed broadly about the many deaths and this was not the case with the flu in our lifetime.

> The flu kills way more than that

The flu kills a lot of people over a long time, this is 500k in just a few months while everyone is actively trying to avoid spreading it. The flu definitely does not kill more than that.

Flu season is about 2-3 months usually and kills, depending on severity, 1-2 million people each year.

Flu season is NOT about 2-3 months usually.

https://www.cdc.gov/flu/about/season/flu-season.htm

The exact timing and duration of flu seasons can vary, but influenza activity often begins to increase in October. Most of the time flu activity peaks between December and February, although activity can last as late as May.

It also does NOT kill Millions of people per season

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6815659/

Until recently, the World Health Organization (WHO) estimated the annual mortality burden of influenza to be 250 000 to 500 000 all-cause deaths globally; however, a 2017 study indicated a substantially higher mortality burden, at 290 000-650 000 influenza-associated deaths from respiratory causes alone

I'll give you an honest answer, but well, it's obvious. Yes, it's a big threat, it's clear if you pay attention. It's killed more than 100k people, a lot more than an annual flue. It's a horrible death with many people not being able to be around their loved ones. There are lots more deaths not counted as covid-19, because they didn't get tested after death. If we look at total deaths vs normal years we are undercounting.

How many sr citizen homes do we need to wipe out 1/3 the population in a month or two before its a serious issue? Mortality rate in the us for people who have it tested via test is 5.7% - https://coronavirus.jhu.edu/data/mortality.

Heinsberg study suggests a mortality rate of about 0.36%. This is high, but not "a big threat". Given that with 0.36% and 15% infection spread in the general population (like flu) you look at ca 162k deaths in the US.

It is not even close 5.7% mortality. If it would be that lethal you would look already at over one million deaths.

Sweden is a good example of a country without a lot of counter measures and we see that it will end in the realm of 0.36% lethality (5,4k deaths). https://en.wikipedia.org/wiki/COVID-19_pandemic_in_Sweden

Sweden regrets their choices https://www.npr.org/sections/coronavirus-live-updates/2020/0...

That's not exactly what the article is saying tho. Also it doesn't change the 0.36% mortality rate and I might add the high age median of around 80. So we will likely see a lower average mortality afterwards.

5.7% was probably referring to fatality rate.

It is extremely irresponsible to say this is not a big threat. At this rate, the death toll of Covid19 in the US is up to as big as 4 times the seasonal flu. In some countries the mortality will end up being more than 0.8%. Almost 1 in every 100 people dying because of a virus seems like a big threat to me, and it is already clear that this will be way bigger than the Swine Flu.

What would you qualify as a big threat?

Remember, a large enough group of monkeys with keyboards will produce Shakespeare given enough time.

Or a broken clock is right twice a day.

Yup yup yup. Glad someone brought this up.

It would take a lot to prove that they actually had good foresight since it’s so likely that they just happened to be the ones who guessed right out of a large population of equally well informed people who guessed lots of other things using similar reasonings.

I think it's probably a strength of science that at any one time there's at least one person yelling about the dangerous of any possible event. It's also a strength that, as in the case of the LHC, the people who were shouting "this will cause a black hole" were drowned about by the ones who were being more realistic.

The philosophy of science precisely prefers knowledge which gives falsifiable, testable theories. A hypothesis indicating every possible outcome with little differentiation is worthless.

> The philosophy of science precisely prefers knowledge which gives falsifiable, testable theories.

Knowledge doesn't give falsifiable, testable theories. Falsifiable, testable theories give knowledge. That's why the philosophy of science prefers them.

I never hear this sentiment on HN when it comes to writing code. Only the other, less-noble job classes.

I definitely understood the parent comment differently than you did.

My read of it is that they are criticizing those predicting low probability events, and then cherrypicking the predictions that turned out correct as misunderstood predictive geniuses.

A similar event in tech is the analysis of a potential AI singularity event, which comes up plenty often. We'll only be able to tell who is right in retrospect, and plenty of folks call each other quacks over that.

You've never heard of the term "code monkey"? Jonathan Coulton even wrote a song about it.

Reply on news.ycombinator.com