The "mad scientist" problem

Matt Yglesias has an interesting tweet:

This image has an empty alt attribute; its file name is Screen-Shot-2021-01-15-at-6.01.30-PM.png

Last May, I said the following:

Occam’s Razor also applies to the lab release theory. We know that dozens of epidemics have come from viruses jumping from animals to humans without any “lab” being involved. Why construct an entirely new theory for this epidemic?

. . . Actually, the CCP would look far better (in a ethical sense) if the virus accidentally escaped from a lab doing valid and useful scientific research, rather than from disgusting “wet markets” that the CCP refused to shutdown. . . .

Of course it’s certainly possible the virus did escape from a lab during research on bat coronaviruses. I really don’t care.

In retrospect, that wasn’t my finest moment. I meant that I don’t care in an ethical sense, and I still hold that view. But Yglesias’s tweet raises an interesting issue. If it was a lab escape, then what are the public policy implications, if any? And I don’t know how to answer that question.

Over the years, I’ve argued that things like accidental nuclear war and bioterrorism were much bigger threats than global warming. (And I view global warming as a fairly big threat.) But I don’t know enough about science to know how policymakers should respond to the risk of bioterrorism.

However, I do know more about science than I did two years ago. As an analogy, before 2009 I thought Western policymakers knew how to handle the zero lower bound problem for interest rates. In 2009, I discovered that they did not, or at least there wasn’t a critical mass that knew what to do. Similarly, last year I found out that we were far less prepared for a pandemic than I had thought. Indeed, my perception of our preparedness seems to fall almost by the day, as I recently discovered our inability to deliver a vaccine to the public that has already been invented, tested, manufactured and distributed to states.

In the past, people who know more than I do told me not to worry about bioterrorism. But their arguments were not persuasive. Now I have zero trust in the public health establishment. I see no reason at all not to fear a “mad scientist” creating a virus 50 times more deadly than Covid-19, and letting it loose.

So to answer Yglesias’s question, it seems to me that if the virus escaped from a lab, then we should conclude that a future scientist with the same sort of psychological problem as that rogue Malaysian Airline pilot might someday unleash another Black Death. Say something as deadly as HIV, where symptoms show up with a long delay (as with HIV), and as easily transmitted as the flu. If you think I am wrong and are able to explain why this cannot happen, I’d love to be reassured on this point.

One thing I know for sure; if something bad can happen, at some point it almost certainly will happen.

Categories: Economics of Health Care Technology Violence and War


Mark Z
Jan 15 2021 at 7:11pm

I guess I’ll give it a try: there’s generally a Darwinian tradeoff between transmission and mortality, so a bioterrorist might design an extremely deadly pathogen that would kill a bunch of people at first but wouldn’t spread very far or fast (and would be fairly containable because mostly you’d containing corpses), or a highly contagious but not very deadly disease. Given how rapidly pathogens evolve, there may be reason to believe it’d be very difficult to outperform what natural selection has already bequeathed us when it comes to pathogens. It may be even conceptually extremely difficult to engineer a pathogen that’s very contagious, very deadly, and not readily treatable, so the limits on how bad bioterrorism can be may be less severe than our imaginations lead us to believe.

That’s the genus of argument I find most plausible, no idea if it holds much water.

Nicholas Decker
Jan 15 2021 at 7:59pm

I really must concur with Mark here. To put it into terms you fellers on this site might grok easier, it’s the difference between centralized and decentralized planning. The actions of millions, if not billions, and viruses far exceed anything that we humans could design.

Scott Sumner
Jan 15 2021 at 11:12pm

Mark,  You said:

“there’s generally a Darwinian tradeoff between transmission and mortality”

I do understand why this applies to natural pathogens, but I fail to see why it must apply to a man made pathogen.

You said:

“Given how rapidly pathogens evolve, there may be reason to believe it’d be very difficult to outperform what natural selection has already bequeathed us when it comes to pathogens.”

I don’t find this argument to be very persuasive.  Nature has already given us very dangerous pathogens like Ebola.  Covid-19 has already infected hundreds of million of people, and it is evolving to be even more contagious.

Also, science is making incredibly rapid progress in this area.  Who knows what we’ll be able to engineer in 20 years?

Sorry Mark and Nicolas, but your arguments make me even more worried, not less.

Mark Z
Jan 16 2021 at 12:29am

“I do understand why this applies to natural pathogens, but I fail to see why it must apply to a man made pathogen.”

Because the more likely (or sooner) a pathogen is to kill its host the less likely it is to spread (corpses rarely make good hosts), and contagion and health hazard are both a function of virulence (among other things). The more a virus reproduces within a person, the more likely it is to kill them, all else being equal; so you can make a virus that’s highly reproductive, so the host sheds lots of virus (or bacteria) and spreads, but it has to be relatively benign for the host not to succumb to the high viral load.

Maybe you could make a pathogen that consistently kills its host after a few months, giving it plenty of time to spread, but how exactly do you make a virus mild enough not to incapacitate the host so it can spread at first, then time it to later become much more virulent? How does it escape the immune system? HIV is so good at this because it has such a high mutation rate, but you can’t control a pathogen with a high mutation rate, it’ll rapidly succumb to natural selection however you design it at first.

The point being there are biological constraints that can’t we can’t necessarily fiat away any more than the obstacles to flying cars. We may yet find ways around them, but I’m not convinced such innovations are just around the corner at least. Here medical pessimism I guess translates into optimism: we still struggle to design a molecule to inhibit another molecule, even when we know the exact sequence and structure of the target molecule. To design an organism to do what you want even initially is much harder, let alone after hundreds of generations (perhaps a few weeks in human time)? Very very difficult to say the least.

john hare
Jan 16 2021 at 5:25am

I believe Scott’s point is being missed. He postulated The Mad Scientist Designed Virus with both a high infection rate and serious delay before symptoms become obvious. To say it can’t be done ignores the point that a rogue individual just might be able to find a way. The real question becomes “what happens next?”.

To say that the government responses world wide have been disappointing would be an understatement.

In unrelated fields, I have stumbled on dangerous concepts* that individuals/tiny groups more capable than myself could develop that seem to be unknown to the mainstream.

*though not as dangerous as a designed pandemic

Scott Sumner
Jan 16 2021 at 2:13pm

Yes, I think he missed my point.  I understand that if a virus quickly kills its host then it cannot easily spread, that’s why I mentioned the HIV example.

Is there some technical reason why HIV cannot be modified to spread easily?  Perhaps that’s what I’m missing.

The Black Death of 1348 actually happened, so we know that nature can produce pathogens that are both deadly and easily transmitted.  So can humans also produce this sort of virus?  I don’t know.

Thomas Sewell
Jan 15 2021 at 9:54pm

If it escaped from a lab, rather than being a “marketplace” accident, a lot more people are going to be considering if it was really an “escape” rather than a “release”.

Not saying they’d be right, just saying there will be folks whose conclusions shift in that direction. A deliberate release (first a bit on their own people, to avoid suspicion and get it spread world-wide) is a more viable conspiracy theory once they admit it this strain was being worked on by government scientists in the location of the outbreak in the first place.

Scott Sumner
Jan 15 2021 at 11:03pm

I find that theory to be extremely far-fetched.  I can’t even imagine why they’d want to do some thing like that, and if they did they’d have released in in another country so that there’d be no suspicion that they were involved.  But again, what possible motive?

In reality, the risk for something like that comes from an individual or a small terrorist group, not a nation-state.


Warren Platts
Jan 16 2021 at 3:19pm

If it were released in another country, it would not take long to figure out that the root virus came from Chinese bats.

As for state actors versus lone wolf terrorists, think about what is involved (the MH370 pilot did not build a B777 in his private laboratory). First you have to find a virus in the wild that is ameanable to selection to make it a good human bioweapon. At a minimum, you’d have to do serial passaging experminents (probably involving hundreds of ferrets). This stuff is expensive, requiring millions of dollars even in China.

But of course, if you REALLY wanted to make a good bioweapon, you would do serial passaging experiments on human subjects. (Petrovshy’s result — that the virus is better adapted to humans than ferrets or other animals — would be explained by this hypothesis (although guys like Andersen & Daszak will quibble). Such atrocities could only be done within a hyper-securitized, surveillance state.

Nor is there a real dichotomy between “mad scientist” actors and nation state actors. The reality is the most likely mad scientists are already working for the state actors.

As for motives, in geopolitics, what counts is relative power. Clearly, China has benefited by that metric. There are news stories that China will surpass the USA as the world’s biggest economy 5 years sooner than was previously predicted, thanks to the SARS2 virus. That should be especially chilling: even if the virus was a totally natural occurrence — no labs, no bioweapons — it shows that countries can benefit enormously from pandemics, relative to other countries.

Scott Sumner
Jan 17 2021 at 1:47pm

You said:

“If it were released in another country, it would not take long to figure out that the root virus came from Chinese bats.”

You miss the point.  The claim is that it’s a man-made virus.  The fact that the first Covid came from Chinese bats is beside the point; it could have been modified in a lab in any country in the world.

Yes, “mad scientists” might get needed resources from a government, but my point is that the danger comes from the technology.  Whether it’s a government lab or a private sector lab is a second order consideration; neither governments nor private companies want deadly viruses to escape.

Peter Gerdes
Jan 16 2021 at 5:45am

But finding out that the virus escaped accidently from a lab doing valid research shouldn’t really affect your probability that a mad scientists would deliberately create such an awful pathogen.  Sure, if you had reason to believe it was a deliberate release than maybe but that seems crazy unlikely even conditional on a lab release.

You might suggest it provides evidence that biology labs lack sufficient security/safety measures to stop a scientists from releasing an engineered strain.  However, we already have plenty of evidence of sloppy lab practices (e.g. the lab in UK where a virus escaped through the drain).  Also there is plenty of evidence that any scientists working at a respected biological research institution can simply order some pretty dangerous samples and then develop them outside of a high biosafety level facility.  Also the fact that a virus can rarely slip out using the normal disposal procedures doesn’t really tell you much about the ease of deliberately smuggling it out.

So yes, maybe worry about deliberately engineered virus but Yglesias is still correct about not making a difference in this case.

Scott Sumner
Jan 16 2021 at 2:20pm

It’s about trust.  If it was a lab creation that escaped, then these labs should probably not be doing this sort of research.  Any potential gains in terms of science are trivial compared to the costs.  So if it were a lab escape, then why not ban such research in US labs?

I’m not saying there should be a ban, it’s just that my faith in the pronouncements of scientists is getting lower each day.  Early in the pandemic, the scientists said this virus was of natural origin.  Now they say it’s probably of natural origin.  Fine, I have no reason to question this new claim.  What I wonder is why have they changed their tune?”  And if it was lab created, should the lab have even been doing this sort of research?  Accidents happen.

Again, I have no answers, but I’m not entirely satisfied with the answers I’m being given.


Jan 16 2021 at 5:55am

To answer Yglesias’s question, if it turns out that the virus escaped from a Chinese lab, then it may trigger actions around enforcement and verification of Chinese compliance with various bioweapons conventions and bioresearch practices.  Or, following Scott’s point, the international community might demand more access to and oversight of Chinese labs to reduce the likelihood of a mad scientist going rogue.

The CCP does not want to invite more scrutiny around their biolab practices.  For example, such scrutiny might involve international inspectors/visitors who might, in turn, discover other violations or learn state secrets.  They would rather have people poking around their food markets than their biolabs.  That’s why the CCP has plenty of motive to cover up any problems with their labs.

Scott Sumner
Jan 16 2021 at 2:25pm

BC,  I’m far more worried about “mad scientists” in the US than in China, as I believe our culture is more likely to produce this sort of behavior.  The unibomber was an American, for instance.

I’m not worried about the Chinese government using viruses as a weapon, I’m worried about the Chinese (and more so the Russian and American governments) using nuclear bombs as a weapon.

For private individuals, I fear bioterrorism not nukes.

Jan 16 2021 at 4:26pm

You have mentioned enforcement, inspection, verification, oversight, and scrutiny. This suggests a negligence presumption. It seems authoritarian regimes are already enthusiastic about activities like enforcement and verification and scrutiny and so on. That’s not to say there cannot be negligence, just that without more info, I’m inclined to be more concerned about a systemic problem.

Specifically, BSL 4 is the highest level of security that exists today but perhaps that just is not good enough. We may need more security innovations.



Jan 16 2021 at 9:08am

Say something as deadly as HIV, where symptoms show up with a long delay (as with HIV), and as easily transmitted as the flu. If you think I am wrong and are able to explain why this cannot happen, I’d love to be reassured on this point.

Long delay (like HIV) but no effective treatment possible (unlike HIV), but easily transmissible (even while the viral load is low during the long delay)?  And do we imagine the mad scientist could perfect such a never-seen-in-nature sort of virus without trial and error?  Or that he could conduct a series of experiments without notice (and remember the long-delay requirement, which means any such experiments could not be carried out quickly)?  This scenario, to me, seems like a risk that’s way out on the tail somewhere.

Scott Sumner
Jan 16 2021 at 2:38pm

You said:

“but no effective treatment possible (unlike HIV)”

That’s beside the point, as it took years for treatments for HIV to be developed.  So that’s not the issue.

The Black Death killed 1/3 of Europe, so nature does produce diseases that are both widespread and deadly.  Yes, we can now treat the bubonic plague, but what if something new becomes widespread before we develop a treatment?

This is why it matters if Covid was created in a lab.  I’m not saying it was  (probably it wasn’t.)  But Yglesias asks us to contemplate how our views would change if we found out it was created in a Chinese lab.  Well, at a minimum we’d know that a scientist created a virus, accidentally let it lose, and it killed millions of people.  If that were a true fact, then I don’t think it’s a stretch to think people might start worrying about a man-made virus created in a lab that could kill hundreds of millions.  The issue would not be China; the issue would be labs.

If it’s true that scientists are unable to create something like Covid-19 (which scientists seemed to be suggesting back in March) then there’d be less need to worry.

Jan 16 2021 at 3:42pm

Well even if it was created in a lab and there are some small risks of accidents or mad scientists releasing it, we could very well still want to continue such research. For example, what if the next black death plague is naturally occurring, but our extensive research on various diseases has allowed us to very quickly create a vaccine against it? Even Covid19 seems to be a pretty good example of that! We have to weigh the potential benefits of such research versus the cost, and it seems likely that greater security to prevent mad scientist leaks and/or accidents could be the best solution rather than stopping research. Although I have no idea how to quantify the potential costs and benefits.

Scott Sumner
Jan 17 2021 at 1:53pm

That’s why I have an open mind on this.  We can do research on vaccines without creating new viruses.  Maybe creating new viruses speeds up vaccine development, I don’t know.  I am not questioning the utility of virus research, I am asking “what is the marginal utility of the part of vaccine research that involves creating new viruses?”

And if we must create new viruses, why not in a facility in the middle of the desert, where researchers must quarantine for several weeks before moving back into society?  Maybe one in Nevada and one in the Gobi Desert.  Normally we’d view that level of caution (correctly) as too intrusive, but if a recent lab accident killed millions?  Now it looks more sensible.

Jan 18 2021 at 11:17am

Right, I was basically assuming it was necessary to create new viruses to come up with effective vaccines or treatments for them but I don’t know. I agree that putting the labs in quite isolated areas would be a good precaution. And on that same topic, building nuclear reactors in the middle of nowhere would also be a great idea to minimize the potential harm from an accident! I expect even though the risk would then be minimal, opposition groups would still find a reason to complain about the tiny risk of making an empty, practically uninhabitable desert…uninhabitable.

Jan 17 2021 at 2:23pm

I think, there is a bit of a conundrum here. The risky research that the Wuhan lab was doing is the “gain of function” research. As I understand it, they take a virus that is not dangerous to people or which is not easily transmissible and see how they can “improve” it. This is incredibly dangerous! On the other hand, it seems like the most straightforward way to learn how the deadliest viruses become this way, now naturally and potentially with the help of rogue scientists, and then use this information to design countermeasures. Maybe combination of lab created monsters with a rapid vaccine development can provide us with the best preemptive defense. The risk of accidental escape is than just a necessary risk to make the whole thing work.

Jan 18 2021 at 7:56pm

The difference would be very important, in that we would worry a little less about wet markets and more about labs, which is probably appropriate given the amount of concern raised by scientists in the past about the same type of research.

If your point is that China looks equally bad in either scenario I don’t care to argue about it.

Comments are closed.

Search Econlib