Why I do not favor variolation for Covid-19

Robin Hanson makes the strongest case for variolation, here is one excerpt:

So the scenario is this: Hero Hotels welcome sufficiently young and healthy volunteers. Friends and family can enter together, and remain together. A cohort enters together, and is briefly isolated individually for as long as it takes to verify that they’ve been infected with a very small dose of the virus. They can then interact freely with each other, those those that show symptoms are isolated more. They can’t leave until tests show they have recovered.

In a Hero Hotel, volunteers have a room, food, internet connection, and full medical care. Depending on available funding from government or philanthropic sources, volunteers might either pay to enter, get everything for free, or be paid a bonus to enter. Health plans of volunteers may even contribute to the expense.

Do read the whole thing.  By the way, here is “Hotel Corona” in Tel Aviv.  Alex, by the way, seems to endorse Robin’s view.  Here are my worries:

1. Qualified medical personnel are remarkably scarce right now.  I do not see how it is possible to oversee the variolation of more than a small number of individuals.  Furthermore, it is possible that many medical personnel would refuse to oversee the practice.  The net result would be only a small impact on herd immunity.  If you doubt this, just consider how bad a job we Americans have done scaling up testing and making masks.

The real question right now is what can you do that is scalable?  This isn’t it.

I recall Robin writing on Twitter that variolation would economize on the number of medical personnel.  I think it would take many months for that effect to kick in, or possibly many years.

2. Where will we put all of the Covid-positive, contagious individuals we create?  What network will we use to monitor their behavior?  We have nothing close to the test and trace systems of Singapore and South Korea.

In essence, we would have to send them home to infect their families (the Lombardy solution) or lock them up in provisional camps.  Who feeds and takes care of them in those camps, and what prevents those individuals from becoming infected?  What is the penalty for trying to leave such a camp?  Is our current penal system, or for that matter our current military — both longstanding institutions with plenty of experienced personnel — doing an even OK job of overseeing Covid-positive individuals in their midst?  I think not.

Under the coercive approach, what is the exact legal basis for this detention?  That a 19-year-old signed a detention contract?  Is that supposed to be binding on the will in the Rousseauian sense?  Where are the governmental structures to oversee and coordinate all of this?  Should we be trusting the CDC to do it?  Will any private institutions do it without complete governmental cover?  I don’t think so.

If all this is all voluntary, the version that Robin himself seems to favor, what percentage of individuals will simply leave in the middle of their treatment?  Robin talks of “Hero Hotels,” but which actual hotels will accept the implied liability?  There is no magic valve out there to relieve the pressure on actual health care systems.  Note that the purely voluntary version of Robin’s plan can be done right now, but does it seem so popular?  Is anyone demanding it, any company wishing it could do it for its workforce?

3. The NBA has an amazing amount of money, on-staff doctors, the ability to afford tests, and so on.  And with hundreds of millions or billions of dollars at stake they still won’t restart a crowdless, TV-only season.  They could indeed run a “Heroes Hotel” for players who got infected from training and play, and yet they won’t.  “Stadium and locker room as Heroes Hotel” is failing the market test.  Similarly, colleges and universities have a lot at stake, but they are not rushing to volunteer their dorms for this purpose, even if it might boost their tuition revenue if it went as planned (which is not my prediction, to be clear).

The proposal requires institutions to implement it, yet it doesn’t seem suited for any actual institution we have today.

4. Does small/marginal amounts of variolation do much good compared to simply a weaker lockdown enforcement for activities that involve the young disproportionately?  Just tell the local police not to crack down on those soccer games out in the park (NB: I am not recommending this, rather it is the more practical version of what Robin is recommending; both in my view are bad ideas.)  Robin’s idea has the “Heroes Hotel” attached, but that is a deus ex machina that simply assumes a “free space” (both a literal free space and a legally free space) is available for experimentation, which it is not.

5. Society can only absorb a small number of very blunt messages from its leaders.  You can’t have the President saying “this is terrible and you all must hide” and “we’re going to expose our young” and expect any kind of coherent response.  People are already confused enough from mixed messages from leaders such as presidents and governors.

6. There is still a chance that Covid-19 causes or induces permanent damage, perhaps to the heart and perhaps in the young as well.  That does not militate in favor of increasing the number of exposures now, especially since partial protective measures (e.g., antivirals, antibodies) might arise before a vaccine does.  This residual risk, even if fairly small, also makes the liability issues harder to solve.

7. The actual future of the idea is that as lockdown drags on, many individuals deliberately will become less careful, hoping to get their infections over with.  A few may even infect themselves on purpose, one hopes with a proper understanding of dosage.  One can expect this practice will be more popular with the (non-obese) young.  The question is then how to take care of those people and how to treat them.  That debate will devolve rather rapidly into current discussions of testing, test and trace, self-isolation, antivirals, triage, and so on.  And then it will be seen that variolation is not so much of a distinct alternative as right now it seems to be.

8. The main benefit of variolation proposals is to raise issues about the rates at which people get infected, and the sequencing of who is and indeed should be more likely to get infected first.  Those questions deserve much more consideration than they are receiving, and in that sense I am very happy to see variolation being brought (not much risk of it happening as an explicit proposal).  That said, I don’t think Heroes Hotel, and accelerating the rate of deliberate, publicly-intended infection, is the way to a better solution.

Soon I’ll write more on what I think we should be doing, but I would not put explicit variolation above the path of the status quo.


What's the URL for the Robin Hanson vs. Gregory Cochran debate on this topic from a few days ago?

Cochran's view was that it probably wouldn't turn out to be a good idea, but it's worth it for somebody to think about how to do it just in case more promising-sounding ideas all fail.


It's nearly impossible to find educated people in this particular topic, but you sound like you know what you're talking about!


Excellent post. The idea has merit as a concept, but an actual implementation has many difficulties when scaled to millions of people.

Maybe... but so does every other thing that is being done. Tyler's argument is, "There's a lot to consider, so let's not do it."

Here's just a stab at it:
1) actually allow this to be done in the first place (recall the problem with testing was that it was basically not allowed to be done)
2) create some sort of public-private partnership and have the government pay a company to do this and provide auxiliary support for quarantine measures
3) you could eminent domain a hotel (or just rent it out, no one is staying there now anyway) in a rural area not far from more populated areas and set up a temporary hospital next to it. Have a helicopter ready to fly seriously ill patients to more well-equipped places. (Note: you wouldn't have to impose any social distancing, so allowing the gym to be open or providing pool and ping pong tables as well as streaming and gaming services in communal areas (in addition to individual access to streaming and gaming)).
4) Post guards on each floor and at all exits.
5) Hire a catering company or two to provide food.

There are surely kinks to work out. It would probably be better to have the Army provide guards and catering (they don't do too badly on those kinds of logistics), but it's not out of the realm of possibility to have them provide all the services you would need for this. They could easily do a pilot test on one of the bases using volunteers and giving the very high hazard pay. Just team up with a clinical trials company to collect the data.

Logistically, this is not that hard. When Tyler says qualified personnel are scarce, he just means in certain areas. There are several medical professional from less-affected areas going to harder hit areas to aid those places. Again, there are plenty of qualified medical personnel in the military that could be used. Hell, you could probably do most of this on a military base.

I don't really care much about either political party, but right now, many failures of the state are mostly failures of leadership, i.e. Trump. Had he intervened with the testing issue earlier, chances are it would have been fixed. All the military top brass don't want to lose their positions due to any subordinate's criticism of Trump, so they silence dissent from air craft carrier commanders. So what you want right now are ideas that Trump's advisers can tell him, "This has a chance of ending the crisis sooner. We can try it out on say 500 young people and it will take 4 weeks to find out if it works. We could just have military personnel volunteer (with the benefit of large hazard pay) so no civilians would be put in danger."

Swapping 'Hero Hotels' for cruise ships would appear to simplify a great deal.

That's a good idea. Those cruise ship companies are looking for additional revenue and they have the logistical wherewithal to do it.

Cruise ships also have the advantage that they can easily be located in international waters to have the process happen outside of the legal barriers of most nation states.

It's an impractical idea.
If you're seeking widespread immunity, far better to concentrate on the various ways vaccine development could be speeded up.
Gates has shown how to prepare for bulk manufacturing; the clinical trials process needs thinking about, as this is now the most time consuming element.

In order to accelerate human trials Gates and Company may borrow Hanson's idea to inoculate 1000 volunteers at a time and put them up in 'Hero Hotels' or Hospital/Cruise ships for observation and treatment if necessary.

If you have 15 minutes of effort to spare, put it into reducing the rate at which the infection spreads now. That's where the energy should be going. Contain it, then eliminate it, and be vigilant while returning to the previous way of life as much as is consistent with safety. There will be no need to intentionally infect people with the risk, time, and resources that entails if this is done.

Masks and gloves are the most practical things to do but it's not sexy. The contrarians love this topic. It's catnip to them. They have this deep inner need for validation by taking an idea that most people naturally dislike and then work it into a state that gains some marginal acceptance somewhere. That's victory for them. To distinguish themselves from the masses and feel a little superior along the way.

Finding a way to feel righteous anger regardless of whether or not it is righteous is what seems to be important. For example, the internet used to be full of "daddies" telling us foolish children electricity grids could never handle more than 20% of electricity being generated from wind and solar, but then they suddenly all disappeared. It wasn't that long ago either. I hope they're okay and weren't all wiped out by a rampaging wind farm.

It is mean to point out just how many loyal MR commenters keep talking out their arse. And it isn't as if they are ever going to stop, because they know better than to believe lying facts.

I doubt they said it was impossible, just costly. And they are probably correct. You are essentially relying on 100% back up from fossil fuel to make a fully renewable grid work, which essentially means you have paid twice for the same capacity. Even if batteries get more widespread they will cost as well.

No, there were posters here insisting that it was impossible to have a stable electric grid with more than 30% renewable energy. Because of whatever reasons they thought reasonable at the time, at least for those who know better than to believe lying facts.

This is only the case if all that we are doing results in no second wave. The Spanish flu stopped because the survivors developed some immunity and it didn't spread anymore. Given that the risk of a second wave seems likely if you lift restrictions, you have at least two options: 1) keep everything shut down like it is now until a vaccine is developed (optimistically this is apparently 18 months away) (not very appealing); 2) you figure out how to make enough of the population "safely" infected and develop herd immunity as quickly as possible so a second wave doesn't hit when you lift restrictions.

A big part of the population had immunity at the start of the Spanish flu. My understanding is that many victims died rather quickly.

I hope for antivirals in the short run. Or that enough of the population gets a mild form. And the rate of new infections slows so that medical supplies and know interventions can deal with caseloads. Other than that wear a mask, wash your hands, keep a distance.

That is certainly not my understanding of the Spanish flu. That around 650,000 people died suggests there was not widespread immunity, and that it lasted for over a year suggest that there was no herd immunity to keep it from spreading. It was much deadlier for 15-44 year olds than previous epidemics, as well. By all means, wash hands, wear masks, etc., but the risk of a other infectious waves when restrictions are lifted is very real. The longer the social distancing measures remain in place, the more economic damage it will do. There's no reason not to try variolation. Seems like something Elon Musk would be willing to fund.

Lack of pre-existing virus-specific and/or cross-reactive antibodies and cellular immunity in children and young adults likely contributed to the high attack rate and rapid spread of the 1918 H1N1 virus. In contrast, lower mortality rate in in the older (>30 years) adult population points toward the beneficial effects of pre-existing cross-reactive immunity. In addition to the role of humoral and cellular immunity, there is a growing body of evidence to suggest that individual genetic differences, especially involving single-nucleotide polymorphisms (SNPs), contribute to differences in the severity of influenza virus infections. Co-infections with bacterial pathogens, and possibly measles and malaria, co-morbidities, malnutrition or obesity are also known to affect the severity of influenza disease, and likely influenced 1918 H1N1 disease severity and outcomes.

Seems like the cut off of 30 is a bit arbitrary as fatality rates were higher through at least the mid-40s in age relative to previous epidemics (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3291398/figure/F2/) and through those aged late 40s relative to later epidemics. The weird thing is the W shape of the mortality curve. I doubt that's due to some differential immunity between those aged 1-14 and those between 15 and 45.

The hypothesis of previously formed immunity is based solely on comparing 1918 mortality rates across ages rather than comparing mortality rates across flu seasons (not based on actual samples). I don't put too much stock in findings that say something is "likely"; likely compared to what? What probability relative to other hypotheses.

But I think where I disagree with you most is your characterization of the potential immune population as a "big part" of the population. Those over the age of 45 made up a small share of the population at that time: https://www.census.gov/dataviz/visualizations/055/.

I should have put quotation marks around my previous post.

Immunity for older people was speculated to be from a previous pandemic, I think a Russian one in the 1890s. Death was very rapid, often within three days. A family friend actually had a grandfather who treated soldiers during this period. He was treating, best they could, 500 patients a day according to his papers. Many of the over age 30 healthcare workers did not fall victim in great numbers to the Spanish flu. Other research that I have read said the quality of nursing heavily affected outcomes. This required nursing staff that was to a significant degree immune. If you prefer, replace "big part" with "significant"

Tyler, there is one crucial parameter you need to address: how much less severe the outcome is if purposefully infected with low dose. For smallpox it’s an order of magnitude lower. How much less severe would it need to be for it to be a good idea? At a certain reduction you would be in favor of it. So what does this parameter need to be for you to support it?
The question shouldn’t be if you support the approach or not. It should be if you support the experiment to understand its severity.

Tyler won't answer this because there is no answer. As far as we can tell, initial dose of CoV-SARS-2 is irrelevant. Small pox comes in two varieties, one much less deadly than the other. Actual variolation is not about the dosage, it's about using the less deadly form of the virus. There is no less deadly form of CoV-SARS-2. This entire discussion is unbelievably muddle-headed.

Exactly. All assuming what they think works for smallpox works for every virus.

We also have data showing large effects for two other viruses. We don't have any data showing small or no effects of other viruses.

I've no oar in this water, but plenty of epidemiologists seem to believe that smaller dose will often (usually?) produce milder symptoms. ceteris paribus.

Plenty? Name a few. Seriously, I'd love to see what their evidence and reasoning is.

"Actual variolation is not about the dosage, it's about using the less deadly form of the virus"

Infection with variola minor would surely work, but, pre-Jenner variolation was not so controlled, and often consisted of infecting oneself by rubbing a variola scab into a small cut made on the upper arm.

And this means of becoming infected made all the difference as compared with acquiring smallpox via respiration, as it allows the body to mount a robust immune response while the viral load is still low enough for one's immune system to manage.

Variolation was still not a decision to be made lightly, as its mortality was about 3%. Which is better than the traditional 1-in-3 mortality from variola major, but (of course) it had to be done community-wide or not at all, and to be effective it had to be done before most of the population had been exposed naturally to the disease.

Perhaps if we still lived in mostly isolated communities ...

Interesting. I wonder whether studies have been done to determine the outcome of covad-19 infection through a skin scratch. Apparently, we already know its morbidity through entering the mouth, nose, and eyes.

Has Robin Hanson self infected yet?

No, and if you understood his proposal, you would understand why not: he doesn't have hospital or medical support, he's in a region where there's substantial corona, it's not being done as part of a widespread program so makes minimal contribution to herd immunity, and he's not young.

One is reminded of Bertrand Russell's comment on Xenophon.

All these gainsayers seem to be ignoring the voluntary bit. If people decided tge risk wasn’t worth it no one would do it. But almost certainly there are healthy individuals out there who are actually looking to get infected to get it over with. Why not harness them to get some decent data? If it turns out that this is pretty harmful for even young healthy people it will probably save lives in future since we will have better data to asses the risks.

I wonder/doubt if we know enough about it? Would an ethics panel approve of such an experiment? (There are plenty around, nearly all (Bio) research Univiersity has them). The problem with subjecting someone else to an unknown risk should be obvious - there is a floor of what needs to be known in order to undertake such an experiment. How many of the 18-29 year olds (say) that have recovered from covid-19 have subsequently been autopsied? Yeah, not very many. So, it's speculation about what organ systems might have significant (long-lasting/permanent) damage. Not to mention reproductive systems which can take years and years (and/or generations) for problems to become apparent.

Maybe so, but that’s just another problem he needs to solve: How to sell, “We, the teleworking class, desire you, the nonteleworking class, to take risks that we’re not willing to take ourselves.”

If the goal is to give health care workers immunity then you'd be taking them offline for 2 weeks....which might be all that's needed for a local wave to peak and then decline.

Why can't we just give people masks and gloves again?

We can’t get masks or gloves because most countries banned exports of PPE. There’s not anywhere near enough domestic production to meet demand.

Also we had a stockpile in 2006 and it was emptied in 2009.

That even after three years, the Trump Administration remains so hobbled by the deep state and Obummer's legacy that he will need four (maybe six) more years to finally drain the swamp.

Trump walked into a house that was burning down and was handed a list of 10,000 things to fix, with "fix the national PPE stockpile" being item 9,500.

Resolving the Covid-19 disease is seeming to take longer than two weeks. About 20 days seems more likely at the moment.

Yes. Two weeks gets you resolution of about 88% of cases. Twenty days gets you to 98%. There are some cases that appear to continue to be infection much longer.

" If you doubt this, just consider how bad a job we Americans have done scaling up testing and making masks."

True. Consider that our globalized, free market system is not structured for something so basic in a public health crisis like $1 masks. This might be a good candidate for nationalization. Here's how a Dallas mask manufacturer explains it:

"The common theme is that during an outbreak like this, everybody wants to be his customer. But as soon as an outbreak subsides, his customers dump him and run back to China. The reason? His masks may cost a dime each, but a made-in-China mask might go for two cents.

“Last time he geared up and went three shifts a day working his tail off,” the mayor recalled. “As soon as the issue died, he didn’t have any sales. He had to pay unemployment for all these people, and he had to gear down.”"


" This might be a good candidate for nationalization."

Nationalization would be an awful idea. The solution to that problem is for the government to put in a large cost+ contract over a 6 month time frame. If in 3 months we've got a handle on the virus, the last 3 months of masks gets stockpiled. If not you tack on another 90 days to the end of the contract.

What's to prevent somebody using variolation to turn themselves into a living bioweapon? There's a few bad actors out there who feel they have nothing to lose and could use this as a way to "get back" at society. There's way too many unknown unknowns here and the law of unintended consequences absolutely dominates the discussion here.

Well, assuming it takes you're only going to be contagious for a few days it seems then your immune system beats down the virus. You'd be a bioweapon that couldn't be very well targeted.

I hear snot reaches its maximum bioweapon potency after 7 days, provided it's stored in a warm place.

At least that way, whatever ideas are exposed to public view will at least be able to be better grounded on data, and one assumes greater familiarity of how the pandemic spreads. NY is simply the among the first, not the last, parts of the U.S. that experiences the pandemic's first wave.

Congrats to Tyler for rejecting an explicitly stupid idea. What's next, being not in favor of self-improvement by hitting one's head against a wall?

The fact that it took this long is kind of disappointing. Look at all the enthusiastic comments here still defending the practice. Never argue with a libertarian if your time is precious.

Never speak with a libertarian if you consider your time worth anything. Unless you are looking for amusement, because they deliver that reliably.

Hopefully Tyler will also get around to discussing the numerous problems with the "stop time" idea.

He needs to banish the very thought of minimum wage laws, a scourge which makes a pandemic seem like a minor problem.

Yep. Tyler's example of the NBA is highly telling. If the idea was ever going to be worthwhile, it would've been for that very small set of very healthy workers whose labor is of sky-high value (yes even higher than most doctors' even during a pandemic).

Personally I'd kind of like to see the NBA experiment with it to see if it works. But they haven't given the idea a sniff AFAICT. This despite there being probably billions, certainly hundreds of millions of dollars that the NBA could make if the idea worked out.

It's unclear that the NBA has much near-term revenue upside from playing games at a remote location without fans.

Quoting a WSJ article from when the NBA suspended games ( https://www.wsj.com/articles/espn-tnt-face-lost-revenue-from-nba-season-suspension-over-coronavirus-11584032874 ):

"It’s unclear whether the networks will be required to pay rights fees for the full NBA season even though they aren’t airing games. Some media executives likened the season’s suspension to the 1998-1999 NBA season, which was shortened due to a lockout. That year, networks still paid rights fees.

Media contracts, including sports-rights deals, often include exit clauses for both sides for catastrophic events, but it isn’t certain that those would apply in this case."

So, the NBA might not directly generate any incremental revenue from playing games solely for television, because they'll get the rights fees regardless. It would have the intangible benefit of making their broadcast partners happy. It would also likely help the NBA's medium-term popularity, since it would be the only U.S. sport on television.

The real revenue boost for the teams, however, would be to play games in front of fans and therefore still have ticket revenue.

Other points worth noting:

- The real goal - for both the NBA and its national TV broadcast partners - is almost certainly to make sure that the playoffs happen. Those have notably higher TV ratings (i.e., TV ad revenue for networks) and also draw more interest. The playoffs had been scheduled to start April 18, with the NBA Finals in June. They can push that back if needed. The NBA was scheduled to have 17 games left in the regular season. They probably want to play at least 5 or so regular season games before the playoffs so that teams get a warm-up after so much time off.

- The NBA players, of course, are represented by a powerful union. The most popular players are celebrities with the ability to get out a message better than the NBA itself. If some of the top players don't want to play, then the NBA trying to steamroll them into playing would be a mess that would make the NBA look bad.

- The logistics of finding 15 remote buildings for the NBA to use and setting them up the necessary cameras and production facilities for high-quality broadcasts are do-able but would take a fair amount of effort. It would involve a lot of people beyond the players and very likely run afoul of the "essential businesses" rules of most states.

"If some of the top players don't want to play"

Which is yet another reason why Hanson's idea is a non-starter. Based on just the business side of things, the players will want very much to play -- remember that NBA salary caps are based on the NBA's revenue. A plummet in revenue means a plummet in salaries.

But I wouldn't be surprised if a lot of players found Hanson's offer of a stay in a variolation hotel less than attractive.

I wouldn't mind seeing the league and the players give Hanson's idea a try. They don't need 15 sites; they can double up and have four teams share an arena, playing double-headers if necessary or just playing on alternating nights. Or have six or eight teams share an arena and play triple-headers.

But we all, except for Hanson, know that's not going to happen. (The playoffs might happen, but not the hero hotel.)

"Based on just the business side of things, the players will want very much to play -- remember that NBA salary caps are based on the NBA's revenue. A plummet in revenue means a plummet in salaries."

See my precise prior points about the NBA's TV contracts:

(1) It is unclear that missing games this season will cause "a plummet in revenue" *except* for ticket sales revenue. And the ability to play games with fans in large arenas is largely out of the hands of NBA decision-making: governors and mayors aren't likely to allow it for some period of time. Also, delaying on the chance (small, IMO) that NBA teams are allowed to have at least some fans in attendance in July or August can be a rational move.

(2) It's an open question how much the agenda-setting, highest-paid players are motivated by the implications for their own salaries. Yes, Lebron James, Chris Paul, Steph Curry, etc. would prefer to have more money rather than less, but the possible loss of part of one's season salary is not a huge hardship relative to their overall career earnings under long-term contracts plus outside endorsement income. The players who are likely to care more financially are the players with low salaries by NBA standards (below the league average of $7.7 million), who have less confidence in long careers and less outside income. They don't drive the conversation from the players' side, however.

(3) MLB is an illustrative comparison. The season hasn't started, so the revenue hit will be larger as a percent of team revenue. The teams and the MLBPA agreed to a deal - https://www.espn.com/mlb/story/_/id/28964249/what-mlb-deal-players-means-2020-season-beyond - that limits players' total compensation to $170 million (about 4% of total annual payroll) if the season isn't played. It pro-rates players' salaries if a partial season is played - e.g. play 81 of the normal 162 games, and players will only receive 50% of salaries for 2020. They've also agreed that one of the criteria for opening the season is that teams are allowed to play in their home stadiums without bans on mass gatherings (though they may revisit that at a later date).

(4) I've thought that any plan to play NBA or MLB games for TV only wouldn't be structured as a "variolation" idea. It would be quite the opposite - frequent and widespread testing to try to prevent infection.

(5) To point 4 - one important consideration for professional athletes would be any long-term health effects. A slight reduction in lung capacity that would be an annoyance for a typical person, for example, could be a major performance problem for an NBA player who would then lack the stamina of other players. Professional athletes are several standard deviations on the right side of the curve in their ability to perform: the fall from "superstar" to "below average" at the NBA or MLB level is very modest relative to population-wide performance. Or, said a bit differently, suppose that the after-effects of COVID-19 for these elite performers are equivalent to 2 or 3 years of premature aging in their athletic performance: that's a huge earnings hit for an NBA or MLB player.

Most of these worries seem pretty surmountable. They don't seem to appreciate how malleable things can be during a crisis (as demonstrated by how much things have changed over the past month).

The obvious way to start a variolation program is with a study of perhaps 1,000 volunteers (give or take an order of magnitude). The logistics of doing something at that scale are obviously much more manageable.

That study would provide information about how variolation works, what the risks are, what dosage and delivery method works best, etc.

It would also provide marketing / momentum towards scaling. If a bunch of people develop immunity with low enough risks, then it'll be a big success putting us a step along the path that we took towards beating smallpox.

It would also allow a few weeks' time to work on the logistics of scaling up. Over the past month testing has scaled up from 1,000 to 1,000,000; a lot can be done in a few weeks. If we want to scale up to herd immunity that'll still be a big challenge, but if even hundreds of thousands of people per month can get immunity at a low enough cost then there are a lot of really useful things that can be done with that.

Of course, that points to the big open question about variolation, which is how well it will work at providing immunity and reducing the cost of infection.

I agree we can give it a try, would actually give us some useful data as to how real an issue this virus is for healthy young people.

You can give it a try, along with every single person advocating this idea. I have heard that Epstein's island is available for rental. Really, at this point, it is not a problem to get covid in the U.S. for anyone who wants to.

The selected strain used in variolation also matters. You want one that induces immunity but is weakly transmissible.

There are ways you could reduce risk - make the virus less effective in modulating the immune system. The original SARS seemed to suppress macrophage function (inflammatory ability and antigen presentation) - so remove/modify genes that do that. Add more proof-reading capability to the SARS-nCov-2 RNA polymerase to reduce mutation accumulation in the genome. I would also suggest add an antiviral susceptibility, again maybe a better binding sites for antivirals in the event the patient developed pneumonia etc.

If we followed this path we would end up with a type of variolation that is a bit more like a living attenuated vaccine.

You got it right with your last sentence. Since it would take the same 12-18 months to do this that it will take to make a proper vaccine, this is a silly discussion.

But this test could be done in a month. Low does is very predictable effect, and very low-tech.

And you think you (or any mol biologist) knows how to genetically modify the RNA polymerase to increase proof-reading capability? Come now.

Tyler, you are right to say that for variolation to work on any broad scale it needs cultural and institutional buy-in within that domain. And we have spent so much effort as a culture pushing the message that Coronavirus is deadly and dangerous for everyone, that more nuanced truths have been lost. That may be a necessary evil to get the masses to act for the public good. (Of course, some truths seem to be able to be reversed rather quickly. Just a week ago everyone was shaming the average person for wearing a mask. Now everyone is shaming the average person for not wearing a mask.)

My contention, however, has been that variolation may be better suited to sub-populations. Inoculating emergency workers in controlled waves, eg, could be used to prevent rapid spread within healthcare workers that would leave hospitals or cities without personnel they need during a covid outbreak. If you could get hospice or other elder care workers inoculated, it would greatly improve the safety to the most vulnerable citizens. Likewise, it may be prudent to inoculate key populations within the government or the military.

This doesn't require broad buy-in from the masses. It only requires limited buy-in from specific groups that are more likely to understand the real risks and benefits inoculation could bring. Many of these people, such as frontline healthcare workers, are already taking great risk just treating covid victims. Variolation with a low dose of the virus may seem like a reduction, not an increase, in risk for them.

While I do agree that as the lockdown drags on, or, more likely, the cycle of heavy mitigation -> relaxed mitigation -> outbreak -> heavy mitigation drags on, people will be more cavalier about the virus. I doubt the number of people who "just want it to be over with" will be significant, but, like with most cultural shifts, there will be a subconscious choice that results in a change in behavior. People will just take more risk and be less inclined to reduce their risk, even if the government tells them to. It's like with cars. As cars have become more safe to drive -- better handling, better safety features, etc --, we as drivers have taken greater risk. We don't explicitly say that because cars are safer we are going to drive more dangerously, but our behavior shows that subconsciously we have made that decision.

Right now it's all a gamble on the future. How long until an effective vaccine or treatment? How long can our economy survive on stimulus bills and liquidity injections? At what point does my wife get so tired of seeing me sitting around the house in my underwear that she runs to the nearest hospital and licks a covid patient? Quien sabe. But eventually we reach the peak of the patience curve and the question is whether it's before or after the peak of the patients curve.

Variolation makes more sense the longer this all goes on. And limited inoculation makes sense with some populations now.

There was an opinion piece in the WSJ a couple weeks ago proposing a more limited version of your idea - https://www.wsj.com/articles/expose-first-responders-to-the-coronavirus-11585067397

Summarizing: the author (Dr. Michael Segal) proposed offering the option of intentional exposure to younger first responders - without a known risk factor - who are otherwise set to enter self-quarantine due to possible exposure. Part of the rationale is that they might otherwise have to go through multiple quarantines if they in fact weren't exposed - and I think (not certain) that they need to test negative after a period of several days in order to go back to work.

He acknowledges that governments would hesitate to implement this plan due to liability concerns.

Such feeble objections to starting a pilot. Many thanks Professor Strauss.

#6 - That was also my first impression: you might be one of the few for whom the virus is extremely dangerous, or even deadly. A few babies and young kids have already died, as did e.g. Adam Schlesinger, the presumably healthy 52-year-old singer from Fountains of Wayne. Is that risk in in a thousand, one in a million? We don't know yet. And I wouldn't risk my and my family's health on such an altruistic gamble, at least not voluntarily.

Then don’t take part. It is voluntary after all.

"The proposal requires institutions to implement it, yet it doesn’t seem suited for any actual institution we have today."

Maybe, military might find it useful, if they feel they need to control trajectory/timing of infections to ensure a sufficient fighting force at all times? This became an issue over the last few days with the outbreak on the aircraft carrier, for example. They might try variolation prior to deployment of units.

I agree with Tyler, though, that doing it at general population scale with sufficient medical personnel to monitor the infected seems unworkable.

Another possibility might be, once we get the present surge under control, use variolation on medical personnel to ensure sufficient, immune personnel for the next surge. However, if we don't get enough medical personnel to volunteer, that would also tell us something about the risk vs. benefits.

Can someone please help me understand why the term 'variolation' is being used? Tyler, in this post, seems to use it to mean intentional infection, like parents having a chicken pox party for their kids when someone got infected in the neighborhood. That's not variolation, that's intentional exposure at a time when the virus is very likely to be relatively harmless.

Variolation refers specifically to exposing people to an a mild form of variola (small pox). There are two forms of variola that infect humans, variola minor (as the name suggests, much less dangerous) and variola major. Immunity from one confers immunity to the other. Variolation was the controlled exposure of variola minor to individuals who were then isolated until they were no longer infectious, thus protecting them from variola major.

There is functionally one form of CoV-SARS-2. There is ZERO evidence that a smaller initial dose of virus leads to a milder form of illness. The people proposing this are engaging in wishful thinking, and revealing their profound ignorance.

Let's call this what it is - intentional infection. Dressing it up with an archaic medical term that doesn't apply makes it seem like a much more reasonable idea than it is.

Some have made claims that viral load may explain the difference between some people who get mild versions and others who get seriously ill. Clear evidence for this is lacking but I think you go too far denying the possibility.

Clear evidence is lacking either way on this question, which itself strongly warns against a "variolation" attempt. Though it's not a nonsensical theory at least that fewer initial viruses equals a weaker infection.

The word "some" is doing a lot of work in your response, Dan. If you mean some economists who create data-free thought experiments, then yes. If you mean some infectious disease specialists and epidemiologists who are referring to CoV-SARS-2, then please post some links.

Yes, clear evidence is lacking. Another way to put this is that the idea is fanciful. Arguing how many viral particles will produce the desired infection is akin to arguing how many angles can dance on the head of a pin.

The link to the initial dose and subsequent severity of the disease is linked with the 1918-19 Spanish Flu pandemic. Simulation models showed that the infectious dose was related to the number of simultaneous contacts a susceptible person has with infectious ones; that severe cases of influenza result from higher infectious doses of the virus; and over-crowded places are the ideal environment for a susceptible person to be exposed to very high infectious doses of influenza.

High infectious dose and influenza disease progression have been shown in experimental animals;
The high infectious dose is associated with a higher viral load; and
The high infectious dose is associated with a smaller period of time to maximum viral load. [2]
[2] Influenza Infectious Dose May Explain the High Mortality of the Second and Third Wave of 1918–1919 Influenza Pandemic. PLoS ONE 5(7): e11655.

An analysis of 147 inpatients with influenza A (H3N2) infection (mean age, 72±16 years) reported major comorbidities and systemic corticosteroid were associated with slower viral clearance. Severe cases of influenza have more active and prolonged viral replication.,,,,,,,,,,,

"Healthcare Workers.

Reports across several countries report healthcare workers are more at risk of catching SARS-CoV-2 and potentially more severe. having COVID-19 disease. We use multiple sources as there is considerable variation in the number of deaths reported, which currently presents a confusing picture

What can we Conclude from the reports on healthcare worker deaths

If readers are confused by the mass of contradictory information, so are we.

What can be desumed by this post is that no one really knows what is going on, least of all governments and professional associations which seem at odds with news outlets as to how many of their members have died.

As our grandfathers used to say, when you do not know what is going on, do nothing. This is what we plan to do from our privileged position: observe and monitor the situation without jumping to conclusions."


" Overall, our data indicate that, similar to SARS in 2002–03,6 patients with severe COVID-19 tend to have a high viral load and a long virus-shedding period. This finding suggests that the viral load of SARS-CoV-2 might be a useful marker for assessing disease severity and prognosis."

Here is an article in the NYT about practical suggestions being made by economists: https://www.nytimes.com/2020/04/05/business/economy/coronavirus-economists.html These include a national order for hospitals to share ventilators. Duh. It's so obvious that its shocking that such an order has to be suggested by an economist. Meanwhile, a debate about variolation takes up way too much time and space.

To your point that this might induce permanent damange, Kaiser Health is reporting heart failures from COVID-19.

"As more data comes in from China and Italy, as well as Washington state and New York, more cardiac experts are coming to believe the COVID-19 virus can infect the heart muscle. An initial study found cardiac damage in as many as 1 in 5 patients, leading to heart failure and death even among those who show no signs of respiratory distress."


I am not a fan of variolation but the story you site is, to me, an argument for variolation. If you can give people a mild form of the disease you might save them from cardiac damage that a more severe form might cause. We just don't know.

"It’s unclear why some patients experience more cardiac effects than others. Bonow said that could be due to a genetic predisposition or it could be because they’re exposed to higher viral loads." From the article.

I agree with you Tyler.

I think the real question is why we don't do forced challenges for vaccine development. That could cut down the phase 3 part of the trial from a year down to several months. Tyler, what I really want you to comment on is what is your opinion on those. (See: https://www.nature.com/articles/d41586-020-00927-3 )

My view is that it’s not going to happen but if we have immunity passes some people will want to deliberately infect themselves and that is best handled with a formal system.

Some technical details on how this virus works:


it seems it lives in the throat first, and then it's kind of a crap shoot about whether it breaks to heart or lungs.

So no, voluntary infection is still nuts.

Viruses aren't exactly niceties sure, but open just any medical book and you'll find out that even the most benign illness sounds horrible.

In the absence of a vaccine, we do variolation in the interim, whether we want to or not.

The only chance the world ever had of containing this ended last November when the first COVID-19 carrier flew out of China and into the rest of the world.

This is a good point, several million people have undergone the experiment already but with poor data collection we cannot be sure of the impact.

You've missed a significant nuance to Robin's post. The proposal is meant to follow during the period after a peak. So:

#1: Is not valid - There should be sufficient medical personnel if you're in a test and trace period with continuing social distancing.

#2: I don't understand this point. The idea was to have the places in advance and to control the process. Why would you need test and trace? Testing yes, obviously. But tracing a stationary population doesn't seem hard. Also, while I'd suggest it being voluntary to enter, exiting would/should not be voluntary. You make a commitment on entering to comply with quarantine. If governments can enforce quarantine for non-voluntary infection, certainly we can enforce a voluntary commitment.

#3: There's not enough evidence yet for an organization with a public image to commit to this. While it might seem like the NBA has a strong motive to restart, they probably don't. They are entertainment. Clearly non-critical. They probably still get revenues for licenses, merchandise, etc, despite all this. They can run a season later, etc. Nothing there compares with the incentive to get some other non-remotable jobs back to work (factories, etc.). There's also no legal system yet to provide any advantage to variolation. If the country/state/city is on lockdown, so are you, prior exposure or not. There's some shift there.

#4,7,8: These all advocate weakly for raising the rate of infection by weakening restrictions. The problem is these are all more risky in terms of health aspects. Their advantages are economic. But they all risk another uncontrolled outbreak. Variolation doesn't create much social risk. Risk is more or less borne directly by those volunteering, and thus capped at the number of volunteers, and if they don't at least have odds in their favor of a net-benefit, they wouldn't volunteer. So the risk of variolation in comparison to weakening restrictions is much much smaller. To the degree that it is a stand in for some of the risk that weakining restrictions will incur, it's net benefit in that larger equation must be to lower the overall risk (this assumes, as Robin does, that it's not done as a forced-coercive-across the board policy, which would then have many more risks).

#5: This is the best, most coherent point. Probably a good reason for this to not start with much fan-fare. Certainly, don't overstate the case too early. Prove effectiveness, use in the most cost beneficial situations, focus on voluntary.

#6: Yes, true, it might happen as such. But it might not. Would the risk from variolation be less or more than risk of waiting. No one has such information.

In the end,

We wanted to flatten the curve in order to reduce the load on the healthcare system. One of the downsides is that this delays the point at which significant percentages of the population have been exposed and develop immunity. Variolation may accelerate the growth in immunity but it will increase the load on the healthcare system. Not sure I can hold both of these ideas. If I'm going to go all-in with variolation, I probably would have just skipped the whole effort to flatten the curve. We'd have an overloaded healthcare system and many casualties to the virus but a far more rapid path to population immunity and a salvageable economy.

The idea of variolation is controlled infections in a health care type setting, but more a hotel. If done in a controlled manner the health care system is overloaded. Of course, there is a problem, how to do it in a controlled setting if you are uncertain about the outcome. Possible that we may see a natural experiment in some third world countries.

We already have it.

In Italy, health care workers who have recovered from the virus work in the wards without masks since they are hopefully immune.

In the US there are many workers who are self-quarantining, either not going home or staying away from family when they do.

While no one is sending off a group dressed in blue uniforms to a 'heros hotel' with a marching band playing in the background, it is essentially happening on its own and assuming an antibody test can be made available at scale, we'll have a list of 'unintentional variolators' that could be assembled.

2. “Will any private institutions do it without complete governmental cover?” This is the part I am most pessimistic about, crisis-wide, in the U.S. To get cover, first you’d need an act of Congress. Good luck getting Congressional Democrats to vote for that. Even if you do, no one could be sure whether the federal law constitutionally preempts state law. So really you’d need each state to pass its own liability laws. Good luck with that. And many state supreme courts have struck down tort reforms as violating their state constitutions. Bottom line is any private institution would be putting itself at risk by participating.

TC & AT seem to have put the cart before the horse. RH is even worse, he blunders in calling a vaccine a "treatment". Such muddled thinking should be embarrassing. I don't go to econ blogs for their opinions about medicine. We lack sufficient data to show that low dose strongly correlates with mild cases of covid-19. It's as simple as that. (That is, assuming the Public Health & Medical Ethics people agree.) Without that correlation, this talk about variolation is hot air. One thing I don't understand, tho, is why TC seems to think liability isn't able to be managed by a properly structured company/organization. Do we need a specific law limiting liability for experiments in this disease? I think it is too early to be talking about induced infections, although one could argue that given the ponderous nature of our institutions, maybe talking about it now enable someone in July or June to make an evidence based proposal. "What did Tommy and Janie do this Summer?" "Oh, we sent them to Virus Camp!"

My main point is that it's too early to either favor or disfavor this set of mitigation strategies.

You see dose effect data from other viruses as completely irrelevant?

If #6 is real, it is most certain fibrosis. Fibrosis can be repaired/reversed with senolytics. What COVID-19 has made obvious as daylight is the need for the development and rollout of not only senolytic therapies (which are cheap and easy) but that of all methods of restoring biological homeostasis. The single greatest risk factor for COVID-19, and all other infectious disease, is decline of biological homeostasis. The restoration of biological homeostasis should be top priority.

Tyler's points in order:

1. Conserving medical personnel by making them immune prior to the peak is part of the point of this. And it's expected that there will be a second wave of this, just as there was for the 1918 Spanish flu, so we should try to get infrastructure in place before that. Hanson himself thinks this pandemic is a "practice run" for an even worse one in the future.

2. Where we put them is the Hero Hotel. We DO NOT release them back to their families like Lombardy, that would defeat the whole point. And since we know where they are, there's less effort required to track them. How can we be sure they're inside? Lock them in, post guards outside. If anyone inside changes their mind and insists they should be free to leave, just prevent them and let them sue, see if the legal system is willing to accomodate a person known to be infected. The reason it's not being done right now is because relatively few people even know about the existing practice of variolation, and for a large-scale organized effort we would want the authorities to be ok with deliberately infecting people.

3. You need a rather large number of players to restart the NBA season, and since they're unionized you'd need to get them to agree to this. Whereas it's much simpler to get some healthcare workers to volunteer for this and then once they're recovered they can go right back to work. There is no reason for the Hero Hotel to be tied to the place where people will go back to work, and instead the places designed to permit people to stay temporarily before being replaced by newer residents are... hotels.

4. Relaxing on activities for the young would not help, because they would still be free to infect older people. Deliberate infection MUST be combined with isolation to reduce spread. And, yes, having more immune people during the peak does help at the margin, as does people getting a smaller initial dose rather than a large one. The actual hotel space isn't literally free, but since fewer people are using hotels now it would be at a discount. And the legal issues are what Hanson wants people to work on now. Some regulatory changes have already taken effect in response to the pandemic, so it's not a complete pipe dream to think this could be permitted if people saw that it could help.

5. The President isn't going to tell society "We're going to expose the young", the people operating the Hero Hotels would say "We're going to expose isolated volunteers and study how the infection progresses under different treatments". It's not even necessary to start this in the U.S: Hanson notes that it's a big world out there and some place may be more open to this. Then once the idea has been tested more people can learn from the example and the Hero Hotels might spread.

6. Some chance that better treatments will arive does not obviate the possibility that we won't be ready before it spreads throughout the population.

7. As your co-blogger Alex Tabarrok points out, individuals deliberately infectings themselves may get a private benefit, but at a social cost if they infect more people. That's why it's important to tie infection together with isolation, as in Hanson's proposal.

Where is this medical staff that hasn't been already exposed to the virus for a second wave? Because of shoddy PPE supplies, we already have a lot of medical personal who have had some level of exposure. If variolation works, they are likely to have some of the benefits.

Maybe you could try this for the red states that are lagging behind social distancing and are likely to get hit with the wave soon, but from their point of view would you want to take a serious number of your medical staff offline for 2 weeks? What is morale going to be *after* two weeks of confinement (sorry 'heros hotel' is still a hotel), maybe many of them going thru it sick, AND then push them into hospitals to work non-stop ? If variolation doesn't work and you end up making many medical people very sick, then the state goes into their wave with even less support than they had before.

I don't think this can be pulled off in time to make a difference and if you had more time you would better use it to make more masks and ventilators.

The place for variolation is the military (as a few other commenters have alluded above.)

1) The military has a clear need for entire units - not just individuals - with immunity to COVID-19. Those units need not be large. Start with Special Forces and some engineering units, which may be sent to build temporary hospitals or refugee camps in epicenters.

2) The military has sealed bases where it can carry out group infection in a practicable manner.

3) The military has embedded medical personnel and enough unit cohesion that it's reasonable to expect the asymptomatic to care for the sick in Camp Variolation, leaving only the extreme cases for medical staff.

Tyler's post doesn't give enough credence to the idea that isolation could be achieved, either voluntarily or (as in the military) coercively. I, for one, would be very willing to enter Camp Variolation, if only to decrease the probability that my entire family gets sick at the same time.

The military already started on this experiment with the SARS-CoV-2 infection on the T.R. Roosevelt carrier. Should this have been allowed to continue? Should those on ship be rotated off once they are asymptomatic? Should new crew be added, increasing the variolation pool (I assume full decontamination of the ship will take some time)?

Live infection isn't variolation. An aircraft carrier is a much worse place to be sick than a military base. And an aircraft carrier crew is not a unit trained for on-the-ground interventions (sort of the opposite). So no, this is not at all ideal.

Still, it's a lot better for the military to gradually experience infections, one unit, base, or ship at a time than all at once.

The disaster of Britain's early herd immunity response, orchestrated by people like Dominic Cummings, should be evidence enough that there is a certain amoral idiosyncratic for its own sake attitude that is very much NOT what we need right now in this epidemic.

"A very small dose of the virus" What does that even mean? Even I can see he has no idea about epidemiology or how viruses work.

There are a number of technical issues here. The time from exposure to infections is variable. The time from infection to asymptomatic is variable. Biggest of all, the time that one who does not present clinical symptoms but still sheds active virus is also variable. The Wuhan scientists have seen viral shedding 20 days post recovery. this ends up as a much longer 'hotel' stay by some percent of individuals (and there is no way to identify those who will be viral carriers prior to them entering the hotel.

This is just another example of otherwise smart economists coming up with ideas that have no practical implementation.

My point by point reply: http://www.overcomingbias.com/2020/04/reply-to-cowen-on-variolation.html

Comments for this post are closed