Tag Archives: Proposal

Deliberate Exposure Intuition

Many have expressed skepticism re my last post on controlled exposure. So let me see if I can’t communicate my intuition more clearly, so we can all examine it more carefully.

Assume we have a virus like COVID-19, highly infectious and substantially deadly, not blocked or cured by any known or soon-coming treatment. It takes up to 2+ weeks from exposure to death or recovery, and advanced medical resources like ICUs can cut death rates. Even with unusually strong quarantine efforts, COVID-19 currently seems to be escaping from its initial region and nation, doubling roughly every week. Even if that growth rate falls by a factor of three on average, it will reach most of the world within a year.

At which point roughly half of the world who isn’t immune gets infected over a perhaps two week period. Medical resources are completely overwhelmed, so ICUs save only a few. And the world economy takes a huge hit; for perhaps months before that point most workers have stayed home from work in an eventually futile effort to avoid exposure. At the worst possible moment, food, trash, cleaning, heating, and cooling may be scarce, increasing the fraction of sick who die.

To deal with this crisis, there are two key kinds of resources: medicine and isolation. With limited medical resources, including medical workers, we can treat the sick, and cut their chance of dying. We also have a limited set of quarantine resources, i.e., places where we can try to isolate people, places that vary in their health support and in their rate of infection leakage in and out. If we put the more likely infected into stronger isolation, that slows the disease spread.

Consider three different policy scenarios, based on three different policy priorities.

First, consider a policy that prioritizes immediate-treatment. This is a common priority in our medical systems today. Each day, medical and quarantine resources are devoted to the individuals for whom they seem most most-effective in keeping that person alive over the next few days. So hospital ICUs hold the patients whom ICUs can most help now. And the best quarantine locations are allocated to the apparently not-infected at most risk of dying if infected. (Such as the old.) Workers are allowed to stay home from work if they think that will increase personal safety.

In this scenario, medical and critical infrastructure workers may not be given priority in quarantines or medial treatment. So medical workers are culled earlier than others due to their extra contact with the sick, and most medical workers may be sick or stay home near the peak of the epidemic, which is a pretty sharp peak. Most workers in critical infrastructure may be home then too, and may have been there a while. Worse, by allocating isolation resources according to a risk of dying if infected, a treatment-focused policy does little to slow the disease spread.

Next, consider a policy that more prioritizes containment. This is the usual priority of public health today facing a new contagious disease. Here more people become more isolated, and the best isolation resources are allocated much more to those most likely to be recently infected, not to those most likely to die if infected. Efforts may be made here to isolate medical workers, even if that results in worse individual treatment.

This priority can make sense given a substantial chance that the disease can be stopped from spreading beyond an initial area. Even if spread seems inevitable eventually, a containment priority also makes sense if that policy makes an effective treatment substantially more likely to be found before this disease spreads to most everyone. Or if more medical or isolation resources can be created in the extra time. Hope springs eternal, and it feels good to assume the best and act on hope.

But what if there is little hope of containing or treating the disease before most everyone is exposed? And what if getting sick and then recovering often gives someone substantial immunity to the disease for a period? After all, if everyone is constantly exposed, the recovered quickly get sick again, and this infection has high mortality, then death is coming soon no matter what. So we must hope for some immunity.

For this situation, consider a policy that prioritizes long-term treatment-resources. Most everyone will be exposed within a year or so, and unless they are immune they will get sick, at which point their chance of recovery instead of death should depend on medical resources, and critical infrastructure, at that time. So this policy seeks to create a pattern of isolation, and possibly deliberate exposure, to increase the average resources available to help people recover when they are sick.

The obvious problem here is that the above scenarios can have a pretty sharp peak in infection rates, overwhelming medical resources at that point in time. And workers who stay home also threaten the availability of other critical infrastructure resources. Yes, if containment slows the rate of growth of the disease, it also spreads out the time period of peak infection by a similar factor. But that could still be pretty short period.

Relative to the containment policy, this long term resource policy would seek to move the time of infection of many people from near the peak, to substantially earlier than the peak. Moving to later than the peak is not possible, if we’ve been containing as much as possible. And the obvious way to infect people earlier is to directly expose them, on purpose.

Of course directly exposing people won’t help spread out the peak if the people exposed are isolated to the same average degree as people are in the containment scenario. That would instead just move the peak to an earlier point in time, and perhaps even make it sharper, by making the disease spread faster. So this long-term treatment policy would have to involve infecting some people deliberately, while giving them much higher than average quality isolation. If their isolation were very good, then they’d use medical resources at an earlier point in time when such resources are more available, without adding much to the overall growth of the disease.

Now, if good isolation resources, and medical resources, were already strained dealing with a flux of likely infected from outside, then there might be little point in adding new infected on purpose. But what if there are many good isolated places available not being fully used to deal with folks very likely to have been exposed, and also available medical resources not fully used, what if recovered folks had little risk of infecting others for a period, and what if we were closer in time to the peak than that average period between reinfections? Well that’s when we might tempted to deliberately expose some, and then to strongly isolate them.

One key idea here is to create a stronger correlation between the strength of isolation of a place and the likelihood that people there are infected. Such a strong correlation allows us to create a population of already recovered folks who are at least temporarily immune. And that can spread out the period of peak infection, so that more medical resources are available to treat the sick. And that can cut the average mortality rate, which means that more people don’t die.

People who work in medicine and critical infrastructure seem especially promising candidates for early deliberate exposure. This is because after recovery they become more available to work during the peak infection period. They are not sick then, and are less afraid of being exposed then, making it easier to persuade them to go to work. The other set of promising candidates are those most likely to die without sufficient help, which seems to be men and especially the old.

And that’s the intuition behind deliberate exposure. Its wisdom depends on some parameters of which we are unsure, and may learn more about soon. So it seems clearer that we should think more about such options than that we should pull the trigger to start one now. And there are substantial challenges in organizing such a policy fast enough, and in gaining sufficient public and elite support to allow it. Maybe this can’t work this time, and must wait until another big pandemic.

But contrary to many loud and rude commenters lately, this option isn’t crazy. And within the next year we may come to see and suffer the full consequences of not working harder to spread out a pandemic peak of maximum infection and medical need.

Added 2p: The obvious easy win policy solution (given that key assumptions hold) here is just to make it easy for people to volunteer for (1) exposure to virus, (2) strong 24 day quarantine, (3) medical help while there, (4) regular checkups afterward. Create a place where people can go to do this, an easy way to sign up legally, and pay to expose and house them there. Maybe even pay them extra if they work in medicine or critical infrastructure. (Btw, as such an option isn’t now available, and I don’t work in critical infrastructure, it wouldn’t help society much for me to just “go infect yourself”, as many have suggested in so many colorful ways. And as I don’t own my family, I can’t volunteer them.)

GD Star Rating
a WordPress rating system
Tagged as: ,

Consider Controlled Infection

In many places long ago, in families with many kids, as soon as one kid caught an illness, parents would put the other kids in close contact, so they could all catch it at once. Because it was less trouble to care for all the kids in a family at once than to care for them one at a time.

Should we also consider controlled infection to deal with our current pandemic? Like controlled burns that prevent later larger fires, it might be a good idea to expose some people early on purpose.

Today a coronavirus is spreading rapidly across China, and the world, and many are trying hard to resist that spread. One obvious reason to resist is the hope that the spread can be completely stopped, limiting how many are exposed. However, once a contagious enough virus has spread to enough people and places, this scenario becomes quite unlikely; the virus will soon spread to most everywhere that isn’t high isolated.

Unfortunately, we are probably already past this point of no return with coronavirus. It seems to spread easily, apparently including via people who are contagious but don’t show symptoms. It already seems to have spread from its initial region to infect many people in a great many other Chinese cities and regions (thousands infected, dozens dead). And that’s with keeping everyone home from work, which can’t last much longer. Once this virus comes to infect most of China, it seems hard to imagine a strong enough China wall (a 24-day quarantine for everyone leaving) to keep it from spreading further. Especially since China & WHO are arguing against such a wall, and we already have confirmed a few hundred cases outside China; they’ve doubled every week for four weeks.

Another reason to resist virus spread is in the hope that a vaccine (or other effective treatment) will be available before it spreads everywhere, stopping the spread at that point. There’s some hope for a drug soon to prevent infections, but the odds are poor and if that doesn’t work prospects are dim. Alas “typically, making a new vaccine takes a decade or longer”, and estimates for this case are at least 18 months. That doesn’t include time to manufacture and distribute it, once we know how to make it.

As of yesterday, total known deaths were 1384, a number that’s had a 6 day doubling time lately. (A very different method estimates 7 day doubling.) At that rate, in four months deaths go up by a factor of a million, which is basically the whole planet. So unless long-term growth rates slow by more than a factor of four, there’s probably not time for a vaccine to save us.

If the virus spreads to most of the world, so most everyone is exposed, then the fraction of the world that dies depends how deadly is the virus, which we just don’t know, and can’t control. Maybe we’ll get lucky, and this one isn’t much worse than influenza. But we are probably not so lucky. The fraction of the world that dies also depends our systems of social support, which we can do more to influence.

I’m not a medical professional, so I can’t speak much to medical issues. But I am an economist, so I can speak to social support issues. I see two big potential problems. One is that our medical systems have limited capacities, especially for intensive care. So if everyone gets sick in the same week or two, not only won’t the vast majority get much of help from hospitals, they may not even be able to get much help from each other, such as via cleaning and feeding. Perhaps greatly increasing death rates. This problem might be cut if we spread out the infection out over time, so that different people were sick at different times.

The other related problem is where many non-sick people stay away from work to avoid getting sick. If enough people do this, especially at critical infrastructure jobs, then the whole economy may collapse. And not only is a collapsed economy bad for most everyone, sick people do much worse there. Not only can’t they get to a doctor or hospital, they might not even be able to get food or heating/cooling. Infected surfaces don’t get cleaned, and maybe even dead bodies don’t get removed. Thieves don’t get stopped. And so on. We can already see social support partially collapsing in Wuhan now, and it’s not pretty.

There’s an obvious, if disturbing, solution here: controlled exposure. We could not only insist that critical workers go to work, but we might also choose on purpose who gets exposed when. We can’t slow down infection very much, but we can speed it up a lot, via deliberately exposing particular people at particular times, according to a plan.

Such a plan shouldn’t just expose random people early, as they’d be likely to infect others around them. Instead, groups might be taken together to isolated places to be exposed, or maybe whole city blocks could be isolated and then exposed at once. Exposed groups should be kept strongly isolated from others until they are not longer very infectious.

Those who work in critical infrastructure, especially medicine, are ideal candidates to go early. Such a plan should only expose a small fraction of each critical workforce at any one time, so that most of them remain available to keep the lights on. If critical workers could be moved around fast enough, perhaps different cities could be exposed at different times, with critical workers moving to each new city to be ready to keep services working there.

Such plans can help even if some people who are infected and recover can get reinfected later. As long as being infected gives enough people enough immunity for a long enough time period, that is enough for this plan to spread out the infections over a time period of similar duration, so medical service needs don’t all appear together. Even an immunity of only two months, which is extremely short compared to most diseases, would allow a lot of spreading.

People selected to be exposed earlier might be paid extra cash, to compensate for perceived extra risk. (Maybe X days worth of their usual wages, so as not to especially select the poor.) Or perhaps they could be paid in extra priority for sick associates if medical help is rationed later. (I’d seriously consider both kinds of offers.) We might even be able to implement a whole plan like this entirely via volunteers, though adding that constraint may make a strong plan harder to design. A compromise might be to let city blocks vote on if to be paid to go early together. I’m willing to help in design work on this, if that could help make the difference.

I don’t have a detailed plans to offer, and obviously any such plans should be considered very carefully. Also obviously, such plans might face strong opposition, which could undermine them. If they were designed or implemented badly, they might even make things worse. But the alternative is to risk having large fractions of the population get sick at once, while the economy collapses due to critical workers staying home to avoid getting sick. A scenario which could end up a lot worse.

So authorities, and the rest of us, should at least consider controlled infection as a future option. I’m not saying we should start such a plan now; maybe that drug will work, and it will all be over soon. But if not, we should start to ask when we might learn what could help us decide, what might be a good time to pull the trigger on such a plan, and how to prepare earlier for the possibility of wanting to pull such a trigger later.

Added 17Feb: See also my next post elaborating the intuition behind why and when deliberate exposure could make sense.

GD Star Rating
a WordPress rating system
Tagged as: ,

Nickname Court

For perhaps a millions years, humans did justice mainly via gossip: the rule was that when you saw a rule violated, you were to tell of it to others, talk together on what to do, and then do it.

Problem was, gossip has long been known to be unreliable. When you hear someone rely an accusation that another person did something terrible, there’s a social “rush to judgment” pressure to immediately agree, to show your disapproval of that terrible thing. Even though you haven’t heard all the relevant evidence, including what that accused person might say in their defense.

Forager bands fixed this vis group discussions. They’d all gather around a campfire, listen listen together to the accusations and any rebuttals, discuss it, and only decide together after everyone had their say. Alas, as social groups got larger this was no longer feasible. So near the dawn of civilization, we invented formal trials: we supported the judgment of a subset of us, who made sure to judge only after hearing all the evidence.

Today we still gossip on a great many topics where there are no formal trials, and on those topics we still suffer big problems with rushes to judgment. So we’d do well to try to find more excuses to make formal trial processes whose judgements we’d be tempted to support.

Note that this isn’t the same as just having some group recommend a judgement. We do that all the time via news editorial boards, expert commissions, interest groups endorsements, etc. The difference is that a formal trial is a public process, airing all relevant evidence, and its judges are not selected to already represent particular sides in familiar debates.

But how? We might declare a new kind of court housing a new kind of trial, but how can we get people to bother to participate as jurors, present evidence as advocates, and to see its ruling as “official” enough to feel pressured to support it?

Here’s my idea: nickname courts. Imagine that people are tempted to use a nickname for a certain person in a particular social context. Such as: the 2nd grade homeroom 3 of public school 117 wants to use the nickname “Stinky Stu” for young member Steward Williams. Perhaps on the basis of recent events wherein young Stu was said to be stinky.

Assume there’s no law against using such a nickname, but that students, parents, and teachers might feel there are relevant social norms against doing something so “rude”. Some of these authorities might even declare their own local rules regulating such nicknames. (Nicknames are going to exist no matter what rules; the best we can do is to better channel that instinct.) I propose that these authorities give substantial deference to a new kind of court, a nickname court.

Someone initiates the process by officially proposing a nickname for a person in a social context, and committing to argue in its favor. That person who is to be given this nickname is notified, and confirms that they will oppose this. Then a “random” jury drawn from this social context is impaneled, which in the example above would be members of this homeroom class. They come together, all present their arguments and evidence, and the jury decides. In the example above, these might all be 2nd grade members of this class, including Stu.

If the jury rules yes, then it becomes more acceptable for members of this community to use this nickname of Stinky Stu for this particular Stu. This ruling may or may not have any legal force, even within the official rules of the school. But people might feel more comfortable knowing that their actions had this “official” support.

If Stu later decided that things had changed enough, he might initiate a new case, arguing that the nickname should no longer apply. Others would be notified, and we’d see if anyone wanted to take the lead to oppose him, starting with whomever initiated that first case. If Stu wins, then it would no longer to be as acceptable to call him Stinky Stu, though it might be okay to call him, “Once-was Stinky Stu”.

This sort of approach could obviously be applied to social media. For example, someone might initiate a case to make it more okay to call Robert Random by the nickname “Racist Rob” on Twitter. As I’ve said before, many of us might prefer a formal process for such labels over the current internet mobs that greatly suffer from rushes to judgment.

I’d bet that kids could really get into nickname courts, and learn a lot about law in the process. And later in life they’d probably think more about how similar processes might apply to other topics. For example, I’d love to have Radical Reform Courts, which evaluate radical proposals for social reform. Today there’s way too much of a rush to judgement, wherein each person who hears of a proposal quickly imagines one potential problem, and then concludes that it can’t work. But I don’t yet know how to make anyone care enough about the ruling of such a court.

GD Star Rating
a WordPress rating system
Tagged as: ,

Open Borders

In their new book Open Borders: The Science and Ethics of Immigration, Bryan Caplan and Zach Weinersmith do everything you’d think that good policy pundits should do.

They don’t just track trends or scold rivals, they identify and focus on a feasible positive policy change. They don’t just pick any old change, but focus on one of the biggest possible gains they can identify. And it isn’t a complex fragile proposal that most people couldn’t understand, or that would go badly if not implemented exactly as recommended; their proposal is simple and robust. They don’t pick a topic that has little emotional-resonance, regarding which few would act even if they were persuaded; their topic is quite emotionally-engaging. They don’t pick a change so abstract (like futarchy) that few can concretely imagine it; one can create concrete vivid images of what would happen if their proposal were implemented.

They don’t use complex technical prose, they write in simple clear language, and even add engaging pictures; their book is actually a well-done “graphic novel”. They don’t just present one side of an argument, but instead respond to many major counter arguments. They don’t just use one favored framework of analysis, they consider the issue from many possible frameworks. They don’t just focus on their favorite policy choice, they consider many possible ways to compromise with others. They aren’t overly confident in their claims. And while they consider many possible details and complexities, their main argument, regarding the main effect of their proposal, is simplicity itself.

Most important, their arguments seem solid and correct. Adopting their proposal could in fact plausibly double world product, over and above the growth rate that we might achieve without it. The main obvious effect seems so huge as to overwhelm other considerations. Relative to that huge gain, other costs and risks seem minor and acceptable. Of course, the real world is more complex than are our models of it, and so we can never be very confident that changes which go well in our models will actually go well in the real world. And all the more so when our models are noisy and partial, as in social science. Even so, this is another case I’d call “checkmate”, at least in argument terms.

So, damn it, Caplan and Weinersmith do everything you might think pundits should do. I remain personally persuaded (as I have long been); I’d pull the trigger on doing large broad tests of their plan, and if necessary making big compromises to get a deal that can make these tests happen.

I very much hope that everyone loves this book, and that it is the trigger we needed to start a larger debate that leads eventually to big trials. But alas, I’d bet against this happening, if I had to bet. The large political world isn’t that responsive, at least in the short to medium term, to the world of elite policy debates, and in the elite world people mainly care about signs of status and prestige. Elites loved Hawking’s Brief History of Time, Dubner & Levitt’s Freakonomics, Piketty’s Capital in the Twenty-First Century, and Bostrom’s Superintelligence not because those offered clear solid arguments that readers understood, but because they came with signs of high status. Many elites talked about them, their style projected prestige, their authors had high status affiliations, and the positions they took were in fashion, at least in elite circles.

I deeply admire my colleague Bryan Caplan, and am proud that he has again gone for the big solid simple intellectual win, as he did before regarding politics, parenting, and school. I hope he can do it another dozen times. I’ll read each one, and usually be persuaded. There’s a small chance he’ll have big effects, and his taking that chance seems a clear win on cost-benefit terms. But I must also be honest; that chance is still low.

GD Star Rating
a WordPress rating system
Tagged as: , ,

Who Vouches For You?

A <600 word summary of my crime law proposal:

Who Vouches For You?: A Radical Crime Law Proposal

The legal system used by most ancient societies was simply A-sues-B-for-cash. But in the last few centuries, states added “crime law,” wherein the state investigates, sues, and imprisons “criminals.” These centrally-run one-size-fits-all bureaucratic systems don’t innovate well or adapt well to individual conditions. And even though most of your “constitutional rights” are regarding such systems, they still seem badly broken.

In the ancient world, a stranger who came to town was trusted more if a local “vouched” for them. We still use vouching today in bonded contractors, in open source software, in organized crime, and in requiring most everyone to get an insurer ready to pay if they cause a car accident. I propose requiring everyone to get an insurer to vouch for them regarding any crimes they might commit. If you are found guilty of a crime, your “voucher” pays the state a fine, and then pays to punish you according to your contract with them. This fine in part pays the private bounty-hunter who convinced the court of your guilt. Competing bounty-hunters obey law because they can’t maintain a blue-wall-of-silence.

To lower your voucher premiums, you might agree to (1) prison, torture, or exile, if caught, (2) prior limits on your freedom like curfews, ankle bracelets, and their reading your emails, and (3) co-liability wherein you and your buddies are all punished if any one of you is found guilty. In this system, the state still decides what behaviors are crimes and if any one accusation is true, and it sets fine and bounty levels regarding how hard to discourage and detect each kind of crime. But each person chooses their own “constitutional rights”, and vouchers acquire incentives and opportunity to innovate and adapt, by searching in a large space of ways to discourage crime.

Some key details:

  1. Judges and juries can retain discretion to consider case details when setting guilt or fines.
  2. If fines vary with wealth or income, then the rich don’t get a free pass to commit crimes.
  3. We could subsidize premiums, or offer a public option, to poor ex-cons for which we feel sorry.
  4. Other poor ex-cons might have to work for a while at isolated ships or mining or logging camps.
  5. Not being vouched is punished severely, not via a fine. Need violations to be rare, as with license plates today.
  6. Perhaps each person/building wears visible QR code or pingable RFID of voucher-client ID (VCID).
  7. It is enough to know VCID to charge with crime, no need to physically detain them.
  8. Key criteria for being a voucher is showing that will have enough money to pay fines.
  9. Vouchers are held to contracts for as long as it takes clients to find new vouchers.
  10. Contracts typically worse for clients over time if not renewed, to cover revealed-criminal scenarios.
  11. Contracts typically do not cover pre-existing crimes or plans, for which prior voucher pays.
  12. Clients can switch at will, though co-liability partners must all agree to switch at same time.
  13. First-to-file bounty hunter has right to prosecute first, though is crime to file with little evidence.
  14. Courts would remain skeptical of both sides’ evidence, with evidence faking being a big crime.
  15. Bounty-hunters access to evidence varies with contract-specified client privacy levels.
  16. Market estimates of fines given privacy levels set extra fine factors paid by clients with high levels.
  17. As immigrants & tourists must be vouched, it matters less if immigrants cause more crime.
  18. Parents must get vouchers for kids, so “majority” age could be when kids can afford it on own.
  19. These or related vouchers might pay for costs to assist folks suffering poverty or unemployment.
  20. May want solemn voucher signing ceremony, after passing test shows that understand contract.

Added: See also this talk video.

GD Star Rating
a WordPress rating system
Tagged as: ,