When analyzing the dangers of COVID or proposing policies to combat it, there is a utilitarian cost-benefit framing which is implicitly assumed. Perhaps the best example of this is the Micro Covid calculator. Every action has a cost and benefit associated with it and the goal of any policy is to choose the best set of actions.

This framework, which I will call “risk based”, is easy to express numerically. This makes it feel objective and rational. However, this framework does not align with how people think about the dangers of COVID and is not how the policy decisions actually get made. By understanding the more intuitive framework which I will call “safety based” it is possible to understand the underlying motivations behind the policies and arguments being made. Understanding the constraints imposed by the safety based framework makes many of the policy failures throughout the pandemic understandable and in fact inevitable.

Safety Based Framing:

When thinking about an activity which involves risk, people do not, even implicitly, do a cost benefit analysis to determine if it is worth it. Instead they ask if it is safe or not. If an action is safe then it can be done without worrying about any residual risks. If an action is unsafe then it can only be done if extraordinary precautions are taken. If you fail to take these precautions then you are considered morally blameworthy for any bad outcomes which arise.

Determining if an action is safe or unsafe cannot be objectively determined. Although it is influenced by the relative risks involved it is inherently socially constructed. To take a simple example consider the difference between driving without a seatbelt and riding a motorcycle.

Riding a motorcycle is a much higher risk activity than driving without a seatbelt. However we treat them very differently. If you are taking the proper precautions (wearing a helmet, proper gear) then riding a motorcycle is safe but risky. On the other hand driving without a seatbelt is inherently unsafe[1].

In March 2020, as COVID cases surged in Italy and New York, there was a shift and COVID became unsafe, which meant that measures had to be taken to eliminate the unsafety. However, three broad camps emerged as to how COVID was unsafe. The seemingly small differences between these perspectives have lead to more wide fissures in rhetoric and policy as the pandemic has continued.

These perspectives could be briefly described as:

  • “Zero COVID” - It is inherently unsafe to be infected with COVID
  • “Focused Protection” - For certain high risk groups it is unsafe to be infected, everybody else is safe.
  • “Flatten the Curve” - COVID is unsafe if it exceeds our capacity to handle it.

None of these perspectives are factually incorrect, they all have some basis in the objective data. Nevertheless, based on these three different lenses the same data is focused and emphasized in different ways to mesh with the overall world view.


The most direct way to translate from the risk of COVID to the unsafety of COVID is to treat COVID as inherently unsafe. This means that an individual is responsible for taking any reasonable precaution to avoid being infected. If you failed to take a precaution and ended up getting infected, then you are morally responsible for any negative effects that this caused.

Proponents of this view will often point to evidence which indicates that COVID unsafe even for those who are hypothetically, low risk. Long COVID means that even those who seem unaffected may be suffering long term serious damage, and the high absolute number of young and healthy people who have died are indicators that getting sick is not an acceptable option.

From this perspective the only way to get rid of the unsafety is to completely eliminate the virus. As a consequence they tend to believe that a very restrictive lockdown would be highly effective as well as logistically feasible (“If everybody stayed home for two weeks this would all be over”).

Focused Protection:

Another camp, took the view that COVID was unsafe for a few particular groups of the population but was safe for everybody else (just the flu). Those who are unsafe are responsible for taking any needed precautions while those who are safe should be free to do what they want.

To support this view proponents will point to data showing an extremely low percentage of deaths for the young and healthy and argue that this means that any residual risks are negligible and can safely be ignored.

From this perspective the pandemic ends when enough people are safe. Proponents of this view will tend to have expansive views of what makes a person safe. Anybody who is young and healthy, vaccinated, or previously infected will be considered safe and thus free from worry.

You can see the difference in perspective in the response to an infection. For those who treat COVID as inherently unsafe there is an emphasis on how scary the infection was, if the symptoms are mild this is credited to whatever protective measures were taken (vaccination, boosters, mask wearing). For those who treat COVID as unsafe for a select few, there is an emphasis on the relative mildness of the symptoms, and provide reasons for why they are inherently safe (being young, healthy, previously infected or vaccinated).

Vaxed and Done:

The arrival of universal vaccine availability added a large group of people to this camp. For those in the focused protection camp, vaccines were one of many ways to make the relatively few unsafe people safe. However the “Vaxed and Done” considered almost everybody unvaccinated unsafe (except for those too young), the availability of the vaccine then meant that everybody was either safe or unsafe by choice.

Vaccine Hesitancy:

On the other hand, most of the vaccine hesitancy comes from this group. There is a core group of vaccine refusers who refuse because they believe that the vaccine is ineffective or dangerous. The majority however are simply those who feel it is unnecessary because they are already safe. Understanding this point resolves one of the apparent paradoxes surrounding vaccine uptake. Republicans have much lower vaccination rates than democrats generally but among senior citizens vaccination rates are extremely high. This is largely because they, unlike younger Republicans, considered themselves unsafe from COVID.

The messaging around vaccination has largely been dominated by the framing that COVID is universally unsafe or by the societal goal of achieving herd immunity. For those who reject these arguments and instead believe that vaccines are important for the sick or elderly it is more difficult to come up with messaging to convince them.

One possible message is to not treat vaccination as required for safety, or as a ticket out of the pandemic. Instead, vaccination is a reasonable risk reduction technique that even those who are safe can take. A safe person doesn’t take the vaccine because they are at risk of serious injury or death. Instead they take it because they make an annoying cold into a more mild sniffle.

Flatten The Curve:

The public health responses to COVID have not been primarily driven by either of these perspectives. Instead the focus has been on the dangers of collapse if COVID gets out of control. For individuals getting COVID is risky but not unsafe, but if cases increase out of control then it will become unsafe for everybody. Based on infectious disease modeling it became accepted that if cases started increasing at all they would very quickly increase out of control without swift action. This meant that the danger of COVID could be simplified to “If cases start to rise, we are unsafe”.

Framing the dangers of COVID in this way has some clear political and logistical advantages.

If every person is responsible for their own safety, then it becomes impossible to come to a stable situation, people who want to be safe will be stuck in their homes for months on end, while those who find that unacceptable will be unsafe and not even take reasonable precautions. By treating it only as a societal safety issue, rules can be developed which provide leeway for needed social interactions. Some people will inevitably be infected if restaurants, airports, or stadiums are open, but as long as cases stay under control that is an acceptable outcome.

Universal Measures:

From this perspective only measures which can apply to a large fraction of the population are of interest. Individual precautions will make little difference unless they can be widely followed.

Furthermore many precautions are not possible for many people to take (e.g. expensive air filtration systems). It is unfair to say that only a few people can be safe and therefore these precautions cannot be necessary to achieve safety. It would also be unfair to give privileges to those lucky few who can take these precautions, so therefore these precautions cannot be sufficient to achieve safety.

You can see this dynamic very strongly with the messaging and policies around vaccines.

In the initial phase, the vaccine was only available to healthcare workers and the elderly. At this point the vaccine could not be necessary for safety, if you followed the proper guidelines and waited your turn, you were still safe. Since only a few had access to it, the vaccine could not be sufficient for safety, so no restrictions could change for the vaccinated.

In the second phase the vaccine was universally available and it appeared that with enough people vaccinated, herd immunity would be reached. At this point vaccines were sufficient for safety, and the process began of removing restrictions on the vaccinated. Since the vaccines were universally available they could be considered necessary for safety and being unvaccinated could be considered like being without a mask at an earlier part of the pandemic.

In the third phase it became clear that vaccination alone was not sufficient to reach herd immunity and prevent cases from rising back up. At this point vaccination remained a necessary part of safety but was no longer sufficient.

There is an example of a “vaccination” which never became universally available and so never could be a necessary or sufficient part of safety. Prior infection appears to provide roughly the same protection as vaccination[2]. Since it was obviously undesirable to make it universally available, it was impossible to consider it sufficient for the purpose of any restrictions[3].

Chicken and Egg Problem:

Since only universal measures can be necessary or sufficient for the purpose of safety. There is a challenge with developing, scaling up or even investigating more effective measures.

Consider the case of N95 masks, which are widely believed to be more effective than cloth masks. At the start of the pandemic they were in short supply and so could not be part of staying safe. Theoretically the solution is obvious, use whatever supply you have and invest in scaling up capacity for the future. But in practice it is not easy to accomplish. Since N95 masks are not necessary for safety, having them won’t make people safer. Since N95 masks are not sufficient for safety, having them available won’t allow any restrictions to be lifted. Since nothing will change with them available, there is no reason to invest in expanding their supply.

You can see this dynamic play out in many different situations. Rapid tests, ventilation systems and contact tracing are all examples where they could not be part of a safety solution at the start and so they lacked investment and so now, even two years later, they are still not integrated into our safety protocols.

Vaccination is the only place where we were able to avoid this trap. Although at the start there was a similar problem (why get vaccinated if you are safe without it and it doesn’t let you do anything?), it was able to make the leap and become universally available and an integral part of safety. Vaccination had a few key advantages which allowed it to make this leap. At the start of the pandemic there was already a consensus that vaccines would need to be created to truly solve the issue. Not only was there agreement that it was necessary, but there was a pre-existing pipeline[4] in place for creating new pharmaceuticals[5] and making them a necessary part of safety.

Avoiding this trap for less well established solutions is a extremely difficult problem, and I don’t think there is an easy answer. One possibility is to leverage something with non-safety purposes and slowly expand it. This may have happened with PCR tests. These tests are useful for diagnostic and population screening purposes. Since they were in short supply, they could not be used for mass population screening. However, their diagnostic uses created a source of demand which allowed the supply to gradually increase. This is not a perfect solution, PCR tests are optimized for accuracy rather than speed, but has at least had some success.


In the past two months Omicron has sent cases skyrocketing, even in heavily vaccinated areas.

This rise has been so rapid that it is considered impossible to stop via the existing public health measures. If the existing definition of safety, that cases do not significantly increase, is retained then it would follow that everybody is doomed to be unsafe.

This is not an acceptable outcome, and so public health officials are abandoning the old definition of safety. However there is no consensus on what the new messaging should be.

For some, the only way to be safe is to be as cautious as possible. Even if cases will still inevitably rise, you should do your best to stay safe. This group is behind the renewed push for measures like N95 masks and frequent at home testing.

For others, however, the way forward is to focus on the safety provided by boosters and vaccines. It is infeasible to keep people safe by preventing cases from rising, but if we focus on the low risk to those vaccinated we can retain normalcy. This group is behind measures such as the reduced quarantine time, and messaging that Omicron is both inevitable and not a big deal if you are fully vaccinated. You can see this with the head of the CDC questioning whether children are sick _with COVID _or _of COVID _and talking about the amount of comorbidities in the vaccinated people hospitalized.

The outcome of this rift in public health guidance has yet to be seen. The current messaging is the result of two distinct groups fighting over their policy recommendations, and coming to some compromise. N95 masks are now considered better than cloth, but are not yet recommended. Asymptomatic quarantine is reduced to five days, but it is still important to get a test if you can.


Taking the distinction between safety and risk seriously has implications which reach beyond COVID. It is tempting to view safety as an irrational cognitive bias, and ignore it in policy making. That is a serious error, people need safety and will mold policy to achieve it.

Early visions of COVID policy imagined governments changing the rules depending on local circumstance, refining their policies to incorporate the latest research and investing in long-term capacity for cheaper, more effective solutions. None of this happened, which was largely inevitable when we understand the constraints imposed by safety.

The actual solutions adopted had the virtue of being immediately, universally applicable. Hand washing, cloth masks, and six feet of distancing may not have been the optimal precautions, but they could be adopted quickly and let everybody achieve safety.

More sophisticated measures may have eventually been more optimal for risk reduction but they were either too slow or too limited to be used. Their proponents, largely analyzing from the perspective of risk, were unable to adapt their solution for the purpose of safety.

Consider the measures taken for ventilation. Advocates have argued that installing expensive HEPA filters can make a building much lower risk for the transmission of COVID. But if we understand the safety framework being followed then it is apparent that this has no chance of happening and instead little will be done about ventilation. If the need and requirements of safety were considered, the proposal might instead be “Open a window, run a fan, or have some cheap filter”. This might or might not work against COVID, but it would at least be possible to try.

Hopefully, the COVID pandemic is nearing its end. But this general problem will remain salient. Proposals designed and analyzed from the perspective of risk need to be analyzed and adjusted based on how they will actually be implemented from the perspective of safety.

  1. I ended up driving ~100 miles to get my first Pfizer dose. At the time I had compared the risk of a fatal accident, which was considered negligible to the risk of a blood clot from J&J which was serious enough to halt its use. This ended up making the car ride significantly more nerve-wracking as now the risk was salient. ↩︎

  2. The scientific literature is ambiguous on whether it is slightly better or worse than the Pfizer/Moderna vaccines. Based on the precedent with the J&J vaccine I think it is indisputable that a vaccine with the protective level of prior infection would be considered equivalent. ↩︎

  3. In Europe, a recent prior infection was considered equivalent to being vaccinated, but before the universal rollout of vaccination, it was not considered at all. ↩︎

  4. Note that deviations from the existing pipeline (such as human challenge trials) were not seriously considered. ↩︎

  5. On the medical side, not only have new treatments (monoclonal antibodies, fluvoxamine, dexamethasone) been developed over the course of the pandemic, but the initial best practice of aggressive ventilation was quickly discovered to be counterproductive. ↩︎

New to LessWrong?

New Comment
14 comments, sorted by Click to highlight new comments since: Today at 3:50 AM

While it is interesting drawing a distinction between "risky" and "unsafe" on moral grounds, I don't know anyone else who does the same, and most of the rest of the post felt like it entirely missed the point somewhere adjacent to that.

I do know some people who have a "zero COVID" viewpoint. Exactly zero of them hold the view you describe as "if you failed to take a precaution and ended up getting infected, then you are morally responsible for any negative effects that this caused". Likewise the distinctions you draw seem to completely miss the point in most of the other categories, where I also personally know people who hold the corresponding views.

This doesn't invalidate your post, but I thought it might be worth presenting counter-evidence, even if it is a very small sample. Perhaps it's a cultural variation between wherever you live and where I live?

Perhaps moral blameworthiness is not the right phrasing. I think there is a mindset where the possibility of catching covid is unacceptable in a way which is qualitatively different than other risks. Does that match with your experiences?

I think you're on-target both about covid and in general, about risk analysis vs. a safety heuristic. There are even degrees of this; even a motorcyclist who drove safely and practiced all reasonable safety precautions would be heckled by someone if they were to crash through no fault of their own, die, and leave behind their family behind. "What did they think would happen?" You could even say that the common victim-blaming tropes are reinforcing a norm that puts safety permanently out of reach, so that they are always "morally blameworthy" or in other words, responsible, for what's happened. 

I think the "qualitative difference" you're describing is just the safety/risk dichotomy. Risk is irrelevant to safety, and safety is not a way of comparing risks.

There possibly is a moral dimension, but much more about risk of spreading it to others than about catching it. Individuals are responsible for limiting how likely they are to spread it on a moral dimension, but only responsible for avoiding catching it on the usual pragmatic dimension like riding a motorcycle.

Compare: "I went to a nightclub and caught COVID there" vs "I had a positive test but was feeling fine so I went to work anyway". The former is more likely to be viewed as risky behaviour while the latter is more likely to be viewed as immoral behaviour.

Even then, this isn't specific to zero-COVID at all in my experience. The distinction between various policies in the people I've talked to seems to be more about different models of disease, costs, and benefits over various timescales than anything moral at all.

An anonymous friend to whom I sent this post writes:

He has a good point that most people just want to do the universal “safety” precautions. I think a big reason that he doesn’t mention is that reasonable precautions are how all businesses defend themselves from lawsuits (e.g. sexual harassment and DEI training); as long as they take the reasonable precautions, then they are immune from lawsuits. But I don’t buy “safety” as an explanation for what policies are possible. It sounds like a just-so story for why we are in the mess that we are in. Vaccines are available and thus a reasonable precaution because the government subsidized them. Respirators and HVAC filters could have been a reasonable precaution if governments had subsidized and encouraged them. I think better leadership could have made a difference.

I wish the federal government made it patriotic to make PPE and safety retrofits. They should have announced programs to set up melt blown respirator manufacturing in America. They should have made a tax credit (like the solar tax credit) for retrofits that improve ventilation in retail or offices.

I think it is plausible that simple ventilation (open a window) could have been a common precaution like masks were. However there are a few reasons why serious ventilation (like HEPA filters) could not have been subsidized like vaccines were.

  1. Everybody agreed at the start that vaccines were the ultimate goal, ventilators would have needed to build consensus at a time when they were unavailabile.

  2. Vaccines only needed money from the government, ventilation would require much more infrastructure (approving ventilation plans on a per building level)

  3. Universal ventilation is much more expensive than vaccines, and for the reasons described in the post non-universal solutions weren't of interest.

I think there is a potential path where it could have happened but i think any such plan to implement would need to address these challenges head on. The reason no government could subsidize ventilation is not because of stupidity but because these pressures were too strong.

Aggressive ventilation is unreachable, not useless or harmful

What do you mean by "unreachable"? Impossible physically? Too expensive?

Too many COVID patients need ventilation device.

This seems related to Zvi's binary of "okay mode" vs. "not okay mode" that he wrote about here.

Your model matches up to my read of how the laypeople I talk to speak about COVID risks. I'm not sure it's the right explanation of the initially strange-looking set of interventions that people perform. For example, some public places have Plexiglass separators to simulate social distancing, and others don't. I think that few would claim that the Plexiglass is necessary to attain safety, but it is still popular. I also don't think this explanation accounts for the dramatic variation in interventions applied in different countries.

Plexiglass separators were a reasonable precaution in the days back when the mainstream view was that the disease was spread mostly by large droplets that mostly fell within seconds. They seem less useful in these days when nearly everyone gives higher credence to primarily aerosol spread.

That said, we still don't have great data on how easily COVID spreads in practice through various transmission routes. Maybe they do significantly reduce probability of transmission after all.

There's fairly decent, real-world evidence that covid spreads almost exclusively by aerosols. There doesn't seem to be much outdoor transmission, and that rules out direct droplet transmission. There's nearly zero evidence for fomite transmission. Also, indoor transmission seems to be required for transmission. All of that (and more) points to aerosols.

For the semi-random set of interventions I don't have a perfect explanation. My best guess is that at the start of the pandemic there was a chaotic period where random measures were tried (like plexiglass dividors) at some point, when things crystallized, removing a safety measure already in place was seen as unsafe.

With regards to the international variation, at least within the Western hemisphere, what examples are you thinking of?

The three perspectives map to some extent on to the political discussion in Germany. Currently there is a debate whether there should be a general vaccination mandate (there is already a mandate for health care workers). The parlament will not decide along party lines (which is quite atypical for the German political system), currently there are three possible bills:

a) A general vaccination mandate (18+ years), the reasoning behind that look somewhat like "Zero COVID" (but maybe it's a combination of "Zero COVID" and "Flatten the Curve").

b) A vaccination mandate for older citizens (50+ years) because those groups are primarily at risk and a high number of infections there puts too much stress on the healthcare systems (leading to capacity restraints for other urgent treatments, e.g. cancer surgery), "Flatten the Curve".

c) The rejection of a general vaccination mandate because everybody has the possibility of protecting themselves by vaccination, "Focused Protection".