In this post, I describe four increasingly strict levels of social
distancing that have been considered in response to the coronavirus
situation. I then talk about how likely each is to succeed, how
sustainable it is, and the relative economic and social cost. I then
talk about the likely time periods that we’ll transition between these
levels. The purpose here is to simply help think systematically about
trade-offs and transitions and identify points of agreement. I’ve
tried to indicate confidence levels for various claims, but am
probably overconfident about many specific things.
The four level of social distancing
The levels
What exactly falls in a given level is subject to debate, but is not
the focus here.
Level 1 (avoid large gatherings) Avoid sporting events, large
lectures, conferences, political rallies, demonstrations, rave
parties, cinema halls, cruise ships, prisons, and other
environments that involve close contact with large numbers of
people.
Level 2 (do all work and recreation remotely where feasible, and
avoid crowding when meeting in person): Run schools,
universities, and workplaces remotely, except those that need
in-person interaction (for instance, retail storefronts and
restaurants). Postpone recreational travel and substitute virtual
entertainment for physical entertainment where feasible. Go to
restaurants and cinema halls sparingly and when they are less
crowded.
Level 3 (flexible lockdown, “stay home except for essential
needs”): Whatever workplaces can be run remotely, run
remotely. For others, shut them down unless they serve “essential
needs”. For instance, shut down nail salons, spas, gyms, etc. Only
things like grocery stores, convenience stores, hardware stores,
and takeout food are allowed. Restrict travel, both regional and
across cities, to essential needs. Go out only for exercise and
buying essentials.
Level 4 (strict curfew-style lockdown, with police
enforcement): Similar to Level 3, but with stricter
restrictions. Mass transit, both intra-city and inter-city, may be
stopped for everybody except essential workers. Even takeout food
may not be allowed. If you leave the house, you may be stopped by
lockdown enforcers and may have to submit documentation or show an
identity card to justify yourself.
Chart 15 in the hammer and the
dance
is somewhat similar in that it lays out various levels of social
distancing. However, it’s much more complicated. If you are familiar
with the jargon of that post, levels 3 and 4 are probably the only
levels that deserve the “hammer” moniker, whereas levels 1 and 2 can
be the relatively relaxed portions of the “dance”.
TL;DR: timelines
Note that the gradation from optimistic to pessimistic for one
transition may not correlate with the super-optimistic case for
another transition (e.g., perhaps going down from level 3 to level 2
earlier increases the expected delay in going down from level 2 to
level 1). The use of “summer” in the table below refers to the
northern hemisphere summer.
Translations:
Super-optimistic case ~ 84th percentile of distribution
Reasonably optimistic case ~ 68th percentile of distribution
Median case ~ 50th percentile of distribution
Pessimistic but non-catastrophic case ~ 16th percentile of distribution
Transition
Super-optimistic case
Reasonably optimistic case
Median case
Pessimistic but non-catastrophic case
Most of the world goes down from level 3 to level 2
mid-April
mid-May
mid-June
mid-August
Most of the world goes down from level 2 to level 1
end of June
mid-August
summer 2021
summer 2022
Most of the world goes down from level 1 to business-as-usual
end of September
summer 2021
summer 2022
summer 2023
More about each level
Level 1 (avoid large gatherings)
Avoid sporting events, large lectures, conferences, political rallies,
demonstrations, rave parties, cinema halls, cruise ships, prisons, and
other environments that involve close contact with large numbers of
people.
Despite the wide range of uncertainty around coronavirus, my guess is
that Level 1 is the bare minimum necessary till coronavirus is fully
under control (for instance, through a widely available vaccine that
has been administered to a large proportion of the population,
including the individuals who want to be part of large gatherings).
Expected duration that we need to maintain at least level 1, and that
there will be a general social consensus that we need to do so:
Somewere between 6 months (super-optimistic) and 3 years (pessimistic).
My confidence in this range: Reasonably high (around 70%)
I expect this to not be too controversial with the LessWrong
readership.
Predictions about businesses: Sports stadiums, cruise ships, and
cinema halls are in for a very difficult time.
Level 2 (do all work and recreation remotely where feasible, and avoid crowding when meeting in person)
Run schools, universities, and workplaces remotely, except those that
need in-person interaction (for instance, retail storefronts and
restaurants). Postpone recreational travel and substitute virtual
entertainment for physical entertainment where feasible. Go to
restaurants and cinema halls sparingly and when they are less crowded.
In the last few weeks, large parts of the world, including many places
in United States, Europe, and South Asia, transitioned from level 1 to
level 2, and some quickly sped to level 3, after staying in level 2
for less than two weeks. Places that have had strong systems of
diagnosis, testing, and contact tracing have been able to stick to
level 2 and not escalate to level 3 (South Korea, Hong Kong,
Singapore, and Taiwan arguably fit this description). Other areas,
where diagnosis, testing, and contact tracing were not in place early
enough, did not have confidence that they could contain the epidemic
at level 2 and escalated to level 3 or level 4.
I believe there’s a good chance that the next few months of data will
show that we need to maintain at least level 2 until coronavirus is
mostly conquered (through a proven treatment or vaccine that has
started being administered). My reason for believing that at least
level 2 is needed: as far as I can make out, a lot of the transmission
of coronavirus happened outside of contexts that can be avoided
through level 1 of social distancing. If the world had followed level
1 consistently, the outbreak may not have reached pandemic
proportions, but it would probably still be growing in numbers.
It is possible that the data, once colllected and studied, will show
that level 1 of social distancing, combined with some specific
precautions gleaned from the data, are enough. However, even if ours
is a world where the ground truth is that level 1 of social distancing
is enough, I expect that rigorously demonstrating this will take at
least a few months. And I expect that nothing short of a rigorous
demonstration will lead to a general social consensus to go down to
level 1.
Expected duration that we need to maintain at least level 2, and that
there will be a general social consensus that we need to do so:
Somewhere between 3 months (super-optimistic) and 2 years
(pessimistic).
My confidence in this range: Reasonably high (around 70%)
I expect this to be a little more controversial with the LessWrong
readership, but not too much.
Predictions about businesses: Distance education and remote work will
thrive relative to their brick-and-mortar counterparts, and the tools
that enable these have tremendous growth potential. Similarly, home
entertainment will make big strides relative to meatspace
entertainment.
Comparing level 1 and level 2 for the economy: Neither level 1 nor
level 2 completely cripples the economy. With that said, level 1
affects only very specific industries, whereas level 2 has
wide-ranging effects, creating both winner and losers.
Level 3 (flexible lockdown, “stay home except for essential needs”)
Whatever workplaces can be run remotely, run remotely. For others,
shut them down unless they serve “essential needs”. For instance, shut
down nail salons, spas, gyms, etc. Only things like grocery stores,
convenience stores, hardware stores, and takeout food are
allowed. Restrict travel, both regional and across cities, to
essential needs. Go out only for exercise and buying essentials.
In the last few weeks, large parts of the world transitioned from
level 1 to level 2 to level 3, or even directly from level 1 to level
3. Some East Asian countries were able to avoid going all the way to
level 3, or needed to go there only for brief periods of
time. Examples of regions that didn’t need to go to level
3
(outside of very specific geographies or short time periods) South
Korea,
Singapore,
Hong Kong, and Taiwan. A region that went to level 3 or 4 and is on
the path to returning to level 2 is China’s Hubei province.
Data in the coming weeks, particularly data on the mostly East Asian
countries that have managed to either avoid escalating to, or
de-escalated from level 3, will be crucial in figuring out how long we
need to sustain level 3.
One question might be: why would the answer to “does it make sense to
de-escalate from level 3 to level 2?” change over time? There are
two mechanisms by which the answer might change:
Preparation time: Staying in level 3 for a little while slows down
the rate of infection a lot, which gives people time to ramp up on
production of medical equipment, and catch up with the backlog in
testing and contact tracing work. Once that catch-up is complete,
quarantining infected or at-risk individuals is good enough, and the
rest of us can go down to level 2.
Information value: Maybe it actually makes sense to stay in level 2
all along, and an omniscient being would know it. But, we don’t know
it yet. From a risk perspective, it makes sense for any part of the
world that has crossed some threshold of infection to escalate to
level 3. Later, when the data on new infection rates is in, it may
turn out that level 2 was “good enough”, and that even with level 2,
the basic reproduction number (R0) was already much less than 1. But
right now, we can’t risk it. Note in particular that this argument
benefits a lot from the fact that some parts of the world are only
at level 2; the data from these regions over the next few weeks and
months will be crucial to making that case that level 2 is
sustainable.
Expected duration that we need to maintain at least level 3, and that
there will be a general social consensus to do so: This is the part I
am most uncertain about, and also the one that I feel has the most
cross-region variation. Loosely, here is what I expect:
I expect that it will take at least till the end of April for there
to be enough data to make a confident case that level 2 is enough in
equilibrium (i.e., the “information value” side of the argument).
I expect that regions where coronavirus can be contained (i.e.,
there are either no cases, or all cases are caught through contact
tracing and there are no unexpected cases) will not need to go to
level 3 at all. But that regions that have needed to go to level 3
will take at least a month to return to level 2, and possibly two to
three months, or even longer.
Concretely:
I expect that already-overwhelmed regions such as New York City and
Italy will stay in level 3 till at least the end of May, and
possibly till July or August.
However, I expect that other regions that are actually not
overwhelmed, will be able to get out of level 3 around mid-June
(super-optimistic: mid-April, pessimistic: August). I emphasize the
“actually” there because there could be many regions that
already have a lot of undiagnosed cases, and are already
overwhelmed; it just isn’t showing up in their statistics yet.
The time range of late April to mid-May is under the optimistic
assumption that level 2 is actually sustainable (and that the
experiences of China, South Korea, Hong Kong, Singapore, and Taiwan
continue to show this) and that the preparation time is
sufficient.
Predictions about businesses: Restaurants, nail salons, spas, gyms,
etc. will see a huge hit for the next few months. Even after flexible
lockdown ends, traffic to them will probably rebound slowly. Those
that lack the cash to get through the few months may go bankrupt, or
get sold.
Comparing level 2 and level 3 for the economy: Level 3 has a much
higher impact on the economy than level 2. Not only does it hurt a
bunch of retail storefront businesses, it also massively shifts
consumer demand patterns, which require a lot of supply chain
reconfiguration (for instance, moving food demand away from
restaurants and toward groceries). The uncertainty of duration of
level 3 further complicates matters; normally, supply chain
reconfiguration takes more time. Thus, even after we de-escalate from
level 3, the reconfigured supply chain may make it harder to go back
to normal.
Level 4: strict curfew-style lockdown, with police enforcement
Level 4 is similar to Level 3, but with stricter restrictions. Mass
transit, both intra-city and inter-city, may be stopped for everybody
except essential workers. Even takeout food may not be allowed. If you
leave the house, you may be stopped by lockdown enforcers and may have
to submit documentation or show an identity card to justify yourself.
The lockdowns in Wuhan (China), India, and many European countries are
at level 4; some European countries straddle the line between level 3
and level 4.
I think the case for level 4 as opposed to level 3 is unclear;
however, this might be a hard matter to study. A justification offered
for level 4 is low compliance with level 3. This may be more of an
issue in some places than others, which makes cross-region comparisons
harder. For instance, reading articles like
this
might lead one to argue that strict enforcement is necessary for
social distancing to be successful in India, making the case for level
4 instead of level 3. However, articles like
this
suggest that the threat or fear of a strict lockdown may itself cause
people to crowd more as they rush to stock up on food or return to
their home town, whereas a commitment to a more flexible lockdown
would lead to less panic and less crowding.
Other than compliance levels, I think there is very little to say in
favor of level 4 instead of level 3. I believe that if level 3 is not
sufficient to bring the basic reproduction number (R0) to less than 1,
and to generally keep the pandemic under control, level 4 will not be
enough either.
Coronavirus: the four levels of social distancing, and when and why we might transition between them
In this post, I describe four increasingly strict levels of social distancing that have been considered in response to the coronavirus situation. I then talk about how likely each is to succeed, how sustainable it is, and the relative economic and social cost. I then talk about the likely time periods that we’ll transition between these levels. The purpose here is to simply help think systematically about trade-offs and transitions and identify points of agreement. I’ve tried to indicate confidence levels for various claims, but am probably overconfident about many specific things.
The four level of social distancing
The levels
What exactly falls in a given level is subject to debate, but is not the focus here.
Level 1 (avoid large gatherings) Avoid sporting events, large lectures, conferences, political rallies, demonstrations, rave parties, cinema halls, cruise ships, prisons, and other environments that involve close contact with large numbers of people.
Level 2 (do all work and recreation remotely where feasible, and avoid crowding when meeting in person): Run schools, universities, and workplaces remotely, except those that need in-person interaction (for instance, retail storefronts and restaurants). Postpone recreational travel and substitute virtual entertainment for physical entertainment where feasible. Go to restaurants and cinema halls sparingly and when they are less crowded.
Level 3 (flexible lockdown, “stay home except for essential needs”): Whatever workplaces can be run remotely, run remotely. For others, shut them down unless they serve “essential needs”. For instance, shut down nail salons, spas, gyms, etc. Only things like grocery stores, convenience stores, hardware stores, and takeout food are allowed. Restrict travel, both regional and across cities, to essential needs. Go out only for exercise and buying essentials.
Level 4 (strict curfew-style lockdown, with police enforcement): Similar to Level 3, but with stricter restrictions. Mass transit, both intra-city and inter-city, may be stopped for everybody except essential workers. Even takeout food may not be allowed. If you leave the house, you may be stopped by lockdown enforcers and may have to submit documentation or show an identity card to justify yourself.
Chart 15 in the hammer and the dance is somewhat similar in that it lays out various levels of social distancing. However, it’s much more complicated. If you are familiar with the jargon of that post, levels 3 and 4 are probably the only levels that deserve the “hammer” moniker, whereas levels 1 and 2 can be the relatively relaxed portions of the “dance”.
TL;DR: timelines
Note that the gradation from optimistic to pessimistic for one transition may not correlate with the super-optimistic case for another transition (e.g., perhaps going down from level 3 to level 2 earlier increases the expected delay in going down from level 2 to level 1). The use of “summer” in the table below refers to the northern hemisphere summer.
Translations:
Super-optimistic case ~ 84th percentile of distribution
Reasonably optimistic case ~ 68th percentile of distribution
Median case ~ 50th percentile of distribution
Pessimistic but non-catastrophic case ~ 16th percentile of distribution
More about each level
Level 1 (avoid large gatherings)
Avoid sporting events, large lectures, conferences, political rallies, demonstrations, rave parties, cinema halls, cruise ships, prisons, and other environments that involve close contact with large numbers of people.
Despite the wide range of uncertainty around coronavirus, my guess is that Level 1 is the bare minimum necessary till coronavirus is fully under control (for instance, through a widely available vaccine that has been administered to a large proportion of the population, including the individuals who want to be part of large gatherings).
Expected duration that we need to maintain at least level 1, and that there will be a general social consensus that we need to do so: Somewere between 6 months (super-optimistic) and 3 years (pessimistic).
My confidence in this range: Reasonably high (around 70%)
I expect this to not be too controversial with the LessWrong readership.
Predictions about businesses: Sports stadiums, cruise ships, and cinema halls are in for a very difficult time.
Level 2 (do all work and recreation remotely where feasible, and avoid crowding when meeting in person)
Run schools, universities, and workplaces remotely, except those that need in-person interaction (for instance, retail storefronts and restaurants). Postpone recreational travel and substitute virtual entertainment for physical entertainment where feasible. Go to restaurants and cinema halls sparingly and when they are less crowded.
In the last few weeks, large parts of the world, including many places in United States, Europe, and South Asia, transitioned from level 1 to level 2, and some quickly sped to level 3, after staying in level 2 for less than two weeks. Places that have had strong systems of diagnosis, testing, and contact tracing have been able to stick to level 2 and not escalate to level 3 (South Korea, Hong Kong, Singapore, and Taiwan arguably fit this description). Other areas, where diagnosis, testing, and contact tracing were not in place early enough, did not have confidence that they could contain the epidemic at level 2 and escalated to level 3 or level 4.
I believe there’s a good chance that the next few months of data will show that we need to maintain at least level 2 until coronavirus is mostly conquered (through a proven treatment or vaccine that has started being administered). My reason for believing that at least level 2 is needed: as far as I can make out, a lot of the transmission of coronavirus happened outside of contexts that can be avoided through level 1 of social distancing. If the world had followed level 1 consistently, the outbreak may not have reached pandemic proportions, but it would probably still be growing in numbers.
It is possible that the data, once colllected and studied, will show that level 1 of social distancing, combined with some specific precautions gleaned from the data, are enough. However, even if ours is a world where the ground truth is that level 1 of social distancing is enough, I expect that rigorously demonstrating this will take at least a few months. And I expect that nothing short of a rigorous demonstration will lead to a general social consensus to go down to level 1.
Expected duration that we need to maintain at least level 2, and that there will be a general social consensus that we need to do so: Somewhere between 3 months (super-optimistic) and 2 years (pessimistic).
My confidence in this range: Reasonably high (around 70%)
I expect this to be a little more controversial with the LessWrong readership, but not too much.
Predictions about businesses: Distance education and remote work will thrive relative to their brick-and-mortar counterparts, and the tools that enable these have tremendous growth potential. Similarly, home entertainment will make big strides relative to meatspace entertainment.
Comparing level 1 and level 2 for the economy: Neither level 1 nor level 2 completely cripples the economy. With that said, level 1 affects only very specific industries, whereas level 2 has wide-ranging effects, creating both winner and losers.
Level 3 (flexible lockdown, “stay home except for essential needs”)
Whatever workplaces can be run remotely, run remotely. For others, shut them down unless they serve “essential needs”. For instance, shut down nail salons, spas, gyms, etc. Only things like grocery stores, convenience stores, hardware stores, and takeout food are allowed. Restrict travel, both regional and across cities, to essential needs. Go out only for exercise and buying essentials.
In the last few weeks, large parts of the world transitioned from level 1 to level 2 to level 3, or even directly from level 1 to level 3. Some East Asian countries were able to avoid going all the way to level 3, or needed to go there only for brief periods of time. Examples of regions that didn’t need to go to level 3 (outside of very specific geographies or short time periods) South Korea, Singapore, Hong Kong, and Taiwan. A region that went to level 3 or 4 and is on the path to returning to level 2 is China’s Hubei province.
Data in the coming weeks, particularly data on the mostly East Asian countries that have managed to either avoid escalating to, or de-escalated from level 3, will be crucial in figuring out how long we need to sustain level 3.
One question might be: why would the answer to “does it make sense to de-escalate from level 3 to level 2?” change over time? There are two mechanisms by which the answer might change:
Preparation time: Staying in level 3 for a little while slows down the rate of infection a lot, which gives people time to ramp up on production of medical equipment, and catch up with the backlog in testing and contact tracing work. Once that catch-up is complete, quarantining infected or at-risk individuals is good enough, and the rest of us can go down to level 2.
Information value: Maybe it actually makes sense to stay in level 2 all along, and an omniscient being would know it. But, we don’t know it yet. From a risk perspective, it makes sense for any part of the world that has crossed some threshold of infection to escalate to level 3. Later, when the data on new infection rates is in, it may turn out that level 2 was “good enough”, and that even with level 2, the basic reproduction number (R0) was already much less than 1. But right now, we can’t risk it. Note in particular that this argument benefits a lot from the fact that some parts of the world are only at level 2; the data from these regions over the next few weeks and months will be crucial to making that case that level 2 is sustainable.
Expected duration that we need to maintain at least level 3, and that there will be a general social consensus to do so: This is the part I am most uncertain about, and also the one that I feel has the most cross-region variation. Loosely, here is what I expect:
I expect that it will take at least till the end of April for there to be enough data to make a confident case that level 2 is enough in equilibrium (i.e., the “information value” side of the argument).
I expect that regions where coronavirus can be contained (i.e., there are either no cases, or all cases are caught through contact tracing and there are no unexpected cases) will not need to go to level 3 at all. But that regions that have needed to go to level 3 will take at least a month to return to level 2, and possibly two to three months, or even longer.
Concretely:
I expect that already-overwhelmed regions such as New York City and Italy will stay in level 3 till at least the end of May, and possibly till July or August.
However, I expect that other regions that are actually not overwhelmed, will be able to get out of level 3 around mid-June (super-optimistic: mid-April, pessimistic: August). I emphasize the “actually” there because there could be many regions that already have a lot of undiagnosed cases, and are already overwhelmed; it just isn’t showing up in their statistics yet.
The time range of late April to mid-May is under the optimistic assumption that level 2 is actually sustainable (and that the experiences of China, South Korea, Hong Kong, Singapore, and Taiwan continue to show this) and that the preparation time is sufficient.
Predictions about businesses: Restaurants, nail salons, spas, gyms, etc. will see a huge hit for the next few months. Even after flexible lockdown ends, traffic to them will probably rebound slowly. Those that lack the cash to get through the few months may go bankrupt, or get sold.
Comparing level 2 and level 3 for the economy: Level 3 has a much higher impact on the economy than level 2. Not only does it hurt a bunch of retail storefront businesses, it also massively shifts consumer demand patterns, which require a lot of supply chain reconfiguration (for instance, moving food demand away from restaurants and toward groceries). The uncertainty of duration of level 3 further complicates matters; normally, supply chain reconfiguration takes more time. Thus, even after we de-escalate from level 3, the reconfigured supply chain may make it harder to go back to normal.
Level 4: strict curfew-style lockdown, with police enforcement
Level 4 is similar to Level 3, but with stricter restrictions. Mass transit, both intra-city and inter-city, may be stopped for everybody except essential workers. Even takeout food may not be allowed. If you leave the house, you may be stopped by lockdown enforcers and may have to submit documentation or show an identity card to justify yourself.
The lockdowns in Wuhan (China), India, and many European countries are at level 4; some European countries straddle the line between level 3 and level 4.
I think the case for level 4 as opposed to level 3 is unclear; however, this might be a hard matter to study. A justification offered for level 4 is low compliance with level 3. This may be more of an issue in some places than others, which makes cross-region comparisons harder. For instance, reading articles like this might lead one to argue that strict enforcement is necessary for social distancing to be successful in India, making the case for level 4 instead of level 3. However, articles like this suggest that the threat or fear of a strict lockdown may itself cause people to crowd more as they rush to stock up on food or return to their home town, whereas a commitment to a more flexible lockdown would lead to less panic and less crowding.
Other than compliance levels, I think there is very little to say in favor of level 4 instead of level 3. I believe that if level 3 is not sufficient to bring the basic reproduction number (R0) to less than 1, and to generally keep the pandemic under control, level 4 will not be enough either.