by Anders Sandberg

In Scientific American Zeynep Tufekci writes:

Preparing for the almost inevitable global spread of this virus, … , is one of the most pro-social, altruistic things you can do in response to potential disruptions of this kind.

We should prepare, not because we may feel personally at risk, but so that we can help lessen the risk for everyone.

…you should prepare because your neighbors need you to prepare—especially your elderly neighbors, your neighbors who work at hospitals, your neighbors with chronic illnesses, and your neighbors who may not have the means or the time to prepare because of lack of resources or time.

I think this is well put. As a healthy middle-aged academic my personal risk of dying from Covid-19 seems modest – maybe about 0.4% if I get it, which in turn might be below 10% depending on how widespread the virus becomes. But I could easily spread the disease to people who are far more vulnerable, either directly or indirectly. Even slowing the spread is valuable since it helps avoid overloading the medical system at the peak of the epidemic.

One of the ways of reducing the spread of the virus is working from home. At my institute the health advice notes in the restrooms clearly state that if one feels under the weather one should stay home. I happen to know that in our case this is a supported norm: our director does want us to follow it, and I think it is also common knowledge that this is so. In other organisations this may not hold. Sure, notices tell you one thing, but how does your boss or co-workers really react? Some work is hard to do remotely, and showing up sends important signals. Some people dislike it. Some people cannot afford to not work.

I hope the current situation firmly establishes home working as the right thing to do and helps build the infrastructure for it (tech companies are of course happy with it, and shares in teleconferencing companies are apparently doing well despite the overall stock market trouble at the time of writing). But I also think we should do more working from home than we might normally think is rational – especially at large organisations.

The unilateralist curse

The reason is that this is a prime example of a unilateralist curse situation.

In a unilateralist curse situation, a number of actors can perform an action and it is enough that one of them does it for the effect to take place for everybody – both the actors and maybe outsiders too. This could be something like revealing a potential spoiler when talking about a film, revealing a security problem in common software, releasing a genetically modified organism into the environment, or undertaking geoengineering. For simplicity we can assume all actors do want to do the right thing and have the same incentives (obviously things tend to go worse if not). Each actor weighs their evidence to make the right decision and performs the action if, all things considered, it looks positive.

The problem is that the more actors there are, the more likely it is that one is going to make a mistake, have randomly biased evidence, or be too optimistic/pessimistic. With N actors there are N chances of mistakes. If the action actually is a net positive, on average more actors guarantee that it will happen – maybe somebody will miss the chance to do it, but on average most actors will do the right thing. But if the action has negative consequences, more actors will also almost guarantee that it happens. Somebody is going to be “that guy”.

This applies to the staying at home situation. When I wake up with a cough, is that enough evidence that I should stay home or not? Here the action is going to the office (rather than staying at home, which may look like an action since it is out of the ordinary: what matters is what causes global effects). Meanwhile many of my co-workers are also considering their health and trying to decide. It is enough that one sick person appears at the office for infection to be possible and consequences to ensue. (A slight variation from the pure scenario here is that if I stay home, I am also protected from that co-worker.)

What to do about the unilateralist curse? The most obvious insight is that it exists, and that one should recognize situations where it is relevant and try to act rationally in them. The rational response if there is no other information is to become more conservative in when to act: knowing there are more actors out there means that I should require a higher threshold for deciding to act than normally. Yes, it may feel I have weighed the evidence well and reached the right conclusion beforehand, but this consideration should make me demand even stronger reasons before I act.

This implies that the sickness or caution threshold for staying home should be lowered the larger the group I could affect. If I work at a tiny office, maybe not so much. If I work at a giant multinational, a lot. The larger the organisation, the more working from home should be the primary option.

Institutions matter (even if we stay home from them)

In our paper on the curse we note that resolving it individually is hard. The more information agents can share the better the decisions can become. The best approach is institutional: ensure that everybody has a framework to act in, share available information, make joint legitimate decisions, and help make them stick. This may be hard for information hazards and geoengineering, but fortunately current institutions are the right size for handling this kind of problem.

Most obviously there should be common knowledge that working from home is a good thing, at least when there is disease afoot – not just the abstract understanding and the reasoning in our paper, but the everyday awareness that this is good for you, me, and our organisation and we all know this. Establishing social norms are important. In large organisations it is also harder, so deliberate effort is needed.

One can support this formally by having policies ensuring that calling in sick will not be punished and ensuring that unpaid sick leave does not force ill people to come to work. One good effect of the current situation is that at least some HR management and labour law regulators are becoming more aware of the problem. But it might be that they should be far more proactive than they think.

The value of information is also high: if it is possible to test reliably for infection, it is worth doing. Healthy (or recovered immune) people can help maintain functions, sick people can definitely stay away. But this requires that the right actors bear the cost of gathering the information: cases in the US where coronavirus tests have cost individuals significant money discourage testing, making spread more likely. This is where enlightened self-interest and care about others make even a minarchist libertarian like me think testing is a public good we should collectively support – and make easy, non-stigmatized, and decision-relevant for everybody.

The cost of mistaken decisions in epidemic disease can be high. This means that one needs to act more cautiously than normally seems rational – especially in large organisations and groups. Rational people and organisations recognize this and adapt as the situation changes. They should also remember how to do this the next time this happens.

Acknowledgement: I thank Owain Evans for pointing out this implication of our paper to me.