If Machines Ruled Us, Lockdowns Would Be Tougher

Last Sunday, our family went to church. 

Ordinarily, there’s nothing for me, a non-practicing Jew, or my wife, a lapsed Russian Orthodox Christian, to do at a Lutheran religious service like the one we attended at the Berliner Dom. But Berlin’s central cathedral has an exceptional organ, and Andreas Sieling plays it with dignity and subtlety. We’d rather have gone to the Philharmonie, but the government shut it down as part of its anti-Covid 19 policy. Closing down churches, however, turned out to be a step too far for the governing Christian Democrats. Chancellor Angela Merkel called on fellow legislators to listen to science in a widely praised December speech; she’s a physicist — but she’s also a clergyman’s daughter.

At the end of the service, the preacher complained that the cathedral was in dire financial straits because it could no longer hold concerts. I don’t think there’s a valid scientific explanation why Sieling’s artistry and the choir’s inspired singing, heard by several hundred properly masked and distanced people, are allowed as part of a church service but not when the event doesn’t include a sermon. God’s word doesn't have a direct effect on the virus, after all: The same preacher reminded the flock to keep the required distance when exiting.

The experience got me thinking about the technocratic approach to handling the pandemic, the “listen to science” approach. Would there be more logic to the policies if they were devised by an artificial intelligence with access to all the available data? Would lockdown rules make more sense if the job of minimizing deaths were left to a machine not guided by emotions, fear of failure, political and religious convictions, greed, ego — in short, by everything that makes us humans different from machines? 

The answer is, the lockdowns would probably be even tougher. That’s because data from the first wave point to the effectiveness of limiting contact in every possible way. And economic considerations wouldn’t be a mitigating factor: Although intuition tells us that when the economy tanks, more people tend to die, data do not quite bear this out. 

Listening to science during a pandemic means setting a simple policy goal: To keep as many people as possible from dying. A purely technocratic approach to this goal would mean designing a complex model that would assess the available data on every possible measure that has been tried in the world since the pandemic began a year ago. The model’s recommendations would then have to be followed to the letter without any discussion. Is anyone against against saving lives? If so, lock them up.

Based on the data, the model would likely decide to go with the toughest restrictions possible. In November, Nils Haug, Lukas Geyrhofer and Alessandro Londei from the Medical University of Vienna ranked the effectiveness of different government interventions using several statistical methods. They found that what works best are “curfews, lockdowns and closing and restricting places where people gather in smaller or large numbers for an extended period of time. This includes small gathering cancellations (closures of shops, restaurants, gatherings of 50 persons or fewer, mandatory home working and so on) and closure of educational institutions.” Other studies also have found some of the most restrictive measures to be the most effective in terms of saving lives. 

Although most politicians see their Covid policy decisions as a trade-off between keeping down infection rates and keeping firms from going under, that’s not how a data-based model would have played it. Joan Ballester from the Barcelona Institute for Global Health wrote in a 2019 paper that countries hit hardest by the Great Recession also saw the biggest drops in mortality rates. Economic gloom tends to reduce work-related burnout and the associated use of harmful substances, cut traffic deaths and workplace accidents, decrease environmental pollution. According to Ballester (and some previous literature), these effects may well counterbalance the opposite trend: While unemployment does increase suicide and crime risks, the overall effect of recent major recessions on mortality appears to be negligible. 

Of course, a model decreeing the harshest contact restrictions and discounting the effects of economic hardship wouldn’t necessarily be right. Data on the efficiency of tough lockdowns comes from the first wave of the pandemic; in recent months, restrictions haven’t been as quick to reduce infection and death rates as they were in the first months of 2020. Besides, the pandemic’s economic damage may well exceed that of the 2008 financial crisis, and the available data on mortality trends may not be directly applicable. In other words, the decisions made by the machine would be subject to what’s known as model drift. But adjusting the model in line with the changing environment probably wouldn’t lead to any major policy reversals, if only because the toughest measures appear to work so much better than milder ones, such as educational programs and even infection tracking.

No politician can publicly admit that saving the maximum number of lives is not really the only goal, and that under certain conditions that goal can even take a back seat to other concerns, not necessarily economic ones. Politicians are often finely attuned to the feelings of their voters; they sense when most people can no longer sit obediently within their four walls, take over their children’s schooling, do without live music, the companionship of a bar, the soothing effect of a shopping trip; they sense when altruism begins to wane, choked by increasingly unbearable restrictions on liberty. Even the most technocratic politicians will only listen to science up to a point — namely, until the policies dictated by the data become unpopular.

In Germany, an overwhelming majority of people still supports the restrictions — and thus it makes sense to govern as the life-saving model would. But even so, there are small mercies like that church service. Thank God for the imperfections and the logic failures of human technocracy.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

Leonid Bershidsky is a member of the Bloomberg News Automation team based in Berlin. He was previously Bloomberg Opinion's Europe columnist. He recently authored a Russian translation of George Orwell's "1984."

©2021 Bloomberg L.P.

BQ Install

Bloomberg Quint

Add BloombergQuint App to Home screen.