Last time on Deep Dive, we learned about the game theory behind experts’ unwillingness to admit when they don’t know the answer to a complex policy question. This week, we’ll look at the opposite question: Why do some legitimate experts —  with real information about how a crisis may unfold — choose not to offer predictions about that crisis even while quacks with no real information make grand pronouncements?

Evaluating predictions is a tricky game. For one thing, predictions are usually made using private information —  that is, information that’s inaccessible to people trying to evaluate the predictions. For another, it’s impossible to know if the prediction will be accurate until events actually take place, at which point the value of the prediction drops to zero because the events predicted are in the past. That makes it very easy for predictors to lie and claim to have access to private information when actually they’re just guessing. To believe a prediction has value, we have to believe that the predictor has access to some useful private information before we actually have the ability to tell whether that’s true.

One way to think about that belief in the predictor’s access to private information is as the predictor’s reputation. If someone claiming to be an expert has a strong reputation within their field, we’re more likely to think they have insights into that field that we lack. Of course, in that environment, reputation is a crucial resource, to be guarded jealously. As it turns out, the need to protect reputation makes for some strange incentives around prediction-making for top experts.

Economists Aleksei Smirnov and Igor Starkov published a paper late last year examining how subject matter experts protect their reputations when making predictions. They made a game-theoretic model that reflects how difficult it is for observers to tell people with real insights —  who Smirnov and Starkov call “seers” —  from people who are taking a shot in the dark and hoping to be proved rights —  “quacks,” in Smirnov and Starkov’s parlance.

In the long run, the quacks get punished. Their predictions tend to be wrong and observers notice that. In the short run, however, differentiating seers and quacks is impossible, unless the quacks decide to answer Isaac Chotiner’s phone calls. As a result, anyone who makes any kind of prediction sees a short-term hit to their reputation. If a prediction is proved right then that predictor recovers their reputation and then some after the fact, but for any prediction the short run reputational cost is unavoidable. The only reliable way to win in the short run, therefore, is not to play. In Smirnov and Starkov’s model, experts who refuse to make any predictions see their reputations gradually grow -- they pay no cost for sitting a given situation out.

Of course, if you can make reputational gains by doing nothing, why does anyone ever make a prediction? The answer, according to Smirnov and Starkov, is that in order for there to be a robust prediction market, the reward for being the highest-rated predictor has to be substantial enough for a quack to want to risk their reputation by guessing at it. Like, shaping-national-pandemic-response-strategy substantial.

Halima Gikandi reported on the threat COVID-19 poses to people in Immigration and Customs Enforcement (ICE) detention. On Tuesday, ICE announced the first confirmed case in one of its detention facilities, but the agency has taken few measures to protect the tens of thousands of immigrants in its custody. Detainees report overcrowding, soap shortages, and lack of masks and gloves for inmates, all of which increase the risk of COVID-19 spreading quickly in ICE facilities.

Kelsey Atherton traced the history of comorbidity between disease and war. Illness often follows the destruction wrought by conflict —  immiseration loves company —  but some commanders have used disease as a weapon. In ancient sieges, early practitioners of biological warfare introduced diseases into cities hemmed in by enemies at the gates, hoping to sap defenses. Today, the Trump administration has increased sanctions against COVID-19-ravaged Iran and trumpeted the virus’s spread there as a victory in the administration’s ongoing attempt to besiege the country.

Durrie Buscaren reviewed the toll COVID-19 has taken in Iran, where the virus has reached a scale seen only in Wuhan, China, northern Italy, and New York City. At least 2,234 died of COVID-19 in Iran as of Friday, and American sanctions are limiting the government’s ability to bring protective equipment and medical supplies into the country. An Iranian academic study predicts the virus may kill 3.5 million Iranians by the summer, which would be three times the number who died in the devastating Iran-Iraq war four decades ago.

Paul Frymer and Jacob Grumbach have a fascinating new paper out demonstrating that union membership reduces racial resentment among white workers. It’s worth a read, but it makes Well Played because of the dance party the Grumbach clan held when it got accepted into the American Journal of Political Science.

How many space trees do you need to make a space forest? Presumably more than one.

This is what Gen-Z survivalism looks like.

Never let it be said that Hollywood lacks historical sensitivity.