Better relevant than right

When George Box famously said “All models are wrong, some are useful”, he forgot to mention that some can also be harmful.

The book The Drunkard’s Walk by the physicist Leonard Mlodinow, tells the story of the OJ Simpson trial. Simpson’s lawyer, the same hired by DJ Trump to take him off the hook, this time for the impeachment process, convinced the jury that since the probability that if a woman is found dead the husband is the perpetrator is low, the probability that Simpson murdered his wife was likewise low. The probability estimate that the lawyer brought to court was correct but irrelevant. The relevant piece of information here is that given that a wife is found dead and the husband has antecedents of beating her up, the probability that the husband murdered the wife is high. Using the probability function, P(wife murdered, husband is the murderer) is low but the relevant probability is P(wife murdered, husband is the murderer| husband is a wife’s abuser). The jury, persuaded by the lawyer, failed to grasp the enormous difference between a true statement and a relevant one, and the accuser was acquitted.

While we may fail to reckon when we are wrong, we can nevertheless be quite apt to point out someone else wrongness. The hall of fame of the pitfalls of predictions contains large errors of famous people. For instance, Bill Gates’ prediction that personal computer will need no more than 4K of memory, Paul Krugman’s wager that the internet’s effect on the economy will not be greater than the fax’s machine, or the space entrepreneur and ubermensch à la mode Elon Musk early 2020 remark that “is dumb to worry about the virus”.

Admittedly these misjudgments and pitfalls do not out and in themselves cause harm, and can even have a positive net effect for being a source of pleasure to schadenfreude adepts and Monday morning quarterbacking aficionados. There are, however, cases in which an opinion or idea from someone in a position of power or professional prestige can result in great pain and suffering. The internet is an indelible register at planetarian scale, to watch the admonitions and forecasts of some experts in virology and public health just a few weeks before entire countries were shutdowns and improvised morgues need to be settled to cope with the incessant number of deceased bodies, is a bizarre experience. Wrong models are not necessarily innocuous, they can deliver harm.

But, it is enough to be right? Elon Musk corrected his previous misjudgment and rightly said that exponential curves do not occur in the natural world. The Space X CEO reminded us Earthians that a realistic curve of contagion would look more like a sigmoid function, that is, the number of cases will grow fastly but it will reach a saturation point, a plateau, or in today’s mediatic parlance, the curve will flatten. This indictment is precisely correct, any biological phenomenon cannot grow exponentially unchecked, metabolic processes are necessarily bounded.

However, Elon’s point while right is not relevant. That the contagion curve has a sigmoid shape is beside the point, what really matters is to be able to tell when the curve saturates. If it saturates at 8 billion people deaths it would tell us that human life on planet earth would have ceased to exist. On the other hand, if it saturates at, let us say, 10 million people, for those unaffected it can be business as usual.

An important lesson for scientists and communicators is that unfortunately, to capture the public’s attention you don’t need to be right, you need to be seen as relevant. And to do not cause harm or misguide the public you need to be both relevant and right.

Nevertheless, there is a final third element to consider. In the attention economy, the principal product of mass media is not education or entertainment but attention, and to capture this most valuable commodity, the message needs to be delivered optimally. Thus, right, relevant, and well communicated, there are the three indispensable elements to produce an effective change for the better. This is the challenge that scientists need to be prepared for if we want to be listened to by the public and policymakers.