Pancake Thinkers — Rationality and Epidemics
Rules and twists
Rules and twists
The word “rationality” comes from the root “ratio”. A ratio is the expression of the simplest relationship between the relative amount of two values. For instance, when you bake pancakes there is a basic recipe in which, for instance, you need 2 eggs etc. If you want to make more pancakes than the basic recipe you just need to multiply the quantity of every ingredient by the right amount. For instance you obtain a recipe with 4 eggs in which every quantity is multiplied by 2 that will lead to twice the amount of pancakes than the basic recipe. Mathematically, you can formalize the pancake recipe as follows. You consider a space with a dimension for each different ingredient. A point in this space represents a particular amount for each ingredient. Then the pancake recipe, is the set of all points that represent a valid pancake recipe according to the basic recipe. What you obtain as the set of all pancake combinations is a line: this is the geometrical expression of the ratio between ingredients. Linear thinking is not only a figure of speech that has invade our ways to express regularities (think of the word “rule” for instance !).
Linear thinking is a powerful tool. From a single pancake recipe you may derive an infinite number of valid recipes. But as every tool it is useful only if you use it the right way for the right problems. The epidemics of COVID19 has brought many instances of ill-applied linear thinking, and more generally ill-applied mathematical rigorous models. I call the practioners of such ill-advised application of mathematical models Pancake Thinkers.
Pancake thinkers are those people who literally cannot think in another way than a linear one (or using their pet mathematical theory). They apply mathematical models that are not adapted to the problem they are trying to solve and cannot change their minds because “mathematics don’t lie”. This is one of the worse combination: an infinite stubbornness in the error. They are trapped by their mental models.
Let us walk through some recent examples of Pancake Thinking that has led to immense damage through this crisis around the epidemics of COVID19.
The chloroquine controversy
Didier Raoult is a world renown virologist (he notably co-discovered giant viruses: mimiviruses, and proposed with his research team new treatments for infective endocarditis) and started an incredible controversy by publishing the 4th of March 2020 in the Int. Journal of Antimicrob Agents, an article recommending the use of chloroquine to treat COVID19 patients.
I am not going to take position on whether the treatment is efficient or not, but on how the controversy has unfolded. For some reasons this proposal has produced a backlash without proportion to its origin. Maybe a part of this finds its roots in the profound divide running across american politics: it is true that Donald Trump and Elon Musk have cited favorably this drug and that they are very controversial personalities to say the least. Another possibility is that the price of this molecule is ridiculously low and that there are no longer license needed to produce: it doesn’t feel like big bucks for the pharmaceutical industry... But I think that the major component of this backlash comes from a very narrow minded approach to science: academism. It is in this respect that I find this example representative of Pancake Thinking. The underlying idea is that if you don’t follow the “academically approved way of doing” you don’t have anything of value to say. The absurdity of such a position is well enlighten by this satirical piece stating that there were no randomized controlled trials to assess the efficiency of parachutes in preventing harm from trauma related to gravitational challenge. It is deeper than it looks at the first glance.
Epidemics fatalities
In these times of generalized lockdown, we don’t have the two minutes of hate but rather the daily show of “official numbers” of COVID19 cases and fatalities. These numbers have replaced meteo forecasts and are considered as
Solid truths accurately describing the reality.
The base on which hard decisions (severity of the lockdown, duration of it, assessment of the efficiency of the response, accurate tracking of the epidemics etc.) are made.
Both points are wrong on many levels. It is clear that the number of official cases is also depending on how many, and following what policy, tests are used. For instance in France, tests are limited to people that shows signs. In other terms we test people we more or less already know they have the disease, the rationale being that hospitals should be preserved from a tsunami of people requesting tests for which there is a severe scarcity (another point, valid from a purely probabilistic point of view, would be that it would reduce the influence of false positive/negative tests). In other countries like Germany or South Korea tests are much more widespread and even opened to people showing no symptoms. Moreover, the definition of a death by COVID19 is not the same across countries (and even across time within a country): how comorbidities are treated ? are post-mortem tests systematics ? etc. For instance in France only deaths registered in hospital are taken into account. Moreover, there is this policy that as soon as 2 COVID19 positive tests are done in a nursing home, they no longer test the rest of the people in this nursing home… I could go on like this for screens. Likewise the slope of cases is barely relevant since the number of tests performed by day may change many fold within a country across time (the extreme example is given by the US that shifted from few hundreds a day to more than 150 000 a day in less than a week).
Pancake thinkers do not even contemplate such remarks. They take published numbers for granted and starts pondering on their signification: how is it that country A has less mortality than country B? Have we reached the plateau of the epidemics? Does this country having a slower rate of increase more efficient than this other country? etc. Those questions are very important, but the way they are dealt with is appalling in the current discourse.
Administrative management of the epidemics
Many strategies have been put in place by the authorities across the globe with substantial differences, though one constant almost universally emerges (with few exceptions like Taiwan, Hong Kong or South Korea) : every authorities have started to downplay the risk of the epidemics, and have adopted a reactive stance to fight it. The general approach has been : life will go on as usual, which is a typically Pancake way of thinking. In its simplest form : a straight line. And it is true that when you look on the left of an exponential curve, it really looks like a flat line. But looking the other way reveals a wall, as the very pedagogical illustration of Luis Batalha shows. The inability of authorities to cope with this simple mathematical reality has produced much delays in the reaction time. Worse, it was clear from the end of January that the virus could not be contained and that a pandemic phase was due at some point in the future. In almost 2 months there were no forecast whatsoever in the west. In the case of France announcements on how to fight against this epidemics have changed every 48 hours.
All those wanderings come from this intuitive, and most of the time wrong, habit of extending curves, using a rule, as straight lines. GDP, inflation, unemployment, stock indexes, forecasts are always (with a consistency approaching the level of physics laws) wrong, this is not big news. The fact that they are still looked at seriously (as basis for fiscal law for once) shows the dark powers of Pancake Thinking.