Antifragile: Things That Gain from Disorder (67 page)
Read Antifragile: Things That Gain from Disorder Online
Authors: Nassim Nicholas Taleb
Alas, all these biases lead to action, almost never inaction.
In addition, we now know that the craze against fats and the “fat free” slogans result from an elementary mistake in interpreting the results of a regression: when two variables are jointly responsible for an effect (here, carbohydrates and fat), sometimes one of them shows sole responsibility. Many fell into the error of attributing problems under joint consumption of fat and carbohydrates to fat rather than carbohydrates. Further, the great statistician and debunker of statistical misinterpretation David Freedman showed (very convincingly) with a coauthor that the link everyone is obsessing about between salt and blood pressure has no statistical basis. It may exist for some hypertensive people, but it is more likely the exception than the rule.
For those of us who laugh at the charlatanism hidden behind fictional mathematics in social science, one may wonder why this did not happen to medicine.
And indeed the cemetery of bad ideas (and hidden ideas) shows that mathematics fooled us there. There have been many forgotten attempts to mathematize medicine. There was a period during which medicine derived its explanatory models from the physical sciences. Giovanni Borelli, in
De motu animalium,
compared the body to a machine consisting of animal levers—hence we could apply the rules of linear physics.
Let me repeat: I am not against rationalized learned discourse, provided it is not fragile to error; I am first and last a decision maker hybrid and will never separate the philosopher-probabilist from the decision maker, so I am that joint person all the time, in the morning when I drink the ancient liquid called coffee, at noon when I eat with my friends, and at night when I go to bed clutching a book. What I am against is
naive
rationalized, pseudolearned discourse, with green lumber problems—one that focuses solely on the known and
ignores the unknown
. Nor am I against the use of mathematics when it comes to gauging the importance of the unknown—this is the robust application of mathematics. Actually
the arguments in this chapter and the next are all based on the mathematics of probability—but it is not a rationalistic use of mathematics and much of it allows the detection of blatant inconsistencies between statements about severity of disease and intensity of treatment. On the other hand, the use of mathematics in social science is like interventionism. Those who practice it professionally tend to use it everywhere except where it can be useful.
The only condition for such brand of more sophisticated rationalism: to believe and act as if one does not have the full story—to be sophisticated you need to accept that you are not so.
This chapter has introduced the idea of convexity effects and burden of evidence into medicine and into the assessment of risk of iatrogenics. Next, let us look at more applications of convexity effects and discuss
via negativa
as a rigorous approach to life.
1
A technical comment. This is a straightforward result of convexity effects on the probability distribution of outcomes. By the “inverse barbell effect,” when the gains are small to iatrogenics, uncertainty harms the situation. But by the “barbell effect,” when the gains are large in relation to potential side effects, uncertainty tends to be helpful. An explanation with ample graphs is provided in the Appendix.
2
In other words, the response for, say, 50 percent of a certain dose during one period, followed by 150 percent of the dose in a subsequent period in convex cases, is superior to 100 percent of the dose in both periods. We do not need much empiricism to estimate the convexity bias: by theorem, such bias is a necessary result of convexity.
3
Stuart McGill, an evidence-based scientist who specializes in back conditions, describes the self-healing process as follows: the sciatic nerve, when trapped in too narrow a cavity, causing the common back problem that is thought (by doctors) to be curable only by (lucrative) surgery, produces acid substances that cut through the bone and, over time, carves itself a larger passage. The body does a better job than surgeons.
4
The core point in this chapter and the next is nonlinearity as it links to fragility, and how to make use of it in medical decision making, not specific medical treatments and errors. These examples are just illustrative of things we look at without considering concave responses.
5
A common mistake is to argue that the human body is not perfectly adapted, as if the point had consequences for decision making. This is not the point here; the idea is that nature is computationally more able than humans (and has proven to be so), not that it is perfect. Just look at it as the master of high-dimensional trial and error.
Wednesdays and Fridays, plus Lent—How to live forever, according to Nietzsche or others—Or why, when you think about it, not to live longer
Whenever you question some aspects of medicine—or unconditional technological “progress”—you are invariably and promptly provided the sophistry that “we tend to live longer” than past generations. Note that some make the even sillier argument that a propensity to natural things implies favoring a return to a day of “brutish and short” lives, not realizing it is the exact same argument as saying that eating fresh, noncanned foods implies rejecting civilization, the rule of law, and humanism. So there are a lot of nuances in this life expectancy argument.
Life expectancy has increased (conditional on no nuclear war) because of the combination of many factors: sanitation, penicillin, a drop in crime, life-saving surgery, and of course,
some
medical practitioners operating in severe life-threatening situations. If we live longer, it is thanks to medicine’s benefits in cases that are lethal, in which the condition is severe—hence low iatrogenics, as we saw, the convex cases. So it is a serious error to infer that if we live longer because of medicine, that all medical treatments make us live longer.
Further, to account for the effect of “progress,” we need to deduct of course, from the gains in medical treatment, the costs of the diseases of
civilization (primitive societies are largely free of cardiovascular disease, cancer, dental cavities, economic theories, lounge music, and other modern ailments); advances in lung cancer treatment need to be offset by the effect of smoking. From the research papers, one can estimate that medical practice may have contributed a small number of years to the increase, but again, this depends greatly on the gravity of the disease (cancer doctors certainly provide a positive contribution in advanced—and curable—cases, while interventionistic personal doctors, patently, provide a negative one). We need to take into account the unfortunate fact that iatrogenics, hence medicine, reduces life expectancy in a set—and easy to map—number of cases, the concave ones. We have a few pieces of data from the small number of hospital strikes during which only a small number of operations are conducted (for the most urgent cases), and elective surgery is postponed. Depending on whose side in the debate you join, life expectancy either increases in these cases or, at the least, does not seem to drop. Further, which is significant, many of the elective surgeries are subsequently canceled upon the return to normalcy—evidence of the denigration of Mother Nature’s work by
some
doctors.
Another fooled-by-randomness-style mistake is to think that because life expectancy at birth used to be thirty until the last century, that people lived
just
thirty years. The distribution was massively skewed, with the bulk of the deaths coming from birth and childhood mortality. Conditional life expectancy was high—just consider that ancestral men tended to die of trauma.
1
Perhaps legal enforcement contributed more than doctors to the increase in length of life—so the gains in life expectancy are more societal than from the result of scientific advance.
As a case study, consider mammograms. It has been shown that administering them to women over forty on an annual basis does not lead to an increase in life expectancy (at best; it could even lead to a decrease). While female mortality from breast cancer decreases for the cohort subjected to mammograms, the death
from other causes
increases markedly. We can spot here simple measurable iatrogenics. The doctor, seeing the tumor, cannot avoid doing something harmful, like surgery followed by radiation, chemotherapy, or both—that is, more harmful than the tumor. There is a break-even point that is easily crossed by panicked doctors and patients: treating
the tumor that will not kill you
shortens your life—chemotherapy is toxic. We have built up so much paranoia against cancer, looking at the chain backward, an error of logic called
affirming the consequent
. If all of those dying prematurely from cancer had a malignant tumor, that does not mean that all malignant tumors lead to death from cancer. Most equally intelligent persons do not infer from the fact that all Cretans are liars that all liars are Cretan, or from the condition that all bankers are corrupt that all corrupt people are bankers. Only in extreme cases does nature allow us to make such violations of logic (called
modus ponens
) in order to help us survive. Overreaction is beneficial in an ancestral environment.
2
Misunderstanding of the problems with mammograms has led to overreactions on the part of politicians (another reason to have a society immune from the stupidity of lawmakers by decentralization of important decisions). One politician of the primitive kind, Hillary Clinton, went so far as to claim that critics of the usefulness of mammograms were killing women.
We can generalize the mammogram problem to unconditional laboratory tests, finding deviations from the norm, and acting to “cure” them.
Now I speculate the following, having looked closely at data with my friend Spyros Makridakis, a statistician and decision scientist who we introduced a few chapters ago as the first to find flaws in statistical forecasting methods. We estimated that cutting medical expenditures by a certain amount (while limiting the cuts to elective surgeries and treatments) would extend people’s lives in most rich countries, especially the United States. Why? Simple basic convexity analysis; a simple examination
of conditional iatrogenics: the error of treating the mildly ill puts them in a concave position. And it looks as if we know very well how to do this. Just raise the hurdle of medical intervention in favor of cases that are most severe, for which the iatrogenics effect is very small. It may even be better to increase expenditures on these and reduce the one on elective ones.
In other words, reason backward, starting from the iatrogenics to the cure, rather than the other way around. Whenever possible, replace the doctor with human antifragility. But otherwise don’t be shy with aggressive treatments.
Another application of
via negativa:
spend less, live longer is a subtractive strategy. We saw that iatrogenics comes from the intervention bias,
via positiva,
the propensity to want to
do something,
causing all the problems we’ve discussed. But let’s do some
via negativa
here: removing things can be quite a potent (and, empirically, a more rigorous) action.
Why? Subtraction of a substance not seasoned by our evolutionary history reduces the possibility of Black Swans while leaving one open to improvements. Should the improvements occur, we can be pretty comfortable that they are as free of unseen side effects as one can get.