I’m reading Taleb’s notebook and thinking about parallelism. He talks about Russian statisticians who inject doubt into predictions, trading (which I know not, nor care about) and most importantly, skin in the game.

He says beware of the mandarins, the grand ecole, who produce administrators who have no skin in the game; not risk, and think they can regulate from academic models. This would include Piketty, who he tartly sums up by reminding us that correlation does not mean causation.

And the parallel I’m considering is being a clinician who works with evidence based statisticians. Someone who writes guidelines, reviews, does clinical trials… and still works in a clinical setting. For most people who I see clinically would never get into a trial. They have too many co-morbidities. They mistrust the system, at times with good cause.

Those who do get into any clinical trial are not the same as the people who end up in a doctor’s consulting room. nor in an acute ward against their will. The population: in technical terms, the sampling frame, is different. And what you conclude in the trial may or may not apply in the clinic.

This applies across medicine, but the issues are greatest in psychiatry which is both the end-point of all those situations society cannot handle, and the one part of medicine where clinical skill has no tests to guide it. (when an illness, such as tertiary neurosyphilis, has a cause and a cure some other group take it from psychiatrists, who gladly gave their general paresis of the insane).

The outsider thinks the psychiatric clinic is inefficient, irrational, and should be easily rationalized. It should fit into their model of the world, be it libertarian (who closed all the US hospitals) or post modern intersectionality (who closed all the European Hospitals). But they forget a few things.

• Our models are not yet explanatory.
• The art of understanding despair and madness is difficult and long
• All our treatments work only in part and have a cost. That includes talking therapy, which does have adverse effects.

And we are dealing with suffering people, and people are complex. Not widgets.

And not amenable to a time and motion study.

Which is why clinicians mistrust a manager who sees not patients. Who has given up clinical work. They have no skin in the game.