Tuesday, May 30, 2017

The challenge of maintaining skepticism in the Trump era

Every morning, before I type these, I scan the news, and come across stuff like this, saying that US intelligence intercepted communications between the Russians about compromising financial information they had on Trump and his people with which to blackmail him.  Is it true?

For any other presidency, we'd have to discount it as so absurd that it shouldn't really be taken seriously.  And that's today's theme, which gets into conspiracy theories and "fake news."

Did you hear?!   There's a new miracle supplement!  Eat as much as you want, but take this pill, and you'll never gain weight!  Doctors don't want you to know about it because then cardiologists will lose business!!!  Extra exclamation points!!!!!  Anyone with a brain should have a decent bullshit detector for things like medical nonsense.  If it sounds too good to be true, it is.

In politics, the basic problem with a conspiracy theory is that conspiracies themselves are really difficult to maintain.  A conspiracy requires more than one person, by definition.  The more complicated the conspiracy, the more people there are involved, and the more likely it is that someone blows the whistle, either intentionally or unintentionally.  That's why conspiracies don't work.  A chain is only as smart as its dumbest link, or something like that.  So, if someone tells you an outlandish story, you should probably discount it.

Fucking pizzagate...


And yet, trying to apply that rule to Trump?  Here's the thing.  These stories are outlandish, but they break because of stupidity.  These aren't massive, alien-concealing conspiracies.  These aren't generations-spanning, Adam Weishaupt-type deals.

And yet, they are difficult to test.  They are absurd, and for any other president, we would discount them completely.

There are two things going on here.  First, what is our standard of evidence, and second, what are our "priors?"

The standard of evidence, from a legal perspective, never changes, but I don't go at things from a lawyer's perspective.  I'm a social scientist.  I shave with my good buddy, Occam.  From a social-scientific perspective, I am fine with loads and loads and loads of indirect evidence as long as there is nothing exculpatory, and right now, the best Trump has is the missile he launched at Syria.  I've already explained by thinking on that.  The point, though, is that there is a difference between social-scientific reasoning and legal reasoning in terms of how we evaluate evidence.  Would I argue for a conviction on the basis of indirect and circumstantial evidence?  No, but as far as I'm concerned, he's already admitted to impeachable offenses anyway (obstruction of justice), so it doesn't matter.  At the end of the day, what I wrote months ago was this:  all Russia needed to have blackmail material on Trump was bribe someone in some office somewhere to get a copy of one of his tax returns.  Now, this story breaks, saying that Russian officials were talking to each other about compromising financial information.  Hmmm...

Then, there's that "priors" question.  I keep coming back to this.  In Bayesian statistics, probabilities are all about the quality of information we possess.  We start with a "prior," which is our initial estimate of the likelihood of something being true or an event occurring, and update that estimate as we encounter new information.  For any normal, sane, competent politician, our "prior" that the politician is controlled or controllable by Russia should be so low that updating that probability upward should require extraordinarily strong information.

The problem for Trump is that, as usual, he's not a normal politician.  He is stupid, corrupt and incompetent.  So, he starts with a higher probability for that "prior."  Thus, it doesn't take as much to move that prior a little bit.

To understand, here's a good one from Randall Munroe.  The basic point is that the Bayesian starts with such a low prior that the sun has gone nova that the device is irrelevant.  If the Bayesian were more uncertain, the device would matter.  When it comes to Trump, we are more uncertain.  It is harder to remain skeptical, even of absurd stories, because he is so obviously stupid and corruptible.

Combine that with the fact that these stories aren't of the kind that involve massive conspiracies where too many people are remaining silent, and yeah, it just gets difficult to keep the kind of skepticism that I keep about medical nonsense and other elaborate conspiracies.

This is our world now.

No comments:

Post a Comment