Friday, April 22, 2016

Hillary Clinton won't be indicted because... math!

One of the questions I keep getting is as follows:  will Hillary Clinton be indicted, and if so, how would that affect the dynamics of a Clinton-Trump campaign?  Is that really two questions?  Yes.  Congratulations you know enough math to follow my answer.

I don't really bother researching accusations against Hillary Clinton in detail anymore.  Why?  Allow me to explain.  With math!

Let us start with some notation.

P(indictable) is the probability that Hillary Clinton is indictable with facts currently publicly available that a lazy observer might not know.

P(accusation) is the probability that Hillary Clinton will be accused of the most heinous crimes in history by her political opponents.

P(X | Y) is the conditional probability that statement X is true given that statement Y is true.  What do I mean by that?  This is about-- warning, buzzword!-- Bayesian statistics!  Bayesian statistics are about uncertainty, and how we update our beliefs as we make relevant observations.  Suppose you don't know whether or not X is true.  Then, your assessment of the probability that X is true is P(X).  Then, you learn that Y is true.  Your assessment of the probability that X is true becomes P(X | Y), which may or may not be the same as P(X).

We've got two of those!  P(indictable | accusation) is the conditional probability that publicly available facts (unknown to the lazy observer) support indictment given that accusations are made.  P(accusation | indictable) is the conditional probability that an accusation will be made given that indictable evidence is available.

Well, actually we have four conditional probabilities.

You see, we also have P(indictable | no accusation) and P(accusation | no indictable evidence).  With that, here's why I ignore stories about HRC's supposed crimes.  With math!

P(accusation | indictable) = P(accusation | no indictable evidence) = 1

Therefore, P(indictable | accusation) = P(indictable).

Ta-da!  OK, let me explain, for those who aren't Bayesian statistics geeks.  HRC's opponents will accuse her of heinous crimes no matter what.  Therefore, the fact that accusations are made provides no information about whether or not indictable evidence exists.  Therefore, my belief that HRC is indictable doesn't change in response to accusations.

Couldn't I have just said "the boy who cries wolf?"  Yes, but then I wouldn't get to talk about Bayes, or use that vertical line thingy on my keyboard, so fuck off and let me have my fun.  It's my damned blog.

So, in historical terms, here's the deal.  HRC's opponents have spent the last 25 years accusing her of everything up to and including murder (Vince Foster), and what has come of a quarter of a century of investigations?  When Ken Starr spent four years and millions of dollars with an open-ended investigation of all things Clinton, he came up with Monica Lewinsky.  Whether or not there is anything indictable has nothing to do with whether or not accusations are made, so ignore the accusations.

Hillary Clinton is sleazy.  She is unethical.  She is a spineless weasel.  Indictable offenses, though?  At some point, paying attention to the accusations becomes a waste of time.  See?  Math is fun!  And it lets you be lazy!

No comments:

Post a Comment