Thread by @erezyoeli

In a post last week, @ATabarrok suggests we think of misinformation as an equilibrium outcome. In Ch 8 of Hidden Games, we do. Here's a short rundown.

Warning: for the sake of brevity, I'm using econ jargon.

marginalrevolution.com/marginalrevolu…

@ATabarrok's post focuses on outright lies, and on the role of behavioral biases (e.g., inattention) in permitting them to proliferate.

We focus on a different type of misinformation, that's arguably more common and pernicious: distorted presentation of information

We consider how people will (mis-)use evidence in the service of persuading others.

What evidence will they show? How will others interpret this evidence?

We make 3 assumptions:

1. People are motivated to persuade, not to inform
2. They have private info about which evidence they have, and how they searched for / generated it
3. It is easier to be punished for fabricating evidence than for omitting details about how it's generated

In equilibrium:

- People only show supportive evidence
- People search hard for supportive evidence, but not at all for unsupportive evidence
- People generate evidence that's only putatively supportive evidence, but not very informative

They behave in these biased ways even though everyone knows it.

And people interpret the evidence they show accordingly.

It's a best response for people to behave in these biased ways.

E.g., no point revealing or searching for unsupportive evidence, as it will just reduce others' posteriors. No use withholding supportive evidence, since people will assume you never got it.

#Nashequilibrium

No behavioral assumption in -> very behavioral result out.

(As is often the case in Hidden Games.)

In Ch. 9, we talk about implications for biased beliefs if, as @TriversRobert @BillvonHippel have argued, people internalize those beliefs in ways that help them be more persuasive.