Art Carden  

Best. Comment. Ever.

PRINT
Mike Lee's Anti-Supply-Side Ta... Of Fossil Fuels, Forests, and ...

A quote from reader Tom West's comment on Saturday's post about why bad stories stick:

Deprived of a narrative when given a bunch of facts, humans will use the facts they're given to compose a narrative, and then adjust the facts they've been taught to fit that narrative.

This is, quite frankly, terrifying. I'm working hard to be extra-skeptical of claims I really want to believe. I don't always succeed.



COMMENTS (11 to date)
magilson writes:

This is, at it's core, a synopsis of Kahneman's "Thinking, Fast and Slow".

Troy Camplin writes:

This is absolutely correct, as Darwinian literary theorists have understood for a while now. We make stories out of what we have, and only out of what we have. This is self-corrective in science, where theories' contents are adjusted as new information comes in, but what works for a spontaneous order like science does not work for individual human beings. Human are resistant, and will adjust their interpretation of the facts to fit the theory for as long as possible.

And what magilson said.

William Newman writes:

Humans often do this kind of aggressive mangling of theory and data in a shockingly flaky way that doesn't match reality very well --- sometimes because they place a higher priority on political side effects than on matching reality, sometimes through laziness or careless or arrogance, sometimes for flakier reasons than that.

However, it is worth noting that some powerful automated strategies for learning about reality --- the kind of thing you can read about in machine learning texts like _The Nature of Statistical Learning Theory_ and _The Minimum Description Length Principle_ --- can also behave in a way that rather resembles West's description, even when they are carefully crafted to do a pretty good job of matching reality. (If not "adjust[ing]" individual facts, at least largely discarding individual facts based on their theory derived from the mass of facts.) If this happens naturally without human impulses, and basically works to help match reality (imperfectly, but in machine learning as in other problem domains utopia is not an option) then perhaps not all behavior matching West's description is necessarily dysfunctional.

Machines can be subject to limitations analogous to human laziness and ignorance, but today's machines are largely incapable of political dishonesty and arrogance, and it looks to me as though their behavior is a natural response to the unavoidable limitations and tradeoffs of inductive learning, not just characteristically-human flakiness.

Thomas DeMeo writes:

Like many insightful economic observations of human behavior, we can't take it for an absolute truth. Better to see this as a tendency, and not be so terrified.

Sometimes humans get narratives right, or at least right enough, to improve our understanding of the way the world works in a positive way. And sometimes humans question narratives, and eventually fix them.

Are you the guy writing a book with Deirdre?

NZ writes:

I think that's actually a bit too generous. Firstly because it's rare that people are deprived of a narrative: we grow up hearing narratives, they are a part of our culture, development, and learning process. Secondly because it's hard to construct one's own narrative and I don't think most people would spend their limited mental energy doing it.

Instead, humans first identify their tribe.

Next they accept the narrative popular among that tribe. It's usually the one that's most flattering to that tribe--or in straight white male liberals' case, it's one that makes their tribe seem remorseful and apologetic.

Then they either adjust the facts to fit that narrative, or rebrand the facts as hateful lies.

Some people are very brave and inquisitive, and do scrutinize or at least contemplate their accepted narratives. Some contrarians may even get pleasure from finding parts of their own tribal narratives they disagree with. But these are rare outliers.

MingoV writes:

I don't agree with the comment. My version: When presented with facts without a narrative, people use the facts and narratives they are comfortable with to override the new facts they dislike. The remaining new facts (if any) are distorted enough to fit within their comfortable old narrative.

Julien Couvreur writes:

I'm not sure that this is terrifying. This sounds like probabilistic reasoning to me (Bayesian inference).
All new information is processed in light of prior knowledge/beliefs.

There are cases in science of new evidence invalidating prior beliefs/theory (relativity is such an example).
There are also examples where the prior theory carried too high a confidence, and it turned out the new data was flawed/incomplete (Mars precession led to the possibility of an unknown planet, Neptune, rather than invalidating Newton's mechanics).

That said, it is correct that human common sense is not as accurate as formal Bayesian inference.
Humans make mistakes (bad reasoning), compute approximate estimation of confidence levels (probabilities are not quantified) and fall into biases.

Bedarz Iliaci writes:

humans first identify their tribe.
Next they accept the narrative popular among that tribe

People are born into a tribe. The tribe forms their minds. There are no free individuals waiting to identify their tribe and choosing to accept or reject the tribal narrative.

The individual himself is largely formed by the tribal narrative. Otherwise, you could not say of a person that he is American and not Italian or Chinese.


Humans make mistakes (bad reasoning), compute approximate estimation of confidence levels (probabilities are not quantified) and fall into biases.

Better unquantified probabilities than wrongly quantified. How do you quantify probabilities that occur outside exact sciences at all?

NZ writes:

Bedarz Iliaci:

I didn't say all people got to choose their tribe, necessarily, I just said they identify it. For most people, the choice is always obvious and choosing atypically is met with cries of heresy, so it isn't really a choice at all. It's just a matter of identifying one's tribe on some level.

However, for some people the choice is not always obvious and there are no serious social consequences for making one choice over another. This is the case with most American Jews, I think. Meanwhile, my black mother-in-law supported Hillary in the primary, and was met with static from certain friends and relatives who couldn't believe she would vote against Obama.

[Comment edited by commenter.-Econlib Ed.]

Comments for this entry have been closed
Return to top