Fact-Checking is Table Stakes

Facts and figures will never beat stories on persuasiveness. Bullshit will always be cheaper than reality. That said… (⚠ These claims are disputed.)

Since I am neither a citizen nor resident of the United States, I feel like I don’t have much of a place commenting on the events of the first week of January 2021, except perhaps one degree removed.

2020 was the year that cemented motivated reasoning as my dominant mental model for how people behave. This is not the OODA-loop, utility-maximizing rational actor operating, but something that looks a lot more like:

  1. You want to do a thing (it doesn’t matter why),

  2. You angle for a narrative to justify the thing you want to do,

  3. You do the thing, before a counternarrative that overturns your justification can emerge,

  4. If such a counternarrative emerges, it doesn’t matter, because the thing you wanted to do is a fait accompli.

I am using the terms narrative and counternarrative on purpose, rather than evidence, because evidence is optional. We are dealing exclusively in the domain of rhetoric, for the purpose providing political cover for our predilections, as well as garnering support from fellow travelers.

The seed for my thinking in this direction is a kinda trippy riff on embodied cognition advanced in the 1970s by Humberto Maturana and Francisco Varela called (later on by Eleanor Rosch) enactivism, whereby living organisms “imagine” their future selves (for a value of the term scaled to the organism) and then subsequently move into those imagined futures—a fancy, hard-to-pronounce term called autopoiesis. And, well, the way you do collective imagining is language.

Evidence is optional, because for the purpose of this kind of justifying narrative, bullshit is more than satisfactory: it is ideal.

Bullshit, as the philosopher Harry Frankfurt so aptly characterized, is a claim or assertion for which its proponent is indifferent to whether or not it is true. This is importantly distinct from a lie, wherein the liar knows the truth but doesn’t want you to know it. Bullshit has no such constraint.

An Econophysics of Bullshit

Not being constrained by evidence—or even reasoning (which I will return to momentarily)—makes bullshit the cheapest kind of speech to manufacture. Even a lie has to be manicured so as to evade suspicion. Alberto Brandolini’s instinct was accurate when he wrote down his eponymous “law”:

“The amount of energy needed to refute bullshit is an order of magnitude larger than to produce it.”

Bullshit is refuted with evidence and/or reasoning, and both are irreducibly costly. Evidence, because you have to actually go out in the field and get it. Reasoning, because it requires everybody in the audience to think out the reasoning process themselves—real cognitive effort you simply can’t force them to do. Bullshit will always be cheaper, and very possibly, as Brandolini suggests, by order of magnitude.

Narrative as Collective Orienting Mechanism

Assuming we agree that bullshit narrative will be a cheaper motivating mechanism than one grounded in evidence and reason, why is a narrative necessary at all? Why doesn’t an argument suffice—even a bullshit one?

I suspect it’s because stories—at least interesting ones—always have a moral valence. The elementary unit of a narrative looks something like:

  • Person (or at least agent) does something (or decisively nothing),

  • The outcome of this decision is good (or bad),

  • For this (in)action, Person should be praised (or blamed),

  • We should therefore like (and/or be like) Person (or not).

Irrespective of the content of this structure, there is always something to be felt. You can construct epics out of this basic narrative atom, of which there are innumerable examples. A natural extension from here is a trope that goes something like “Person has committed an injustice and We Must Get Them™.” Which, fair, is an age-old societal regulatory mechanism, but there is enough historical record, to say nothing of famous myths and legends, that this kind of response is often enough predicated on bullshit.

Stories short-circuit our analytical faculties and hit us straight in the feels. The trope of an injustice that demands a corrective (or revenge, it is not clear to me that there is a meaningful distinction in this frame) is a psychic irritant similar to the utterance of a swear word, which we can think of as the same thing at a smaller scale. Stories like these, for lack of a better term, are triggers.

Reasoning Costs

The Roman statesman Cicero (as well as Aristotle and others before him) also remarked on the inferiority of reason as a tool of persuasion, the basis of which being, roughly, that reason only works on an audience who is already sympathetic, or at least receptive. The paper Why Do Humans Reason?✱ from about a decade ago by Hugo Mercier and Dan Sperber, suggests this is precisely the ecological niche reason evolved to fill: Reason is for swaying people who are neutral to, or already have doubts about the opposing story. Logic, math, and the scientific method are only fortuitous side effects.

✱ They could have called the paper An Argument For Argument as the Reason for Reason; such a lost opportunity.

To speculate on why this tantalizing proposition might be, let’s consider that following a line of reasoning means holding your undivided attention all the way to the end. Vacillating over whether to take one position or other with respect to an issue of potential social consequence to you when you haven’t already made up your mind is similar, except it never terminates, so in this situation reasoning is preferable. If, however, you have already made up your mind, why spend your own cognitive resources to potentially introduce doubt?

War Is Organized Theft

Over the last few years I have kept going back to a particular clip from Bronowski’s Ascent of Man on the origin of war, which he characterizes as “a highly planned and cooperative form of theft”, that…

“…began 10,000 years ago, when the harvesters of wheat accumulated a surplus, and the nomads rose out of the desert to rob them of what they themselves could not provide.”

A particularly dangerous combination of motivated reasoning is a motivation to plunder justified by a narrative that is bullshit. It is dangerous because there is no graceful off-ramp: once you start fighting, your only rational way forward is to try to win.

This situation is doubly dangerous when these people are allowed to live in their bullshit fantasy world for a protracted period with no social consequences.

Well, Actually…

We should be able to tell by now that fact-checking motivated reasoning is utterly impotent as a rhetorical strategy. Fact-checking is bringing evidence to a narrative fight. People who hold ridiculous beliefs to justify their invidious tendencies need to be mercilessly shamed and ridiculed precisely because their positions are disingenuous. They need to feel pain: lose their jobs, friendships and social connections, book deals. You have to go after these people like they’re eyeing your stuff and sizing you up for a fight, because on some level, that is what they are doing.

It’s like that Gandhi quote with the roles reversed: if you ignore them too long and don’t laugh at them hard enough, you will have to fight them, and losing would be Bad™.

Bless Daniel Dale’s heart for his tireless effort, but his role, and that of people like him, is purely janitorial.

Fact-Checking is Table Stakes

This is to say that I do believe there is utility in mopping up the Augean stable of bullshit, it’s just not what we think it is. (It is emphatically not to smugly wellactualize maga-chuds with factual gotchas.)

It shouldn’t be controversial to write that accurate information about the world is good, and having the means to know when information is bad is also good. I submit that the value of a fact-check amounts to “yes, somebody trustworthy has looked at this.” At the very least it informs you where to direct your ridicule.

The problem with fact checking, in my opinion, is that it’s too retail. It needs to scale: Snopes and journalists on Twitter are not efficient enough. Moreover, while any given unit of bullshit is cheaper than a unit of truth, it has to be repeated in order to spread. This can be used to the advantage of the proponents of reality, because you only need to check a given claim once.

What also needs to happen is the separation of the fact-check data itself from its presentation. Make a fact-check protocol. Heck, I’m even willing to say with a perfectly straight face that a system like that would be one hundred percent appropriate to design around a blockchain. The wellactuallygram annotations on social media, on the other hand, are not especially good. In the spirit of mocking the target, perhaps instead set the bullshit claim in purple Comic Sans, à la PZ Myers. Maybe do a daily roundup of the dumbest takes for public pillorying (but not like in a smug John Oliver way). Make ’em the main character of Twitter for a day.

How would I implement a fact-check protocol? Spitballing:

  • A copy of the document or resource containing the claim (we should really be archiving these anyway),

  • The address(es) where said document was retrieved✱,

  • Some kind of formal representation of the claim (TBD lol)

  • Some kind of thumbs-up/down status indicator,

  • At least one time stamp (e.g. when the check was logged),

  • An identifier that picks out the fact-checker,

  • All wrapped up in a (public-key) cryptographic signature,

  • …and published on some kind of distributed ledger, mainly so there are many copies that no single entity can control.

✱ Tweets and analogous social media posts and comments tend to be directly addressable. There are likewise ways to address content directly but it would likely have to be accompanied by some kind of standard normalization function to be really useful. This is a topic for another time.

Obviously there will be the usual problems with a scheme like this. Partisan operators will call bullshit on each others’ claims. Fact-checkers will get hacked (or defect!). The bullshit-peddlers will probably try to arms-race or otherwise game the system somehow. Metacrap—bullshit in the system itself—is definitely a hazard. Lots of details need to be figured out for sure.

This Is Called Structured Argumentation

I started by talking about how as artifacts of persuasion, stories beat arguments. I’m going to close by picking on news outlets for being too wedded to the discrete story or article as their basic unit of work product.

  • When you don’t need persuading, just the facts will do. It would be very interesting to publish the specifics in a structured form—the who/what/where/when/how/why—alongside, or embedded within conventional news articles, so their contents can be operated over programmatically. This would be the “formal representation” I handwaved above.

  • Actual stories tend to be much bigger than just one “story”. New information is just a delta against what is already known. It would be useful to be able to zoom out and see the progression of events over time—especially across outlets—punctuated by news articles.

News outlets do publish metadata on their articles, but it’s mainly for search engines and social media, and not very useful beyond that. Likewise, they also occasionally do longitudinal multimedia presentations, but these are big expensive one-off undertakings. I’m talking about a permanent fixture.

Part of policing bullshit—and protecting yourself against it—entails knowing what the real situation is. I am deeply interested in developing ways to make the consumption of situational information much, much more efficient. There is plenty of tech just lying around to help with this, enough to hopefully put a dent in the bullshit asymmetry gap.