The Strengths and Weaknesses of Social Predictive Heuristics

A while back there was a writer known in communities I frequent (call her Jane) who wrote a think-piece explaining she was very, very upset about an issue. It was, by Jane’s telling, something like the great national sin of our age; it was a human rights violation of epic proportions. This was during the Trump presidency, when ostensibly genocide-equivalent human rights violations were a dime a dozen and dominated every moment of every cable news broadcast.

Several more right-leaning people called bullshit; you aren’t actually upset about this, they said. Their claim was that Jane didn’t like the particular president and it was motivating her to exaggerate her feelings, sort of the “Anti-Trump Hysteria” accusation as applied to a single view. The sentiment was that she wouldn’t care about it if the current president was on the left, kind of like how if you are on the left you might draw attention to the fact that nobody on the right ever asked for Trump’s birth certificate.

She immediately wrote a response article explaining that she was in fact incredibly and sincerely upset; she could not eat or sleep, walked around in a daze because of this issue and cried until her pillow was soaked every night. Since it’s not possible to prove someone’s internal emotional state beyond a reasonable doubt, normally this would be the end of it; without video evidence proving she was mostly living her life as normal besides the one article saying she wasn’t, it would be something of a stalemate.

Then someone did the raddest thing I’ve seen anyone do in years.

Basically, a guy did what everyone else did in guessing that she’d stop talking about the issue as soon as it wasn’t politically convenient. But he took it one step further: he held onto the prediction for months and months until the office of president changed hands. He then noted (very publicly) that while she’d written about a great many things during the new administration, she conspicuously no longer thought the previously-maximally-bad subject made the write-worthy-subjects cut.

This of course still didn’t solve anything absolutely; there was a big argument about what it meant. But the power of a documented prediction of one side (that she’d stop talking about it, since it wasn’t important) coming true while the assumed prediction of the other side (that since her sadness was apolitical and her sorrow was so intense, it would continue to be visibly so outside of a narrow partisan argument) failed to materialize was significant enough that it generated an instant controversy. Jane still had people willing to defend her, but they were visibly on their back foot throughout the conversation.

Jane eventually had to show up and explain that she often was very, very upset about things that more or less faded into the background of her consciousness a few months later, and that this was normal for her. It seemed that she would still have preferred people to take her argument seriously, but the power of the opposing side’s prediction was such that she had to give up a ton of ground to stem the reputational bleeding.

Note that this wasn’t an out-and-out drubbing or an absolute victory in the nothing-left-to-gain sense. Jane lost a lot of credibility, but most in the crowd were reluctant to say she was dishonest in any way. While she had to enter the dialogue to defend herself, showing up to the argument won her a lot of goodwill with a lot of the people present.

This story is a bit vague, and purposefully so; the forum in question doesn’t really want outside attention and Jane has probably already taken an appropriate amount of flak. The reason I’m sharing the narrative is more about the technique than parsing who was right or wrong; despite not quite getting to salted-fields levels of devastation, the make-a-prediction-and-wait tactic had ten times the effect the initial argument had and was as close to a trouncing as I’ve seen.

---

Quick: Tell me exactly what a white supremacist is.

Even quicker: Tell me precisely what a bleeding-heart liberal is.

Now: What is “woke”? What qualifies as CRT? What’s a “MAGA guy”?

I’m not saying you don’t have a definition of some kind for most of these; you probably do. Somewhere in your heart, each of those phrases means something fairly precise to you even if you haven’t thought about it a lot. And that definition is actually probably useful to you for internal use; it describes a group of people, even if that definition isn’t more complex than “the bad group I hate”.

But the private meanings labels carry aren’t usually enforceable once they leave your head; I wasn’t the only person to write an article about the disaster the Left brought on themselves by trying to tell people theirs was the only acceptable definition of CRT, for instance. Trying to call someone who has been caught multiple times saying factually untrue things a liar around someone who has reason to defend them (even if the reason is just general niceness) usually ends with the discovery that an exact definition is surprisingly hard to get others to agree on.

Is someone who says something untrue while they are upset lying, or are they just not thinking? Is someone who confidently claims something to be true having not done basic research that would show it false being dishonest, or just a little weak on their how-to-know-things fundamentals?

---

You only need to have a few conversations about what is (or isn’t) being taught in schools or about whether someone who regularly makes factually false statements should be thought of as a liar to know how incredibly easy it is to get mired in the semantics of things. You might argue for hours about the correct thing to call someone who uses “BIPOC” in casual conversation even while agreeing on everything you’d expect that person to do or believe.

Framing your statements about a person as a prediction avoids a lot of these problems. At least part of what you are trying to do when you call someone a liar is to relate to your audience (or yourself) an expectation; this is a person who will tell you things that aren’t true. If you say just that and avoid the word liar entirely, you get to circumvent all the discussion about where to draw the threshold of when that label becomes appropriate (if you should even call someone that at all).

¹

Once in a great while you see a person do something that defies your expectations of hypocrisy; maybe someone crosses partisan lines to stay consistent on an issue, or someone calls out intellectual dishonesty even where it weakens their side. It’s usually a good thing when this happens. We like it because it strips away a lot of uncertainty what to expect from them and lets us have more trust in their positions no matter what their political alignment might be. If what they said was important to them actually was, we will see by their actions; if they do anything else, we know what their actual principles (rather than those they claimed) are.

The idea of the predictive heuristic here is something like that in reverse; as opposed to making decisions based on principles in defiance of labels, it strips away the baggage of labels we might use to try to identify the principles which actually guide their behavior. At the same time, it forces us to make actual predictions of what we think a person’s actual behavior will be in a way that’s far more agnostic of value judgments we might make about them. At a basic level, we are making a falsifiable statement of what we think their principles are in the barest, most stripped-down form we can.

I’ve been presenting this as a tool used in communication, but I actually like it even better as a medication prescribed for internal use. It’s tempting to think of someone as a Democrat or Republican, full stop, and to hate them based on whatever kinds of negatives you’ve dyed that term in. It’s work to actually force yourself to think enough and know enough about an individual to develop a model that accurately predicts what they will do in a given situation, but it’s also better on (almost) all counts. It’s more forgiving where your blanket judgments would have condemned some poor soul and more damning where your cold, dry predictions of misbehavior (or good behavior, in the case of someone on your team) are proved untrue.

---

While the predictive heuristic brings a lot of advantages to the table, in terms of simplification and productive focus, it’s not without some significant disadvantages. To the extent the transition from labels brings downsides, the lost utility is mostly related to norm enforcement.

There’s a classic SSC article called The Whole City Is Center which deals with some of these same themes. It’s mostly a discussion between two fictional people, one of which believes that the term lazy carries a lot of unnecessary negative baggage, and one of whom points out that it accurately describes a person who can’t be counted on to do things they should.

You should read the article, but the relevant upshot is that the article makes the assumption that at least some people who are termed “lazy” only act that way because of a basket of psychological things they can’t control. It’s not necessarily involuntary in the sense that they’d still be lazy if someone had a gun to their head that fired at any instance of shirking, but in the article’s telling is still powerful enough and common enough that considering someone shitty for letting a dog they were sitting starve to death is misguided moralism.

It’s on this that the author builds the argument that the term shouldn’t be used for any of them. More importantly (and much more subtly), he’s asserting that the term “lazy” carrying any negative connotation at all is an accepted negative - we shouldn’t be looking to make people feel bad, no matter what else we might do.

²

But even in an extreme version of “this person didn’t have any choice, people are just bundles of influences” type-thinking, we can’t fully escape that how one is viewed by others is included in the bundle and is often one of the bigger influences found there. Liar is a pretty rough term; people don’t like being called that, and they especially don’t like being thought of that way.

Scott thinks it’s very unfair to call someone lazy because he thinks it’s mostly Not Their Fault and that the word itself carries a bunch of negative baggage that isn’t appropriate if laziness is really a bundle of responses to incentives and psychological hardwiring. Scott-style anti-label folks are absolutely right that there’s a danger of labels becoming mere slurs; the issue isn’t that this is a danger, but instead moving past any discussion of the normative function those labels can serve.

There’s no obvious way to acknowledge that there’s a significant negative impact to labelling someone with a term like lazy or liar and still claim it wouldn’t be a part of the decision-making process of a theoretical person, even if they are completely controlled by their baseline psychology and external influences. Unless you discount incentives entirely, “avoiding the negative label” is an incentive that pretty much everyone in the argument seems to agree is pretty powerful.

There’s a segment of our culture where being unracist, unhomophobic and untransphobic are terminal values; to the people who rely on them for self-worth, they are the very most important things. In response to those norms being popularized, everything from media to social media all the way to presidential politics are broadly affected. There’s a whole Twitter army working overtime to burn heretics at the stake all to avoid being thought of as bigot, racist, or homophobe in favor of whatever substitute for woke pops off the euphemistic treadmill next. Whether that norm is right or wrong, it’s powerful to the extent it’s transformed how our entire society functions.

Now imagine a world where the word liar is banned, no substitute labels are developed, and everyone is forced to say describe things using predictive heuristics. We can still point out that Jane can’t be trusted to relay accurate information on hot-button issues of the specific type she proved untrustworthy on here, but not letting that leak into judgments about her character or the generalized judgments we’d apply to her honesty.

In the no-label world, we lose all the impact we’d get from Jane’s concern with how she’s viewed as a person, keeping only the effect we’d expect if she cares about us thinking she’s trustworthy on the specific issue at hand. She might be known as untrustworthy on some particular subjects but would be safe from any more global judgments of what that untrustworthiness means about her as a person. Agnostic of everything else, we’d expect Jane (and everyone) to lie more in general in response to opinions on their character being insulated from any particular misbehavior coming to light.

The Whole City Is Center view of things either rejects or ignores the idea that, say, having a society where there was a strong norm to view liars negatively and where lying was thought to have strong implications on someone’s general character might result in less lying overall. But in a world where norms do matter, where a word like “racist” is so powerfully negative that a part-time army works on getting supposed racists fired or ostracized to distance themselves from it, you give up something by giving up the power to label a person rather than a specific behavior.

---

What good is it, my brothers, if someone says he has faith but does not have works? Can that faith save him? If a brother or sister is poorly clothed and lacking in daily food, and one of you says to them, “Go in peace, be warmed and filled,” without giving them the things needed for the body, what good is that? So also faith by itself, if it does not have works, is dead.

But someone will say, “You have faith and I have works.” Show me your faith apart from your works, and I will show you my faith by my works. You believe that God is one; you do well. Even the demons believe—and shudder!

I like the passage above because it’s James desperately trying to disambiguate a label; he’s working as hard as he can to make sure people reading this immediately and across subsequent centuries don’t have an excuse to mistake the baggage they attach to a particular label to the actual truth of the matter.

James’ logic as stated here says that you should be able to predict your own actions if you have the kind of faith the Bible cares about; if you do, you will go around and do good stuff. If you say you have faith but don’t do any good stuff, James says you and he aren’t talking about the same kind of thing; you’ve let the label get in the way of the definition.

For better or worse we live in a post-agreed-upon-definition world. I’d like it if we all agreed on what a liar is (or that it existed as a meaningful term in the first place) and applied that definition consistently within and outside of our tribal affiliations, but I think we are far past that; I’m going to push for it, obviously, but I’m not confident at all that I’ll get what I want. With a few exceptions, labels are broadly useless now; to the extent they work at all, it’s a slur used between members of an ingroup to help hate an outgroup member better.

Despite the norm-enforcing value of labels being broadly lost, I think we still get benefits from transferring their function to predictions. For a bipartisan example, it may not be useful to call your out-group party hypocrites; even if labels did still work, it’s too easy to brush off as a slur. But waking up from a coma not knowing what year it is, you can make the almost-for-sure prediction that the minority party is pro-filibuster and the majority party considers it a stain on Democracy without even knowing which party occupies which seat.

---

Lest you get away without some moralizing: I’ve talked about all this so far as either a communications tool or a better way of thinking about others, but if any of this has rung true for you then you can’t escape what it means for yourself. It’s easy for someone like me to apply the label good husband and then not do a few loads of dishes for my wife so she has less to worry about coming home from a long trip. If I want that label to mean anything, I need to tie it to actual predicted behavior that justifies the pride I take in it.

If labels ever had a normative value where people would avoid bad behavior to protect their greater reputations, we’ve at least mostly lost that by now. But by tying predictive models of behavior to how we talk about and assess people and ourselves, we can get back some of what we’ve lost. We might not have labels, but we can still have standards.

(Special thanks to Nick for proofreading/editing efforts on this article).

---

If you are reading this, there’s a good chance you are a newsletter type of person. If so, I recommend The Sample; they send you a random newsletter a day from a curated list; if you like a particular writer, it’s one click to subscribe from your inbox.

Full disclosure: I get traffic from sending them traffic; it’s a barter thing. But with that said, if you want “good random” newsletters sent to you for finding-new-things purposes, it’s the best service I’ve tried for that purpose.

1. In the interest of communicating useful information, most people don’t use “liar” to mean anyone who has lied at least once, even if it’s arguably technically correct to do so. But the numerical threshold of when someone has lied enough that the label starts being appropriate (if it ever is) is unclear.

One qualifier some people use has to do not with the amount of lies, but the perceived size/impact of the lies; white lies are accepted where “bigger” lies wouldn’t be. When you apply that logic to other unusual unacceptable behaviors outside of lying, you get the kind of system of moral assessments that produces jokes like this:

Dave is visiting a small Irish village on vacation, but notices there’s one man who is somewhat of an outcast. The man doesn’t talk to other people in the village where he can help it, and the residents of the village point, laugh and sneer at him when they see him. Dave is curious as to why, and eventually tracks the man down to ask him about it.

“Do you see the thatched roofs of the village? I thatched each one.” Says the man. “Hundreds of them! But do they call me Joseph the Roof-thatcher? No. See the fine cobblestone streets? I laid thousands of stones myself to make them. Do they call me Joseph the road-builder? No. But you screw one goat….”

2. In the article, both characters work under the assumption that calling someone “lazy” is always pretty negative, in that it carries with it a judgment. Scott doesn’t explicitly state it, but he’s the kind of compulsive-gentle person that doesn’t generally believe that people can do negative things on purpose. So he says stuff like this:

There’s a popular mental health mantra that “there’s no such thing as laziness” (here are ten different articles with approximately that title: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10). They all make the same basically good point. We shame people who don’t work very hard as “lazy”, and think they should have lower status than the rest of us. But actually, these people don’t just randomly choose not to work. Some of them have psychological issues, like anxiety and trauma that are constantly distracting them from their work, or a fear of success, or self-defeating beliefs about how nothing they do matters anyway. Others have biological issues – maybe hypothyroidism, or vitamin deficiencies, or ADHD, or other things we don’t understand that lower their energy and motivation. Still others just don’t want to do the specific thing we are asking them to do right now and can’t force themselves to work uphill against that gradient. When we call people “lazy”, we’re ignorantly dismissing all these possibilities in favor of a moralistic judgment.

The set-up is something like pointing out that you wouldn’t call someone with joint problems that cause them pain “race-averse” in a sneering way; they are to some degree disabled and working from a different baseline than you. Scott thinks (or seems to think) most “lazy” people are like this; they to a great degree aren’t in control of their behavior. Once the truth of “laziness isn’t anyone’s fault, nobody has choices” is established as relevant to this case, his characters both use it as a baseline assumption:

Simplicio: I think we’re treating the word “laziness” differently. I’m thinking of “lazy” as a way to communicate a true fact about the world. You agree that the true fact should be communicated by some word, but you’re interpreting “lazy” to mean some sort of awful concept like “a person who avoids responsibilities in a way not caused by anything whatsoever except being bad, and so we should hurt them and make them suffer”. Are you sure this isn’t kind of dumb? Given that we need a word for the first thing, and everyone currently uses “lazy” for it, and we don’t need a word for the second thing because it’s awful, and most people would deny that “lazy” means that, why don’t we just use “lazy” for the very useful purpose it’s served thus far?

Sophisticus: I think…

Simplicio: And it’s the same with “judgment”. I’m using it to mean a reasonable thing that everyone does and has to do. You’re demanding we reserve it for some kind of ultimate judgment about everything that doesn’t really make sense and probably should never happen.

Sophisticus: I think you’re wrong about common usage. I think a lot of people – maybe not you, but a lot of people – really do use “lazy” to mean the second thing. And that even for good people like yourself, “lazy” has a bit of a connotation of the second thing which you can’t avoid letting slip into your mind.

Simplicio: If you’re right, I worry you’re going up against the euphemism treadmill. If we invent another word to communicate the true fact, like “work-rarely-doer”, then anyone who believes that people who play video games instead of working deserve to suffer will quickly conclude that work-rarely-doers deserve to suffer.

Sophisticus: Then let’s not invent something like “work-rarely-doer”. Let’s just say things like “You shouldn’t have Larry as a dog-sitter, because due to some social or psychological issue he usually plays video games instead of doing difficult tasks.”

I acknowledge that everyone starts from a different baseline (some people are naturally hard working, while some people are naturally less so) but it’s worth noting that Scott has a much different view of the accuracy of the assertions of psych fields than I do. He works in them; he likes (or liked) the kind of articles he links and finds them to be broadly accurate-seeming. I’ve internalized the replication crisis in such a way as to (I think accurately) consider the entire non-pharma sector of the field to basically just be people on the left designing studies to confirm what people on the left already broadly want to be true.

It’s not super relevant to the article since I go another direction with it, but I thought I should note the difference in viewpoint here; if you don’t think that anybody can control anything they do no matter what (Scott doesn’t necessarily claim this, but the first quote in this footnote asks us to behave as if he did and that this claim is true) then you are going to see a lot of what I say differently than if you think it’s possible a situation exists where a person might rightfully catch judgment for something they chose to do.