On (Noticing) IQ (When You're Allowed To)

Always noticing.

I was writing a reply to @TowardsPanna 's question in the SSQ thread that got a bit out of control. Large enough that I decided to just post it here too.

For what it's worth, none of these ideas are particularly new, at least they aren't if you compulsively browse LessWrong (or even read a lot of Scott), but the older I get, the more I realize that novelty is often just a nice-to-have.

Does an IQ taboo (established for political reasons associated with another taboo around HBD, or any other reason) contribute to more of a reliance in many people on the heuristics of social class, physical features, clothes, sociolect and prosody, credentials/profession, or even ideological conformity - and thus more of an opaque and effectively hierarchical society? A society where appearances become more important than the underlying reality, and where presumptions are not challenged? Where the average individual, who insofar as they've been taught anything about mental horsepower, has come to believe that it's about the development and growth mindset - any child can join any profession if they work hard and choose through free will to develop smartness; and knowledge - the person in higher education studying e.g. psychology becomes smart and competent through their acquiring of knowledge and routines and joining into a professional group? The flipsides of these coins being that someone who didn't go to college/university has stupidly chosen not to become smart and grow their brain and thus can blame themselves and should not be given much time of day? And if appearances are what matters, someone with perfect grooming and high class speech and all the shibboleths should be assumed to be smart rather than looking under the hood?

Yes. Of course?

Humans aren't stupid. We’re expert pattern-matchers. We’re distinctively evolved to be relentless Bayesian updaters, constantly scanning our environment for correlations that offer a survival or status advantage. We’re Noticers™. The problem is that we’ve built a civilization where noticing the most predictive variable for life outcomes is considered a faux pas at best and a moral failing at worst. It’s common now to explicitly state, in corporate policy or legislation, that specific classes of Bayesian evidence are "fruit of the poisonous tree" and have to be ignored.

Like many well-intentioned interventions that hinge on obscuring reality, it doesn't work: when you outlaw the most accurate map, people don’t stop navigating. They just buy worse maps.

In the ancestral environment, we didn't have the Raven’s Progressive Matrices (only the Modern Hunter Gatherer has had the privilege). We did, however, have proxies. We had vocabulary size, wit, speed of learning. In the modern environment, we’ve muddied the waters. We’ve got proxies that look like intelligence but are actually just measuring conformity to the upper-middle-class Blue Tribe aesthetic. We look at credentials, which are increasingly just a measure of "willingness to sit still and follow instructions for four years."

(Please note that I don't think that education is purely credentialism. This is a disease that affects midwits first and foremost. A bachelor's in physics is a stronger signal than one in sociology.)

So when we agree to pretend that "mental horsepower" is a myth, or that it’s entirely malleable via the Growth Mindset, we don’t actually create a flat hierarchy. We just create an opaque one. We create a hierarchy based on the performance of competence rather than the reality of capacity.

This leads to a specific kind of societal gaslighting. The narrative is that anyone can be a doctor or a quantum physicist if they just work hard and acquire knowledge. If someone fails to climb the ladder, the implication is that they stupidly chose not to grow their brain. They lacked grit. They lacked character. In a world where aptitude is taboo, failure’s always a moral deficit.

The obvious analogy is a world where the Paralympics and the "normal" Olympics are wrapped up into a single event. Anyone, including the disabled athletes, who points out that missing a leg makes one very unlikely to win in the 100m sprint is immediately walloped and then piled-upon, with cries of "the Science! Not-Murray et al clearly showed, in their landmark 1998 paper, that the number of remaining lower limbs had no relation to performance in the sleighing, shooting and arm wrestling categories. The 2012 Olympics were studied in great detail, and it was confirmed that Usain Bolt had the same number and arrangement of legs as his closest competitors."

When the person with no legs fails to win the 100m sprint, the commentators nod gravely and say, "He just didn't want it enough. He failed to manifest the victory."

I see three main mechanisms driving this, and why the average person (even the smart average person) underestimates the sheer weight of g.

The "All Men Are Created Equal" Overcorrection:

We’ve engaged in a concerted propaganda campaign for decades suggesting that because moral equality is mandatory, biological identity must also be mandatory. This makes it practically unthinkable to consider that it’s perfectly possible to construct a political system that grants equal rights to all citizens while acknowledging that not everyone can visualize a rotating 3D cube with equal fidelity. But biting that bullet feels dangerous to many. It feels like opening the door to old aristocracies. So instead, we pretend the differences don't exist, which inadvertently creates a new aristocracy of "people who know the right shibboleths."

The Bubble and the Range Restriction:

This is the big one. We live in intense cognitive bubbles.

If you’re reading this, you probably spend your life surrounded by people within one standard deviation of your own IQ. You went to a selective university. You work in a cognitive field. You live in a zip code priced for high-earners.

You can go days, maybe weeks, without having a meaningful conversation with someone significantly below an IQ of 100, or even +- 1SD outside of yours.

This creates a statistical illusion known as range restriction. If you look at the NBA, height doesn't correlate very strongly with points scored. Everyone’s already tall; therefore, the variance in performance seems to come from practice, grit, and shooting form. If you looked at the general population, height would be the single biggest predictor of basketball ability (because the 5’5” guy isn't getting the rebound).

Imagine if the NBA had a rule that you couldn't mention height. You'd still need to pick players, so you'd start using proxies. You'd talk about "reach" and "wingspan" and "vertical leap." You'd notice that certain players had better "court vision" or "positional awareness." And all of these would be real skills, but they'd also be correlated with height, and height would still be determining who made it into the league in the first place. The difference is that now you'd be pretending you weren't noticing height at all, which would make you more likely to mistake the packaging for the product.

In your life as a student or a professional, you look around at your peers. You see that Alice is doing better than Bob. Since Alice and Bob are both roughly equally smart (they both made it this far), you attribute Alice’s success to her conscientiousness, her charm, or her work ethic. You conclude, erroneously, that "intelligence doesn't matter much, it’s all about hard work."

You don’t see the people who were both dumber and less hardworking than you; they’re in a different social class entirely. You rarely see the people who are simultaneously smarter and more hardworking; they’re running hedge funds or solving alignment theory and don’t hang out at the hospital cafeteria. These are far more likely to be acquaintances rather than peers.

The problem is when we generalize from this filtered view. We start believing that because IQ doesn't predict success within our bubble, it must not predict success period. We see a colleague who's a bit slower but works incredibly hard and does fine, and we extrapolate that to everyone. We forget that the slow colleague is still in the 85th percentile of the general population, and that the person in the 30th percentile who works just as hard isn't a doctor at all, they're doing something else, probably something that pays worse and doesn't sound as impressive on a Hinge profile (see proxies re-emerging?).

The Opacity of Alien Competence:

Some professions are more segregated than others. An ER physician arguably sees a more representative slice of humanity than a family medicine specialist, who sees a more representative slice than a Google L10, but the direction of judgment is difficult.

When a patient comes to see me, I’m performing a distinct ritual. I listen to symptoms, I peer at them significantly, I type things into a computer. To the patient, the output seems to appear from thin air. A sufficiently competent doctor makes it look like they never broke a sweat. A less experienced one (like me) makes up for it by projecting confidence and then panic-Googling the interaction effects of MAOIs in the bathroom later.

The patient can't judge my raw processing power. They can only judge my bedside manner, my clothes, and my degree on the wall. They’re judging the paint job.

When I'm on call and a patient is rambling in a disorganized way, using neologisms, showing thought disorder, I don't need to know their IQ score to recognize that something is cognitively wrong. But when I'm in the doctors' lounge, listening to two consultants argue about whether a patient's depressive symptoms are primarily biological or reactive, I also don't need IQ scores to know who's making the sharper arguments. The difference is that in the first case, we're allowed to talk about cognitive impairment. In the second case, we're allowed to talk about "clinical judgment" or "experience" or even "medical education," but we're not supposed to talk about the underlying mental horsepower that makes one doctor better at diagnostic reasoning than another.

(Ideally, the whole point of the medical education system and associated exams is to put the gatekeeping before patient contact. When you see an MD in the wild, you ideally want to not need to bother with asking for a transcript of their grades, nor fear that they don't know the safe limit for paracetamol. The reasons why this idyllic state of affairs doesn't hold are too lengthy to fit on this margin. Be smart about things.)

Conversely, I can't judge the competence of a senior software engineer. I know enough Python to automate a spreadsheet, but can I distinguish between a decently competent programmer and a 10x developer without referencing their credentials or status? Probably not.

This brings us back to the original shower thought. If we can't discuss the engine, we obsess over the trim.

In a "blue" environment where social class is ostensibly "over" and never discussed, it’s actually the primary sorting algorithm. We use heuristics like "do they speak with a Received Pronunciation or General American accent?" or "do they know the current correct terminology for this specific social justice concept?" to decide who’s smart.

This favors the people living and breathing inside the dominant culture. It favors the legacy admission who knows how to dress and speak. It subtly closes out the dissenting voice from the outsider who might have raw supercomputer-level processing power but speaks with a regional accent, wears ill-fitting clothes, and hasn't learned the subtle dance of feigning humility while signaling status.

Some professions, like programming, are comparatively more sane/honest. You can have a perfectly decent career in FAANG if you don't shower regularly and speak with a lisp, assuming you are actually good at your job. Hell, like the dude with the MLP (pony, not perceptron) resume, you can counter-signal by being incredibly eccentric. If you're still drawing a seven figure salary, then you're worth it.

Speaking very generally, I think society (a vague term, can't help it) needs to move in the direction of more meritocracy, not less. In practice, that looks like more standardized testing, with reduced focus on vibes.

The vibes are, paradoxically, easier to fake than the exam results:

Consider the standard critique of the SAT: "It just measures how rich your parents are."

This is statistically false. The correlation between SAT scores and parental income is real but moderate (around r = 0.4). But compare this to the correlation between parental income and "being captain of the fencing team" or "having a polished personal essay about your volunteer work in Peru" or "speaking with the correct sociolect during an interview."

The SAT is a partially gameable metric. "Holistic" traits are entirely gameable products.

If you are a rich parent with a dim child, you cannot tutor them into a 1600 SAT. You can maybe get them from a 1000 to a 1150. But you can buy them a spot in a prestigious internship. You can hire a consultant to write their essays. You can ensure they have the "right" hobbies.

Since this is my soapbox, and you're presumably still reading this far, I also want to speak out against another approach towards modern fetishization of pseudo-meritocracy, or more accurately, aristocracy.

I should probably elaborate on the @2rafa position, since she's the strongest advocate (only advocate?) She's old money. My impression being not "my parents paid for Stanford" money, but "my great-grandfather's trust fund paid for Stanford's new rowing pavilion" money. She'll happily tell you that the problem with modern society isn't that we sort by intelligence, but that we pretend to sort by intelligence while actually sorting by a particularly vulgar form of striving that she finds aesthetically repulsive.

Her critique goes something like this: The modern PMC striver is a grotesque creature. The Tiger Mom. The résumé-padding, LinkedIn-posting, "crushing it" bugman who measures his self-worth in LeetCode problems solved and whether he made partner by 35. This, she argues, is what you get when you tell everyone that anyone can be elite if they just grind hard enough. You don't get equality. You get a soulless arms race of performative busyness, a cargo cult where the striver apes the surface of elite competence without acquiring the substance.

Better, she says, to return to a system where everyone knows their place. Where the sorting happens early, quietly, and permanently. Where you don't try to be elite, you simply are. Where excellence is demonstrated through effortless grace, not desperate optimization. The boarding school accent. The understated wardrobe that costs more than a car. The ability to discuss Proust and quarterbacks with equal nonchalance. The aristocratic assumption that if you have to ask, you're not one of us, and that's fine. It's a system that rewards being rather than becoming.

From her vantage, this is obviously superior. And I get the appeal. If you're already at the top of the heap by accident of birth, a system that freezes the heap in place is very comfortable. You don't have to worry about some brilliant kid from a refugee camp out-hustling your mediocre son for the last spot at Harvard Medical School. Your son's spot is secure, not because he's particularly gifted, but because he's yours. The system can quietly acknowledge his inherited position without anyone having to say the quiet part out loud. The "genteel" facade is the point, it transforms raw inherited advantage into a question of taste.

This is far from the worst approach. The elites are elites for a reason. That reason is often hereditary. Even total regime change and societal upheaval usually has the dispossessed children of nobility (or the elites) almost inevitably percolate back to the top. This is evident in data from places as far-flung as China, where the grandsons of pre-revolution landlords often outperform their peers. The previous status-quo was bearable, in some ways superior.

However, the aristocratic alternative often ends up parasitic on the very meritocratic machinery it despises. You still need high-g selection somewhere. You still need the engineers, the surgeons, the generals who can think. You can dress that up in tradition and patronage, you can recruit them as client talent, you can offer them a place in the court, but you cannot run a technical civilization on inherited gentility alone. At some point, reality reasserts itself, usually via catastrophe.

But here's where the rubber meets the road: I am that brilliant kid's son. My dad did out-hustle the mediocre sons of privilege. He clawed his way out of a refugee camp because somewhere, a grinding, impersonal system looked at his test scores and said: "This one. This one is worth plucking from the mass and polishing." He wasn't sorted by vibes. He was sorted by a standardized exam that didn't care about his accent, his hand-me-down clothes, or whether he knew which fork to use at a state dinner. It cared whether he could solve the problems in front of him, quickly and correctly.

The "genteel sorting" system that @2rafa prefers would have written him off before he started. He didn't have the right pedigree, the right consonants at the end of his name, the right summer internships. He had the wrong everything except the one thing that actually matters for medicine: the ability to hold a thousand variables in his head while making a decision under pressure. The exam caught that. The "holistic" process would have missed it, distracted by his lack of polish.

So yes, I have a personal bias. I believe in meritocracy because meritocracy is the only reason I'm here, writing this, instead of hauling crates in a warehouse or pulling weeds out of a farm. But my bias aligns with a principle: if we're going to have sorting, and we are, because complex societies require it, then let the sorting be honest. Let it be based on the thing that actually predicts performance, not the cultural markers that predict comfort for the existing elite.

The aristocratic approach pretends it's avoiding Goodhart's Law by refusing to articulate its metrics. But the metric is still there: it's called lineage. It's just a metric that can't be improved upon, only inherited. And while it's true that modern meritocracy is imperfectly gamed (that's what this entire post is about) the solution isn't to replace an imperfect but theoretically climbable ladder with a walled garden whose keys are handed out at birth.

The modern PMC striver is indeed a pathetic figure in many ways. But he's pathetic because he's been lied to. He's been told that credentials are everything, then handed a system where credentials are increasingly just proxies for the ability to acquire credentials. He's been told that growth mindset will make him a doctor, then sorted by an IQ test disguised as the MCAT. His sin isn't striving. His sin is believing the official story, and optimizing for the proxies rather than the underlying reality.

@2rafa's genteel system doesn't fix this. It just makes the proxies even more opaque and even more heritable. It replaces the MCAT with the recommendation letter from your Exeter headmaster. It replaces the LeetCode grind with the unspoken assumption that of course you'll summer at the firm your father's college roommate runs. It removes the last remaining pressure points where someone like my father could punch through.

What we need isn't a return to aristocracy. It's a return to honesty. Acknowledge that g exists, that it matters, and that it's largely heritable. Then build a system that finds the people who have it, wherever they are, however they present. Make the exam harder to game, not easier. Make the credentials less important, not more. And stop pretending that the alternative to vulgar striving is egalitarianism. It's very much not. The alternative is feudalism with better manners.

---

Sadly, none of this particularly matters in the long-run. The AI will "meritocratically" take your job, and will eventually do it better than you can imagine. My kids aren't going to college. Yours probably won't either. I find that reassuring, in some ways, short timelines taken seriously make a lot of squabbling moot. You can stop running so damn hard, the winner has a jetpack. Isn't that oh, so reassuring?