Elon Musk and spiky intelligence

Like a lot of successful people, he’s world-class in some dimensions of intelligence but deficient in others.

A couple of quick site announcements before we proceed to today’s main event. First, February is a short month, which means it’s already almost time for the next edition of Silver Bulletin Subscriber Questions. Paid subscribers can submit questions in the comments of SBSQ #17.

Second, I’ll be doing a Substack Live video chat with Nobel Prize winner — and fellow New York Times refugee — Paul Krugman at 11 a.m. Eastern this Thursday, February 27. Paul needs no introduction, but you should definitely check out his Substack if you haven’t yet. We expect to talk politics, media, economics, and all that other good stuff. You’ll need to use the Substack app to see the conversation live, but I’ll also send it out over email once it’s done.

---

There’s been a debate raging on Twitter — Noah Smith can run you through the parameters — about the intelligence of the platform’s owner, Elon Musk. My contribution was to suggest that (i) Elon is obviously pretty bright, and (ii) this shouldn’t be conflated with moral judgment — highly intelligent people do lots of bad things.

You’d think this wouldn’t be especially controversial, but since it involves Elon and intelligence — well, it was. Elon Musk has run, founded, or cofounded Tesla, SpaceX, OpenAI, Neuralink, xAI, PayPal — and more recently, Twitter (excuse me, “X”). He’s also managed to steer himself into a position where he’s now the de facto Chief of Staff to the President of the United States. I do not doubt that Elon has gotten lucky in various respects: some of these were long-shot bets, and Walter Isaacson’s biography of Musk documents how he thought he’d be ruined if there had been one more failed SpaceX launch. The success of some of these enterprises might also be debated. Twitter was a canny play for cultural and political influence but probably won’t produce a positive financial ROI, and we’ll see how the whole co-presidency thing goes. But I don’t care what Elon’s SAT score is (1400, according to Isaacson). He’s clearly some sort of outlier in ways most people would associate with intelligence. Probably even a “genius.”

And yet, when my partner and I were heading to dinner the other day and saw some tweet that Elon sent — I forget which one because he tweets so much — we were both like, “Man, he’s such a dumbass.” Yes, someone can be both a genius and a dumbass. Welcome to what I call “spiky intelligence.”

What is spiky intelligence?

The term “spiky intelligence” isn’t entirely original, although it isn’t commonly used. Interestingly, many of the instances online refer to people on the autism spectrum — Musk has publicly stated that he has Asperger’s syndrome. But the concept is simple. While intelligence is a multi-dimensional phenomenon, the scientific consensus is that there’s also something known as a “g factor,” sometimes also called general intelligence. As an empirical matter, most traits we’d associate with intelligence are positively correlated — for instance, math and verbal scores on the GRE are correlated.

But the correlations are loose enough that you will wind up with all different permutations on the spectrum of human behavior. Some people are highly proficient or even world-class in some areas but deficient in others — think of the absent-minded professor stereotype. There may even be trade-offs between certain types of intelligence: Simon Baron-Cohen and other researchers, for instance, have postulated that there’s an inverse correlation between systematizing and empathizing: people tend either to be interested in abstract ideas or interested in other people, but not so often both.

My book, On the Edge, describes a certain community of intelligent people that I call “the River.” These people, who occupy a range of professions from AI research to poker to venture capital, are bright but in spiky ways. In Baron-Cohen’s dichotomy, they lean heavily toward the “systematic” side of the equation. They are good at abstract, analytic reasoning. But, they may lack other forms of intelligence like empathy, judgment, and self-awareness. They also have some distinctive characteristics largely unrelated to intelligence — for instance, they tend to be extraordinarily competitive and somewhat “contrarian.”

For better or worse, this typology is strongly correlated with high achievement in certain highly lucrative professions, especially tech and finance. It is also associated with high variance: Sam Bankman-Fried built FTX into a company that investors valued at $32 billion before the house of cards collapsed and he was sentenced to 25 years in prison (which he’s now tweeting from for some reason). I interviewed SBF several times for the book, and I can tell you that he very much falls into the “genius but dumbass” category. I’m sure he has a high IQ, but he demonstrated both abysmal judgment and a complete lack of a moral compass.

So, it’s important to avoid two pitfalls when encountering people with spiky intelligence. Namely, neither their worst traits nor their best ones tell the whole story:

• That Elon is highly intelligent in several ways does not mean that everything he does is brilliant. Some things he does are exceptionally dumb or dangerous — and we shouldn’t make excuses for these or pretend that it’s all part of some master plan.

• But likewise, it’s absurd to suggest that Elon isn’t brilliant in many respects just because he isn’t in others. And if he has merely very good SAT scores, I don’t care; he’s demonstrated his intelligence through his accomplishments. This is a bit like criticizing Tom Brady because he had mediocre ratings in the NFL combine. The SAT may be correlated with general intelligence, but it’s an imperfect proxy for it, and test-taking skills like figuring out verbal analogies¹ aren’t too important on their own.

A quick inventory of Elon’s intelligence

I wasn’t able to track down Elon for the book — believe me, I tried — so none of this is based on first-hand observation. I could flatter myself by saying I know Elon second-hand — I know a lot of people who know Elon — but those people are mostly peripheral contacts rather than close friends. So, I’m mostly modeling and extrapolating from the many other Silicon Valley bigwigs I have met. Plus, Musk maintains an exceptionally public profile — he’s turned X into a running diary of his innermost thoughts — and there are various highly detailed biographies of him, both from Isaacson and a previous biography by Ashlee Vance.

One more caveat here: I will try to evaluate the overall trajectory of Elon’s career — not just his recent antics. I don’t know precisely the extent to which he’s “always been this way” versus having changed — though his former friend, Sam Harris, says Elon has changed, and that’s my guess, too. It’s not hard to think of reasons for this. Politics and social media poison a lot of people’s brains. Having that much wealth and power has to be intoxicating², especially if (as Harris’s experience suggests) Musk ostracizes people who might keep him grounded. More sympathetically, he is taking on an incredible array of responsibilities — doing several really hard jobs at once, each of which would be highly stressful on their own — while still managing to father 13 children and tweeting hundreds of times per week. With that out of the way, here’s my assessment.

Dimensions where Musk has exceptionally high or genius-level intelligence

Cognitive load capacity and overall horsepower/“RAM.” He’s always “on”. I mean literally — look at how often he’s tweeting:

Is this sustainable? Well, I doubt it, and we’re probably seeing the effects of that. But the Isaacson biography convinced me that Elon isn’t just sitting in an empty room tweeting all the time; he’s tweeting while jaunting around the world looking after his various enterprises and interests, hosting dinner parties, fathering children, and so on. And now he’s sort of co-president of the United States, too.

In NBA terms, we’d say this is a player with an exceptionally high “motor.” And this is undoubtedly a valuable trait as the world becomes more complex. Last fall, I was simultaneously doing an extensive book/media tour, running the election model, trying to build up Silver Bulletin, plus some intensive consulting work. Even if I mostly kept my wits about me, it was an incredible amount of mental and physical strain that would only have been sustainable for a short burst. But Elon is taking on — I don’t know — approximately 1000x more stress than that and has done so for years. You can at least give him some credit for being the ultimate “multitasker” — few other people would even attempt this.

Rapid cognition and thin-slicing ability. There are a lot of testimonials that Elon is — or was — good at this, and it’s a skill highly prized by Silicon Valley. Indeed, in a capitalist system with a significant premium on being first to market, making decent judgments fast is often more important than making better judgments slowly. Canonically, VCs imagine themselves rapidly filtering through potential founders as though on Shark Tank, relying on well-honed gut instinct. But this also often gets people in trouble, as I think it has for Elon. Their snap judgments might be poor in areas outside their domain of expertise, and they can become overconfident in their instincts and slow to correct their mistakes.

Abstract problem-solving ability. This is related to the idea of “creativity,” although in Musk’s case, it seemingly doesn’t manifest itself in artistic prowess. However, the Ashlee Vance biography suggests that Musk can see through engineering problems exceptionally well, keeping an eye on the project goal while identifying workarounds that others might miss.

Instrumental rationality. Philosophy nerds like to distinguish between two types of rationality. Instrumental rationality is aligning means with ends — basically, figuring out the most efficient ways to get what you want. (We’ll cover the other dimension of rationality later.) For this category, I think you have to point toward the scoreboard: Musk has some unparalleled accomplishments and isn’t about to let anybody or anything stand in his way. It’s also a category often associated with manipulativeness or even being an asshole, not one for nice guys — but there’s a type of intelligence there. Other terms for it might be willpower or shrewdness.³

Dimensions where Musk has unexceptional or even below-average intelligence

Before we get into these, I should admit to some ambiguity about the benchmark that we’re measuring Elon against. Some of these might be categories where he’s below average relative to other successful people but still above average relative to the population. But they are, at least, not comparative advantages for him.

Erudition and “book smarts”. I think of this as someone who would be a good Jeopardy! contestant or a witty guest at a dinner party, regaling you with his or her worldly knowledge. It’s a commonly recognized type of intelligence, though one more often associated with “the Village.” (The Village is the rival community to the River in the book, basically the liberal “establishment,” including fields like media and academia.) Elon has deep knowledge in some areas — engineering questions, particularly — and some hobbies and interests (sci-fi, video games). Still, they’re pretty middlebrow, and his tweets make clear that he’s frequently outside of his depth in other areas.

Emotional intelligence. In principle, this should be high-ish for VCs and, to a lesser extent, founders. VCs must persuade founders to sign up with them, while founders must have a persuasive enough story to win over investors and recruit talent. However, because Silicon Valley is small and close-knit, many founder types are only persuasive within relatively narrow confines, i.e., to other VCs and founders, which often explains why they have poor political and social instincts outside these circles. Elon fits this description to a tee. He can evidently motivate some fraction of employees to work incredibly hard for him, but many quit or burn out. I don’t think Elon would even claim to have high EQ — he’s often cited his Asperger’s for his social awkwardness.

Judgment and wisdom. This trait is often in tension with some dimensions where Elon rates highly. In fact, VCs often deliberately select for spiky, high-variance skill sets. Wisdom typically rises with age, but founders are usually young and lacking in life experience. (Or, in Elon’s case, they “never grow up”.) However, this fits the industry's incentives. The bets VCs make are asymmetric — tails, you lose only 1x your investment, but heads and you might make 500x. They want people who will pursue “crazy” ideas that may have a high failure rate — but nevertheless have positive expected value because the payoff might be huge.

Epistemic rationality. In contrast to instrumental rationality — choosing means that effectively suit one’s ends — epistemic rationality is seeing the world accurately for how it really is. This would be demonstrated, for instance, by the ability to make well-calibrated probabilistic predictions — or, more generally, being highly motivated by the pursuit of truth. Although a strength for some Riverians — it’s an important attribute for poker players, for instance — epistemic rationality and the related epistemic humility are often weaknesses for founders. And especially for Elon, who has a long history of outlandishly wrong predictions, broken promises, and overall grandiosity. It’s also a trait that often worsens for people as they achieve more success. If you’re an exceptionally competitive individual who has recently won some huge contrarian bets, it’s hard for that not to go to your head.

Why spiky intelligence is common among successful people, with sports metaphors

Although I’ve listed an equal number of categories in the good and bad buckets for Elon, I don’t mean to imply that this averages out to merely “pretty good” intelligence. For one thing, I assume that these traits follow some sort of normal distribution — if you’re on the far right tail in several categories, 2-3+ standard deviations above the mean, that can often make up for several shortcomings. Wilt Chamberlain, for instance, was a notoriously poor free-throw shooter (51.1 percent for his career), far below the standards of not only the NBA but even high school boys’ basketball. It didn’t matter because free-throw shooting is only moderately important and Wilt was extraordinary enough in other areas to more than make up for it.⁴

There’s also a statistical phenomenon called Berkson’s paradox that helps to explain why, in any given field, successful people often seem to have some sort of Achilles’ heel. Here’s some data that I simulated, for instance. Imagine that you take 1000 college basketball players — say, rotation players on major conference squads, plus some players of equivalent ability from smaller schools. Your scouting service ranks them along two dimensions — defense and shooting⁵ — which you regard as equally important. Analogous to g or general intelligence, these skills are somewhat positively correlated.⁶ From this list of potential recruits, you select the 50 best players to be drafted into the NBA based on their combined shooting and defensive ratings. But look what happens:

In the truncated data set — only the players who reach the NBA — there’s a strongly negative correlation between defense and shooting, even though they are positively correlated in the population overall. (That is, players who are great shooters tend to be below-average defenders relative to the rest of the league and vice versa.) The intuition is that as hard as it is to win the genetic lottery once, it’s much harder to win it twice.⁷ It’s really hard to have world-class ability — several standard deviations among the mean — in any one area. Only once in a blue moon do you get a Shohei Ohtani type who’s great at everything, including skills (batting and pitching) that are normally relatively uncorrelated. So most of the time, even in highly competitive fields, you settle for people with one or two truly outstanding skills — and then tolerable skills in the other departments so as not to drag the team down.

Silicon Valley’s selection process for founders is much less efficient than the NBA. There are fewer “tangibles” to measure, and there’s a lot of guesswork. There’s a lot of crude pattern-matching, including a bias toward white men of certain ethnic backgrounds. (I say that as someone who’s not super woke; there’s more about it in the book.) For every founder who gets a round of Series A funding, there are probably 10 or 100 would-be founders who would do roughly as well if they got the same chance. That doesn’t mean founders the VCs do invest in aren’t smart or talented — most of the ones I’ve met are super smart and dedicated. There just aren’t enough lottery tickets to go around.

However, Silicon Valley is also competing against other professions. Let’s take the same graph as before but shift from basketball skills to different types of intelligence. Respectively, these represent the departments that Elon is comparatively good and bad at. Along the Y-axis, we have raw mental horsepower coupled with instrumental rationality. On the X-axis, there’s wisdom, EQ, and epistemic rationality.

Even if Silicon Valley values EQ, other professions might value it more. Doctors and lawyers also have plenty of intellectual horsepower, for instance, but — depending on the type of doctor or lawyer — they’ll also need a heavy dose of empathy and “people skills.” So, these fields will often win the “draft” for talent higher on the EQ dimension. Meanwhile, the 1600 SAT types with “pure” abstract intelligence may gravitate more toward game-like fields like quantitive trading or AI engineering, regardless of their social intelligence.

From that point forward, the culture of each industry can be self-reinforcing. Silicon Valley is notoriously tolerant of “difficult” behavior from its founders, and perhaps that’s wise given the talent they recruit. But that can also repel people who are more well-rounded and less cutthroat. So, each industry quickly becomes a stereotype of itself. It becomes spikier, in other words.

All that might work out just fine as far as Silicon Valley goes. I’m mostly a techno-optimist, notwithstanding my concerns about AI. It’s certainly a lucrative industry, and it’s able to recruit talent from all around the world, often getting the top picks on its draft board. But being co-president requires a much more rounded sort of intelligence — not the type that Elon has.

1. Analogies have since been dropped from the SAT, though they were present when Elon would have taken the test.

2. Executives and investors at Elon’s companies have also expressed concern that he is sometimes literally intoxicated.

3. President Trump probably also rates highly along this dimension if lacking in other areas.

4. There are even some theories as to why traits that allowed Chamberlain to excel in other areas (such as his height) made free-throw shooting harder for him.

5. I specify “shooting” rather than “offense” because, in basketball, certain types of offensive skills are naturally related to defensive abilities. Being tall makes you more fearsome in the post, for instance, as well as a more intimidating rim protector. And rebounding skill helps both on defense and offense.

6. Specifically, a correlation of about 0.4 throughout the player pool.

7. Even if the draw is rigged in such a way that you have above-average chances on each spin of the tumbler.