To attract talent, pay asymmetrically

Hiring advice for the fiscally challenged

TL;DR

• Money is attractive to both the right and the wrong hire.

• Certain goods are more attractive to the right hire than to the wrong: probably prestige, interesting coworkers, and interesting (or important) work.

• To disproportionately attract the right hire, you should roll these goods into your compensation package.

• This is akin to price discrimination, except that instead of targeting a proxy for higher willingness to pay, you’re targeting a proxy for higher competence (caring about prestige, interesting coworkers, et cetera).

• Asymmetrically compensating the right hire is arbitrage. Signals of an interesting company culture can be “bought” in a market that charges the right and the wrong hire the same, and effectively “sold” in a market where the right hires value these signals more.

• Other proxies for competence: self-confidence and fluency in obscure technologies. You can target these in your hiring, too.

---

Why does offering money attract better talent?

One model is that it makes trying harder worth it for talented candidates. (This can look like just bothering to apply in the first place.) If a 50% shot at a $125,000 job is not worth the effort, a 40% shot at a $250,000 job might be. As, say, a tech employer, your hope is that this effort translates to creating flashcards and solving LeetCode an hour every night for at least a few weeks — enough effort for your hiring process to pick up on the (noisy) signal that the engineer is smart and likely a good hire.

The trouble is that increasing compensation makes the wrong hire try harder, too. For all sorts of reasons, this could mean that you make the wrong hiring decision. For example, the wrong hire could be more conscientious and solve every LeetCode problem they can get their hands on. This could lead to better interview performance while remaining worse at real-world engineering than the right hire. 

The fundamental problem is that a greenback is worth roughly as much to the competent engineer as it is to the incompetent engineer. If only you could pay checks that just the competent can encash… Then, you could advertise your compensation scheme front-and-center, and only the competent would apply, and your hiring process would have a lot less error.

If there were, you could do something akin to price discrimination. You cannot directly ask people how much they’d pay for a good (unless you’re a US university) — the buyer would just lie. But you can track proxies for higher willingness to pay, and charge people matching those proxies a higher rate. The classic example here is a company spending money to artificially slow its printers, and selling the non-slowed models for more. The idea is that people who care the most about a speedy printer are also those willing to pay the most for any printer. In India, and I suspect in most developing countries, foreign tourists have to pay three to ten times as much to visit government-maintained monuments. I think the official explanation is that these visitors don’t pay income tax, a portion of which goes towards maintaining these monuments. But really India is trying to get residents of first-world countries to pay something closer to their maximum willingness to pay. And, of course, museums the world over have discounts for people who are disproportionately likely to be broke — students.

You can compensate asymmetrically — or practice compensation discrimination, if the analogy with price discrimination is to be highlighted at the cost of the term sounding bad without context. The key is to ferret out which goods are valued disproportionately higher by competent hires, and roll those into your compensation package. 

I can think of at least three currencies for asymmetric compensation: prestige, interesting coworkers, and interesting (or important) work. 

Prestige

It wouldn’t surprise me if the right hires care disproportionately more about the world recognizing their intellect, their social impact, and so on.

An example of paying asymmetrically in prestige is AI companies popularizing (or, at the least, not countering) the narrative that they’re creating God or “intelligence too cheap to meter”. This narrative has finally stanched the years-long flow of tech talent from software to quantitative finance. The appeal of quantitative trading firms is not fully explained by their high compensation, either. If it were, college students not working at Jane Street wouldn’t be wearing Jane Street t-shirts. Wearing Jane Street merch signals intelligence (you probably got the t-shirt from an internship, a trading competition or a networking event, all three of which are at least nontrivial to get into). But more subtly, it signals a certain personality type — one that plays mathematically involved card games to kill time. By allowing their internal culture and extremely high hiring bar to be advertised, Jane Street has associated its name with a certain personality type.

But if you’re looking to raise the prestige of your field, I claim you can do no better than finance a film. Even if the video games industry makes seven times the money that movies do, books are a more prestigious format, and blogs and newspapers are consumed by the most people, nothing shapes the cultural zeitgeist like films do.

Would you invest in his peer-to-peer music-sharing program?

Byrne Hobart calls The Social Network the most important film ever made because it made founding startups cool. The underdog Mark Zuckerberg showing up to the vaunted gates of Sequoia in pajamas turns startups from “Oh? Like Pets.com? I remember what that did to my 401(k).” to “Oh, like that movie?”

But The Wolf of Wall Street has done for finance at least as much as what The Social Network did for tech. If my experience in college is any yardstick, every investment banking aspirant switched off the film before the last 30 minutes. And, of course, before either film, All The President’s Men made news reporters cool.

But perhaps the most convincing example of films influencing behavior is when said behavior is negative. Tech, finance and news are at least zero-sum, so they’re relatively easy to glamorize. But part of the reason the American Italian mafia is so much more popular than any other crime organization has to be Scorsese and Coppola films. I cannot find the interview now, but I remember Al Pacino and Robert de Niro doing ground research for Heat by talking with criminals, only to find that they’d been making gangster films for so long that the criminals had begun aping them. And I’m talking about films elevating the social appeal of crime, so I won’t even include cases of films popularizing the modi operandi of crimes — which is a thing. For example, one of Joan Didion’s best-known essays is about a Double Indemnity-inspired murder. CSI arguably educated people on how to get away with crimes. I’m only surprised that there haven’t been bank heists inspired by Heat popularizing the idea that the contents of bank vaults are insured.

And, to be clear, I am quite serious that you should finance a film. The billionaire Jeff Skoll funded Contagion, a 2011 film that inspired the British Health Secretary Matt Hancock to order enough COVID vaccine for every adult to have two doses — and made twice its production budget. (Skoll has been doing this for some time now: his other hits include the incredibly influential An Inconvenient Truth, Fast Food Nation, and Spotlight.) Another example of a film influencing an important person: the film HackerWars got Ronald Reagan concerned enough about cybersecurity to pass an anti-hacking law that morphed into the present-day Computer Fraud and Abuse Act. Finally, though they’re print fiction, not films, Isaac Asimov’s Foundation got Paul Krugman into economics (he even wrote the Folio Society edition’s introduction), and Upton Sinclair’s The Jungle launched Teddy Roosevelt’s investigation of the unsanitary conditions in Chicago’s meatpacking industry and the eventual formation of the Food and Drug Administration.

Sure, if you finance a film, the press might find out. But historically, that hasn’t diminished a film’s impact. Barbie was a resounding success, and F1 is going to be one. You don’t need to hesitate even if the film is going to be about a disaster. Just ensure the lead-up to the disaster is sufficiently cool, and as The Wolf of Wall Street has shown, you will be fine.

Interesting coworkers

Perhaps more effective asymmetric compensation is interesting coworkers. Jobs that pay well seem to automatically become prestigious: law, investment banking, consulting. If you’re a startup and your talent acquisition strategy banks on prestige, a bigger competitor can outmaneuver you by simply paying more. Cultivating an interesting team, however, probably requires more care. Patrick Collison, for example, reports spending six months making Stripe’s first two hires — he reasoned that the “lightcone” of hires those first hires made would shape company culture just that much.

In another interview, Collison discusses a problem faced by a startup he knew in Boston. Its greatest challenge: it was in Boston. Not being in the Bay or in New York City is typically death for a tech company. The company’s solution was to start an engineering blog. The things they blogged about were only tangentially related to their work, but a strong signal of their technical sophistication. So, technically sophisticated engineers wanted to work for them.

In general, it seems that having an engineering blog is extremely low-hanging fruit. The overwhelming majority of successful hires at Wave mention CTO Ben Kuhn’s blog as a reason they applied. Dan Luu writes that it should be shocking that his blog gets more visitors than many official company blogs. Companies are working on more varied and interesting problems, and in Luu’s own opinion, have more clever solutions for them. Yet, they’re oddly slow about approving blog posts, and conservative about what goes in them. Cloudflare is a trailblazing exception here, who have attracted “at least a few … generations of folks who interviewed because they saw a blog post”.

Another example: if you want people who like reading, you can stuff workspaces and hacker houses with bookshelves, offer an unlimited office book budget as a job perk, et cetera. Just the fact that you care about books is a signal, but the choice of books will matter, too. If you want mostly technical people, do textbooks. If you want more well-rounded people, include art.

This last example shows another benefit of asymmetric compensation: it is a kind of arbitrage. By buying goods in a marketplace that charges both the right and the wrong hire the same (Amazon/AbeBooks/what have you), and selling them in a marketplace where these goods are valued more by the right hires, you get bargains. Arbitrage speakers: Adam Marblestone, Michael Nielsen, Robin Hanson and Anders Sandberg are indistinguishable from other respectable academics to the average person (and therefore as much effort to invite), but the right crowd will go crazy for them. (Not a slight to their work; I am in that crowd, and think their work is more important than the average respectable academic’s.) 

Interesting work

If you want your field to be exciting to the layperson, finance a film. If you want your field to be exciting to the right candidate, finance an Aaron Sorkin film. With The Social Network, he made startups interesting to would-be founders, but his masterwork in elevating a profession’s sexiness is The West Wing. 

Do you know which is the only fruit that has seeds on the outside? Jed Bartlet does.

Practically no one, least of all Sorkin himself, thought that the day-to-day work of executive staffers is the stuff of drama. It was while making a movie on the President — which everyone agrees is a job worth writing drama about — that he realized that the work of staffers is interesting.

And his skills at making it interesting to the right kind of person were incredible. Few other media not about war evoke as much patriotism and a doe-eyed optimism about personal agency. The President welcomes one of his most impactful staffers into the fold with a Margaret Mead quote: “Never doubt that a small group of thoughtful, committed, citizens can change the world. It is the only thing that ever has.” 

The characters in the show are whip-smart. To prove this, Sorkin’s characters talk jargon when they talk business, and talk fast. I mean, look at this. Which other TV show deliberately makes conversation hard to follow? The point of the jargon is not that you get what’s going on, but that you get that the characters get what’s going on. Very often, even “show, don’t tell” is abandoned. The entirety of the third season is about how it’s a problem that the President looks too smart running against his Republican opponent. He can speak Latin, convert from Fahrenheit to Celsius in his head, and knows the three English words with “dw” in them. If that weren’t enough, he has a Nobel prize for development economics. His staffers are no slouches, either. At one point, the President says that two of them have “300 IQ points between them”. Another went to Princeton, edited the Duke Law Journal, and calls one of the other staffers his favorite nonfiction writer (his favorite fiction writer is Dickens). 

Unsurprisingly, the show’s intellectualism and idealism drew many staffers to the West Wing, some of whom would’ve had “zero interest” otherwise. More surprisingly, a West Wing episode directly inspired the mechanics of a coup against a Tony Blair bill. The West Wing popularized the idea that White House staffers matter a lot, more than most elected politicians. I wonder how many takes like Noah Smith and Andrej’s are causally downstream of it.

If the work itself cannot be made interesting, you can settle for having a mission. As Byrne Hobart explains, a mission-driven company is in a sense “very expensive consumption by rich people” because it has to swallow opportunity costs in service of a grand long-term goal. But having a mission is also a subsidy — it’s plausible to me that the right hire has the ego to care about their social impact, and will work for a mission-driven company at personal cost. Hard startups are easier than conventionally easy startups in this sense. Neuralink may find sourcing talent easier than another “Tinder for X” simply because talented engineers like working on important things.

Other ways to discriminate for competence

I’ve focused in this post on compensation. However, the insight from price discrimination to target proxies for the thing you care about is a pretty general one.

Drop the CSS. It’s cleaner.

• Companies artificially slow their printers, disable the math coprocessor on their chips, write software to reduce their discs’ storage, and spread rumors about their chemicals containing arsenic. All to identify those with a higher willingness to pay for a superior good. Similarly, rather than offer the right hire more, you can offer the wrong hire less. Artificially reduce your firm’s prestige: misspell words on your website and use a WordPress default theme. If you’re an AI company, don’t get a .ai TLD. Also, don’t add “Labs” to your name. Make having worked for you look as unimpressive on a resume as possible. The people who still want to must really care about what you’re doing.

• Tying pay to performance is an obvious win — the employee who expects to perform poorly will not accept performance-based compensation. According to Sebastian Mallaby’s More Money Than God, one of the great innovations of Alfred Winslow Jones — the first hedge fund manager — was figuring out how to compensate brokers for their stock picking skill rather than their market exposure. But the cleverest example of performance-based pay may be the lionization of the scientific method. As Steven Landsburg explains (hypothesizes, really) in The Armchair Economist, if you’re trying to source your theories only from smart scientists, of course you pick those who have enough confidence in their models to make predictions before looking at test data. Being confident about one’s smarts is a decent proxy for actual smarts, and having smarts is a good proxy for having good theories.

• Use obscure technologies. Back when Python was up-and-coming, Paul Graham mused that companies that use Python must fare better. This is because only developers who care about their craft would bother learning the language — these are disproportionately likely to be good hires. It’s no accident that the highest-paying languages on the StackOverflow survey are rather obscure: the top three are Zig, Erlang and F#. By using OCaml, Jane Street filters for engineers who like functional programming; an engineer with an opinion like that may or may not be right, but they’re bound to have written a lot of code.

The takeaway is that it is worth asking yourself: what’s something the right hire finds much more valuable than the wrong hire, and can you give them that?