“Bullshit Jobs” is a Terrible, Curiosity-Killing Concept

Plus! Sanction Economics; Exiting; Building Complements; Shares and Votes; Quality Problems; Diff JObs

In this issue:

The Diff July 8th 2024

Today's issue of The Diff is brought to you by our sponsor, Warp.

“Bullshit Jobs” is a Terrible, Curiosity-Killing Concept

It's every nonfiction author's dream to stick a name on a concept that's so incisive it literally outlives them. David Graeber died in 2020, but "Bullshit Jobs" still gets thrown around a lot. I was uncomfortable with the concept, but had never read the book so I didn't feel especially comfortable being dismissive. This was a mistake. If you don't own a copy, don't bother buying one. If you own a copy, consider reading it an act of meta-anthropology, exploring why a professional anthropologist could be so relentlessly, aggressively incurious about the lives and experiences of others. In fact, reading the book in that light, as an exploration for how someone could come to believe strange things, and what approach they might use to marshal statistical and anecdotal evidence in their favor, is a great exercise. We won't all make the same mistakes, but we're all capable of making the same kinds of mistakes.

The observation the book explores is that there seem to be many jobs that just aren't necessary. Graeber lists a few of these bogus occupations[1]: tax lawyers, marketing consultants, actuaries, HR consultants, financial strategists, etc. Since the book was written to expand a previous article, he has room to backtrack on at least one of those, conceding that actuaries may do something useful, something he learned through pushback. Here we have our first anthropological datapoint: he didn't learn this by asking himself "Is there, after all, some kind of social utility in knowing how long someone is likely to live? In an advanced economy where people aren't working from the first moment they're capable of it until they're incapacitated or dead, might we expect such a job to exist, to create value, and to be paid accordingly?" No, what happened according to Graeber, is that people who read his claims responded and set him straight.

That drives a lot of the empirical research in the book. He cites some surveys, which show that many workers feel alienated from their jobs, don't believe those jobs are worth doing, don't like their jobs, etc. But a lot more of the content in the book comes from a novel source: He set up an email address people could use to send him anecdotes about how much they hated their jobs, and he quotes these at length. Some of the respondents mostly focus on how their specific job is nonsensical, while a few dive into the theory.[2] What they also do, surprisingly often, is discuss their interest in radical politics and the arts. Nothing wrong with liking either, but surely after the tenth person told a story about living on a commune or partying with anarcho-syndicalists, Graeber should have wondered if this is a truly representative audience.

It's possible that there's a large cohort of people who define themselves by what they do with their time, but who also have interests that aren't very lucrative. Of course these people will feel alienated! There are people right now who are miserable at work because they'd much rather be hiking, playing video games, getting into Wikipedia edit wars, etc. That doesn't make their jobs fake. It just makes them jobs, i.e. the kind of thing you have to do for money because the things that you enjoy doing get done for free. That's not some kind of social catastrophe, just a reflection of the fact that the modern economy has many goods and services that are more tempting than living a life of the mind while eating rice and beans in an apartment you share with roommates in a not-especially-great neighborhood. The way most people seem to resolve this is some combination of satisficing on money and then optimizing for meaning (i.e. find the most inspiring or least time-consuming job you can that hits some required income threshold) or optimize for income and expect consumption to offset the cost of not really enjoying the job. If you happen to genuinely enjoy something that pays very well, you don't have this problem, but most people are not so lucky.

But Graeber is not just talking about people disliking their jobs in general, even if that seems to motivate a lot of the people who provided him with their stories. Instead, it's about these jobs being utterly meaningless. He presents an economic theory for how this happens, connecting it to the medieval practice of creating face-saving make-work jobs for talentless aristocrats, like a master of the horse or a lady of the bedchamber. In Graeber's telling, a monarch who mostly gets paid taxes in kind has a calorie surplus, and needs to spend it by feeding assorted lackeys and hangers-on. And the modern rich are much richer than medieval rulers, while human vanity is a constant. So of course they hire a lot more such lackeys, right?

No! Of course not, because economic growth makes stuff cheap and time precious. And unstable political institutions also make safety precious. If there's a cheap way to keep lots of allies handy, in case they're necessary, and to keep enemies close, in case they want to try something, why not?[3] That doesn't apply today; Shantanu Narayen doesn't have to worry that the illegitimate son of an Adobe cofounder is going to sail to San Jose, mercenaries in tow, and depose him. And even if he wanted to be surrounded by loyal lackeys, it would be a lot more expensive than it used to be. A country where the default is subsistence, and a small elite extracts 10-30% of output, that elite can pay a lot of people to just hang around. But the richest Americans are, in their best years, capturing far less than 1% of GDP. People are just a lot more expensive than they used to be. And there are easier ways to signal wealth than wasteful hiring. Buy a $10m house, and you've made it obvious that you're rich. Spend $10m on servants instead, and you've established that you're rich and very weird.

This idea that rich people create fake jobs in order to have an impressive-looking number of economic dependents runs into a few other problems. The anecdotes that Graeber cites can be divided into two categories: people whose jobs illustrate why a particular field is not known for its efficiency, and jobs that are executed ruthlessly but that Graeber believes add negative value to society. In the first case, one of his examples—one which he explicitly cites as evidence that these jobs are not just in the public sector—is an IT consultant who is tasked with managing an incredibly baroque process for allowing workers and the client organization to change offices. That parent organization is the German military. Do these rules come from a company arbitrarily trying to inconvenience its customers, or from an unaccountable rules-maker somewhere in the German government? Another example came from someone who worked in publishing, also a field where it's historically hard to make much money, in part because people really enjoy the glamour of the industry. The fake job in question is basically a contribution to that glamour: a receptionist who doesn't have much work to do. But this could end up being a money-saving proposition if the company is able to attract workers, and pay them less, by treating the presence of an assistant as a perk. The big wealth-generating industries tend to be intolerant of this kind of thing; when competition is fierce, it's hard to justify arbitrarily accepting low margins, and when some roles require long hours, it's demoralizing to employ people who are visibly going to shirk.[4] Graber suspects that these jobs exist to keep people convinced that they have to work all the time, and spend a lot, until they die. It would be very convenient for any given rich person if every other rich person were in on this conspiracy, but the conspiracy also requires these rich people to keep it going by spending some of their riches giving other people meaningless employment. It's a perpetual motion machine: at best, if this employer monopolized all private industry, they'd be able to recapture the after-tax income of the people doing make-work jobs, but even in that extreme example, an easier way for that rich person to maximize their money would be to keep it instead of hiring someone.

The reasoning isn't just wonky for fake jobs. Graeber also plays a bit fast and loose with his discussion of what he considers real jobs. One interesting side point is one job he cites as indispensable: public transportation workers can, indeed, shut some cities down if they decide not to work. But this is not a characteristic of the job, but of how employment is structured: if all the workers are declining to show up at once, the term is a "strike," and their employer can't just swap them for someone else. There are plenty of people who would do these jobs, at their current pay, if that were an option, so the ability to paralyze a city like this is a function of unions, not of the job itself. This is one reason people stick around in these roles for almost three times as long as they do in typical private sector jobs. (Elsewhere, he talks about a bureaucratic tangle involving his efforts to get a carpenter to fix his office bookshelf. Apparently part of this violated some "work rules," so the carpenter who showed up couldn't work. Private sector employers don't have a very strong incentive to construct elaborate rules about which tasks people can't do, unless they're worried about legal liability or that's something they trade to unions in exchange for a concession somewhere else. It is, again, very hard to blame capitalism for something that capitalists love to complain about.)

Graeber points out, correctly, that the person who cooks your food makes a lot less money than the person who decides which appetizer will get a limited-time discount; the person driving your Uber makes less than the person who designed the frontend for the app. There is a rough correlation between how real-world a job is and how little it pays, or, more kindly, a correlation between how abstract a job is and how well it pays. But some of this is because those more tangible jobs strike us as legitimate jobs because they've existed for a long time. You're descended from a long line of people who all knew what a "farmer" was but would be befuddled by the concept of a Director of Social Media Strategy. Those old jobs are more prone to partial or complete automation—if it's something people did in a pre-literate society, you don't need college or even high school to do that kind of work today. You may need other things, like patience, willingness to literally get your hands dirty, physical strength and endurance, etc., but these traits are widely-distributed enough that there are many people who are willing to do such jobs, so the price of a replacement is low. Tax lawyers, distributed computing experts, and exotic derivatives traders are newer professions. They're also jobs that require substantial training, including apprenticeship-style training that's only available from working alongside other expensive workers. And they require substantial local knowledge—doing these jobs usefully in a given organization requires extensive information about how the organization already operates. All of these features mean that the people who do these jobs are hard to replace, and they get a wage premium in return.[5] If all the world's bus drivers disappeared, it would be enormously inconvenient for a while, but the net result would be some labor shifting, some retraining, and a status quo similar to the past one. If all of the people who've spent the last decade obsessing over video compression algorithms, rapid matrix multiplication, ways to describe certain kinds of income as capital gains, etc. disappeared, it would take much longer to replace them, and the replacements would be more visibly different.

But do we actually need those jobs? At one level, the answer is no: we got along fine for most of human history with zero people doing this work. We could get along that way again if we needed to. But "we," collectively, do not seem especially interested in merely getting along—we generally want to get more. And that tends to happen through accumulating productive physical and intellectual capital and relentlessly specializing. That process produces specialists whose work has a highly tangential relationship to the finished product; you can't point to any specific feature of an iPhone that can be credited to anyone in Apple's HR, marketing, or finance departments. But those jobs are all complementary to the work that does create such nice devices, and there's a long history of companies deciding that some category of job is completely pointless and then grudgingly admitting that there's a reason the competition had it. In fact, many of these jobs act as interfaces for other complex systems; Apple has an HR team as an alternative to every manager carefully memorizing the labor laws of every jurisdiction in which the company operates, they have marketers because a good product the right customer never hears about is only hypothetically good, they have finance teams because capital markets are complex and it's hard to build a good phone if the designers' paychecks sometimes bounce.

And: some of these jobs may be fake, or fake-ish. Some may be the result of corporate empire-building, or might exist to help create and sell products that customers would be better-off not buying. That's always a valid suspicion! But it's also valid to ask: when you encounter other people's behavior, and find it surprising, is it more likely that you noticed something, with only a few moments of thought, they've missed for their entire career? Or that they've figured out something you don't understand after years of work? Do they happen to have preferences that you don't share? (There are plenty of products that I don't think people should consume, but the more products make you feel that way, the worse off you'd be in a system where all this stuff is up for a vote—put it up to a plebiscite, and history books will lose to sitcoms every time.)

In a sense, the book is a work of pathological optimism about the capitalist system. Graeber estimates that roughly half of all work fits his fake job categorization, which implies that the economy's productive capacity is roughly twice the output we actually get. It would be a pretty big deal if this were true: we could have a lot more leisure, and a lot more stuff. And there are people motivated to make this happen! The strongest single argument against Graeber's book is: did anyone at Bain or McKinsey read it? What about KKR and Blackstone? Did any owner of any business of any size read it and say: "What a sec! That's right! Most jobs really are fake jobs designed to make rich people feel good about themselves. But what makes me feel good about myself is having more money, so I'm going to start firing people and keeping the money." The closest you can get is Elon Musk at Twitter, which did reveal that the service could keep running, and ship new features, with a lower headcount. But that happened at a company that was notoriously inefficient, for years, and one where it's widely-agreed that they unnecessarily blew their lead in short-form many-to-many communications, and took too long to get into messaging. If there's one large-scale example of the thesis playing out, and the thesis holds that it's describing a ubiquitous phenomenon, something doesn't add up.

It’s just sad to go through life dismissing everything you can’t immediately explain as some sort of emperor’s new clothes phenomenon that everyone else is too oversocialized to call out. Some fraction of the time, that’s true; there really are pointless jobs that nobody would miss if they were gone. But these jobs fall into two categories:

  1. Minor roles in big institutions, where there’s too much inertia to ask if the job really needs to stick around. There was a story a while ago about someone who stopped showing up for work for at least six years, possibly fourteen, before he was caught. He was a government employee in Spain, and the story was an org chart mix-up: “the water board had believed García was the responsibility of the city council for most of the period of his employment, while the city council thought he was working for the water board.”
  2. Obviously fake roles given out as a favor. If the owner of a small business gives his son-in-law a job with “strategy” in the title, that’s just a nice way for the son-in-law in question to save face while living off his wife’s allowance.

What these stories have in common is not just that they nominally validate Graeber’s thesis, but that they refute his approach. These case studies are both interesting. You could get a respectable short story out of either of them, possibly a novel, maybe some Oscar-bait movie! There’s tension, there’s humor, there are all-too-human motivations and funny misunderstandings.

They’re not just stories of random interactions leading to weird effects. They have specific causes, and those causes are illuminating. For example: governments try to specialize in things that the private sector doesn’t do, or doesn’t do optimally. One category of those things: cases where the inputs are easier to measure than the outputs. Which question leads to a more factual, verifiable answer: “How much did the city’s budget for parks increase in the last year?” or “How much more beautiful is the park than last year?” It’s good and worthwhile to have a beautiful city, and that’s an investment that we can collectively make and collectively feel pride in, but I would not want to be the person creating a rigorous rubric for rating how well it’s performed. In the small business context, there’s another mix of incentives: small companies are partly profit-maximizing entities, but the ones that do that well cease to be “small” in short order. So you can assume that the CEO of a twenty-person company, who was running that same company at roughly that headcount a decade ago, is motivated by something other than maximizing net worth.[6] One great reason to keep running a not-optimally-profitable business is ego. It’ll keep you going, and probably lead to a better experience for customers and employees. And one element of that is, sometimes, doling out fake jobs because it’s easier to pay someone money than to have a difficult conversation with them.[7]

The world is full of mysterious economic phenomena. You should expect it to be! A world where you can consider a random career or business for a few seconds and instantly identify a way to double its efficiency is a much weirder world than one where those mysteries tend to have satisfying answers. It's also a world whose sizable and growing aggregate wealth is a big mystery: if we're wasting more and more of our time, shouldn't we be getting poorer? Subjecting Bullshit Jobs to any serious intellectual scrutiny, whether asking if the broad theory makes sense or if the evidence for it is reliable, reveals that there isn't much substance there. It's a book that would read better if more declarative sentences were turned into question marks, and if it spent at least a little time asking employers "what is this person's job for?" Some time gets wasted, for a while; sometimes the 95% of your time that you waste waiting for the phone to ring or the inbox to ding is worth less than the 5% of someone else's time that you're saving by doing this work. And sometimes, the existence of a job, whether actuary or anthropologist, only makes sense in light of a deeper appreciation for the complexity of the modern economy. The world is full of people who are doing things that don't make sense to you, but do make sense to them. And if you start with that as a premise and keep pulling on threads, you end up with a view of the world that makes sense, too.

  1. In the interest of keeping things readable, and because swearing a lot is low-class, I'll be freely using a variety of euphemisms for Graeber's titular concept. ↩︎

  2. He tends to present both the facts and theory uncritically, even when that's just not true. One correspondent claimed that the existence of strippers is demoralizing because it "tells the vast majority of its female population that they are worth more dancing on boxes between the ages of eighteen and twenty-five than they will be at any subsequent point in their lives." (For what it's worth, their median pay is apparently $46k per year, below the BLS's median wage numbers for women in every age bracket past that.) That same person also claims that we could solve climate change solely by directing the revenue spent at strip clubs. That's about $8bn/year, so either a) this correspondent was very early to hearing the good news about stratospheric aerosol injection, or b) nobody Googled the numbers, or even speculated about the orders of magnitude. Another correspondent claims that campus jobs are part of a conspiracy to load students up with debt so they'll be easier to control in the future. Neither that correspondent nor Graeber wonders how paying people money while they attend college induces them to borrow more than they otherwise would. Intuitively, it would seem to go in the other direction, unless every student is somehow operating like a large-cap CFO and always trying to keep their leverage at some multiple of current-year cash flow. ↩︎

  3. There's some more dubious history here, where Graeber claims that pre-modern economies could do just fine when a large share of the male labor force went off to war. This overestimates the size of pre-modern militaries—in the Hundred Years War, France was fielding an army of 10-25,000 at a time when its population was 14 million. It's also not true that this wasn't disruptive: extended military campaigns were notoriously destabilizing, and are a classic feature of the decline and fall of empires. The closest exception might be steppe nomads, who a) had high complementarity between the skills they used to fight wars and the skills they used between wars, b) were typically poorer than their opponents, and c) tended to have small populations, unless they conquered new territory and settled there, rendering a sort of market verdict that this lifestyle had less appeal than a more sedentary one. ↩︎

  4. I've had a job like that: I worked one summer at a warehouse, where my job was to look at the scale when trucks showed up, look at the scale when they left, write down both numbers (important!), and answer the phone. My boss could easily have done this himself, but the business required a fair amount of phone-based wheeling and dealing. If the phone was ringing and he wasn't answering, money would steadily evaporate. So he outsourced the interruption-prone part of the job—also the part of the job that could be done badly but couldn't be done uniquely well—to me. I'd usually bring a book and the Wall Street Journal, and as long as I didn't miss any incoming trucks, that was fine. But the purpose of that job wasn't to make a great first impression on visitors. And I would have been slightly more productive if I'd brought along fewer distractions. To the extent that the job was "work for your entire shift," it was a mostly fake job. To the extent that it was "be here during your shift, and do the following straightforward but mission-critical things," it wasn't. ↩︎

  5. One minor piece of evidence for this is that people doing these kinds of jobs tend to see their incomes rise quite fast early in their career, and to take a long time to level out. Once a bricklayer or janitor has shown that they can show up on time, and do the job as assigned, there's just less room for them to double or 10x the economic value of their output, even if they are improving for a long time. And in these jobs, it's much more straightforward to replace N high-skill people with >N low-skill people. The best workers will still capture some of their upside, but less. And that's a market signal, too: if you're good at the meta-skill of becoming more skilled as a janitor, you can apply that meta-skill to a domain where getting better at the job has a bigger impact. ↩︎

  6. Especially now, when the world of small businesses is crawling with search funds that want a deal. It’s a tough time to be a search fund operator, but a great time to be a business owner thinking about when to retire. ↩︎

  7. There's also a tax angle: the cost of the fake job comes out of pretax income, and then the income from that job accrues to someone who is probably at a lower tax bracket than the head of the company. ↩︎

A Word From Our Sponsors

Still running payroll the old way?

Warp is built for founders and startups who have less time on their hands. Warp bundles payroll, compliance, and benefits into a single dashboard, automating annoying registrations that come with hiring across states or offshore.

Unlike bloated platforms pushing unnecessary features like applicant tracking systems and performance reviews for your employees, Warp is laser-focused on the essentials. Get in, get it done, get out.

Say goodbye to endless payroll reminders and cryptic notices from tax departments. Reclaim your time and focus on what truly matters – growing your startup.

Get started now at joinwarp.com/thediff.


Sanction Economics

The US is considering stricter sanctions against Russian oil exports, but is worried about the inflationary impact. The calculus is pretty straightforward:

  1. A price cap doesn't much impact the supply of oil, because Russia still has to sell it if they can. But if that price cap is imperfectly enforced, buyers who violate sanctions simultaneously pay less than the prevailing global price for oil and more than Russia would get elsewhere. So a price cap, over time, becomes an energy subsidy for any country willing to look the other way.
  2. Oil supply is inelastic in the short term, so any sanctions that actually reduce the supply of oil will cause the price to spike. When Iran attacked Saudi oil assets in 2019, targeting one field that produced ~1m barrels/day and a facility that processed 7m barrels/day, oil prices briefly spiked 10-20% before dropping. Russia produces ~9.5m barrels per day.

One cynical answer is that the US could attempt to limit Russia's energy exports, and provide an offsetting subsidy to households. This wouldn't eliminate the inflation hit, but would make it more diffuse—if the subsidy perfectly offset the higher impact, households would shift some of their spending to energy but would be able to afford the same basket of goods. At least in the US; elsewhere, it would be more of a direct financial hit. The problem they run into is that it's hard to punish producers of some good without impacting consumers of that good, and the world still consumes a lot fo oil.


In Friday's issue ($), we briefly discussed the question of exit strategies for the CEOs of companies with an overpriced stock. The profit-maximizing choice is to exchange that stock for something of more value, with obvious candidates being 1) the cash you get for selling shares at their current market price, and 2) a more viable business that's willing to sell for stock. The advantage of the first option is that it's more certain, but the disadvantage is that it's a strong signal that management thinks the stock is too expensive. The second option implies that, but it also makes plenty of sense for a company with a fairly-priced stock to prefer share-based acquisitions, both to defer taxes for the seller and to align the seller's incentives with the rest of the company.

So, Trump Media and Technology Group is acquiring a small media technology company, WorldConnect, and paying for it with stock ($). This is an extra-special edge case: TMTG is worth so much because it's a meme stock on fiat rails, representing the fortunes of someone who is still a popular social media topic. and that fandom can extend past retail investors (and voters!) to include business owners. Getting someone to sell their company in exchange for overpriced stock is partly a matter of convincing them it's fairly priced, and if they're inclined to like it regardless that's an easier task.

Building Complements

ElevenLabs is an AI company whose core product converts text into voice, for arbitrary voices. You've experienced this if you've ever clicked the "play" button on the audio version of a Diff piece. Their most recent product launch is a wonderful case study in commoditizing the complement ($, Diff): they're offering a tool that isolates voice content from other audio. This is a generally useful thing to have, and there are certainly plenty of applications. But one of those is providing clean audio files to companies that train AI text-to-speech models.

So it's a nice reminder that commoditize-the-complement is a viable strategy. But it's also a good reminder of why that matters: AI business models don't just challenge the immediate use cases; they also change the economics of adjacent products. If ElevenLabs can dominate text-to-voice (still very much an open question) it's in their interest to make voice samples as abundant and cheap to acquire as possible, and that means making free or cheap tools that obsolesce existing products.

Shares and Votes

The standard way to set up a company is that it's controlled by its owners, who vote for a board, which selects an executive, who then runs the business. One feature of this is that, by default, there's a 1:1 correspondence between economic ownership of equity and voting control over the company. That's not striclty required, of course, and last night Paramount agreed to a new majority owner, who paid a 53% premium for voting shares with the same economic interest as the non-voting stock. If you look at a business whose shares have multiple classes with different numbers of votes, one question that might come to mind is: could I create a synthetic version of this? If you're a retail investor, you might think of buying the stock and writing call options that neutralize most of your exposure, but typically retail investors aren't swinging contentious shareholder votes, and institutions have more tools: they can borrow stock and vote it, or buy shares and simultaneously set up a swap transaction that offsets this exposure. Or, to cover your tracks, you might ask a friend to engage in these maneuevers on your behalf. That's what hedge fund Politan Capital Management has accused a shareholder of Masimo of doing: a broker voted 9.9% of the shares outstanding, despite controlling fewer shares than that, apparently by temporarily borrowing the shares. The mechanics of stock lending make it hard to avoid this possibility, since there will always be room to 1) temporarily acquire votes, and 2) have some semi-plausible excuse for this. There was a company, the Shareholder Vote Exchange, which tried to turn this into a market, but shut down a few months ago ($, WSJ). Trading votes separately from stock sounds dubious, but if it's going to happen anyway, it's better for it to happen at a known price and with more transparency by default.

Quality Problems

The WSJ highlights the decline in quality among small-cap stocks ($, WSJ). The Diff has previously covered this: years ago, someone carefully trawling through small-cap stocks would find plenty of companies with good economics and decent management that had happened to be unlucky, or unnoticed. There was a lot of money in noticing, though, and that led a lot of these companies to either grow into larger businesses or get acquired by private equity firms and competitors. The Journal adds that this coincided with the rise of index investing: a Russell 2000 tracking fund doesn't have an explicit mandate to care about whether it's a portfolio of mid-sized manufacturers or a portfolio of money-losing biotech companies, and if that flow of funds pushes stock prices above their private market value, there's a strong incentive for these companies to go public. It's a good reminder for investors that counterparties' behavior is informative. If a stock is cheap, it's good to know why. And similarly, if one class of companies is likely to go public and another is likely to go, and stay, private, it's good to know why.

Diff Jobs

Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.