October 2023One of the most important things I didn't understand about the worldwhen I was a child is the degree to which the returns for performanceare superlinear.Teachers and coaches implicitly told us the returns were linear."You get out," I heard a thousand times, "what you put in." Theymeant well, but this is rarely true. If your product is only halfas good as your competitor's, you don't get half as many customers.You get no customers, and you go out of business.It's obviously true that the returns for performance are superlinearin business. Some think this is a flaw of capitalism, and that ifwe changed the rules it would stop being true. But superlinearreturns for performance are a feature of the world, not an artifactof rules we've invented. We see the same pattern in fame, power,military victories, knowledge, and even benefit to humanity. In allof these, the rich get richer.[1]You can't understand the world without understanding the conceptof superlinear returns. And if you're ambitious you definitelyshould, because this will be the wave you surf on.It may seem as if there are a lot of different situations withsuperlinear returns, but as far as I can tell they reduce to twofundamental causes: exponential growth and thresholds.The most obvious case of superlinear returns is when you're workingon something that grows exponentially. For example, growing bacterialcultures. When they grow at all, they grow exponentially. But they'retricky to grow. Which means the difference in outcome between someonewho's adept at it and someone who's not is very great.Startups can also grow exponentially, and we see the same patternthere. Some manage to achieve high growth rates. Most don't. Andas a result you get qualitatively different outcomes: the companieswith high growth rates tend to become immensely valuable, while theones with lower growth rates may not even survive.Y Combinator encourages founders to focus on growth rate ratherthan absolute numbers. It prevents them from being discouraged earlyon, when the absolute numbers are still low. It also helps themdecide what to focus on: you can use growth rate as a compass totell you how to evolve the company. But the main advantage is thatby focusing on growth rate you tend to get something that growsexponentially.YC doesn't explicitly tell founders that with growth rate "you getout what you put in," but it's not far from the truth. And if growthrate were proportional to performance, then the reward for performancep over time t would be proportional to pt.Even after decades of thinking about this, I find that sentencestartling.Whenever how well you do depends on how well you've done, you'llget exponential growth. But neither our DNA nor our customs prepareus for it. No one finds exponential growth natural; every child issurprised, the first time they hear it, by the story of the man whoasks the king for a single grain of rice the first day and doublethe amount each successive day.What we don't understand naturally we develop customs to deal with,but we don't have many customs about exponential growth either,because there have been so few instances of it in human history.In principle herding should have been one: the more animals youhad, the more offspring they'd have. But in practice grazing landwas the limiting factor, and there was no plan for growing thatexponentially.Or more precisely, no generally applicable plan. There was a wayto grow one's territory exponentially: by conquest. The more territoryyou control, the more powerful your army becomes, and the easierit is to conquer new territory. This is why history is full ofempires. But so few people created or ran empires that theirexperiences didn't affect customs very much. The emperor was aremote and terrifying figure, not a source of lessons one could usein one's own life.The most common case of exponential growth in preindustrial timeswas probably scholarship. The more you know, the easier it is tolearn new things. The result, then as now, was that some peoplewere startlingly more knowledgeable than the rest about certaintopics. But this didn't affect customs much either. Although empiresof ideas can overlap and there can thus be far more emperors, inpreindustrial times this type of empire had little practical effect.[2]That has changed in the last few centuries. Now the emperors ofideas can design bombs that defeat the emperors of territory. Butthis phenomenon is still so new that we haven't fully assimilatedit. Few even of the participants realize they're benefitting fromexponential growth or ask what they can learn from other instancesof it.The other source of superlinear returns is embodied in the expression"winner take all." In a sports match the relationship betweenperformance and return is a step function: the winning team getsone win whether they do much better or just slightly better.[3]The source of the step function is not competition per se, however.It's that there are thresholds in the outcome. You don't needcompetition to get those. There can be thresholds in situationswhere you're the only participant, like proving a theorem or hittinga target.It's remarkable how often a situation with one source of superlinearreturns also has the other. Crossing thresholds leads to exponentialgrowth: the winning side in a battle usually suffers less damage,which makes them more likely to win in the future. And exponentialgrowth helps you cross thresholds: in a market with network effects,a company that grows fast enough can shut out potential competitors.Fame is an interesting example of a phenomenon that combines bothsources of superlinear returns. Fame grows exponentially becauseexisting fans bring you new ones. But the fundamental reason it'sso concentrated is thresholds: there's only so much room on theA-list in the average person's head.The most important case combining both sources of superlinear returnsmay be learning. Knowledge grows exponentially, but there are alsothresholds in it. Learning to ride a bicycle, for example. Some ofthese thresholds are akin to machine tools: once you learn to read,you're able to learn anything else much faster. But the most importantthresholds of all are those representing new discoveries. Knowledgeseems to be fractal in the sense that if you push hard at theboundary of one area of knowledge, you sometimes discover a wholenew field. And if you do, you get first crack at all the newdiscoveries to be made in it. Newton did this, and so did Durer andDarwin.Are there general rules for finding situations with superlinearreturns? The most obvious one is to seek work that compounds.There are two ways work can compound. It can compound directly, inthe sense that doing well in one cycle causes you to do better inthe next. That happens for example when you're building infrastructure,or growing an audience or brand. Or work can compound by teachingyou, since learning compounds. This second case is an interestingone because you may feel you're doing badly as it's happening. Youmay be failing to achieve your immediate goal. But if you're learninga lot, then you're getting exponential growth nonetheless.This is one reason Silicon Valley is so tolerant of failure. Peoplein Silicon Valley aren't blindly tolerant of failure. They'll onlycontinue to bet on you if you're learning from your failures. Butif you are, you are in fact a good bet: maybe your company didn'tgrow the way you wanted, but you yourself have, and that shouldyield results eventually.Indeed, the forms of exponential growth that don't consist oflearning are so often intermixed with it that we should probablytreat this as the rule rather than the exception. Which yieldsanother heuristic: always be learning. If you're not learning,you're probably not on a path that leads to superlinear returns.But don't overoptimize what you're learning. Don't limit yourselfto learning things that are already known to be valuable. You'relearning; you don't know for sure yet what's going to be valuable,and if you're too strict you'll lop off the outliers.What about step functions? Are there also useful heuristics of theform "seek thresholds" or "seek competition?" Here the situationis trickier. The existence of a threshold doesn't guarantee thegame will be worth playing. If you play a round of Russian roulette,you'll be in a situation with a threshold, certainly, but in thebest case you're no better off. "Seek competition" is similarlyuseless; what if the prize isn't worth competing for? Sufficientlyfast exponential growth guarantees both the shape and magnitude ofthe return curve — because something that grows fast enough willgrow big even if it's trivially small at first — but thresholdsonly guarantee the shape.[4]A principle for taking advantage of thresholds has to include atest to ensure the game is worth playing. Here's one that does: ifyou come across something that's mediocre yet still popular, itcould be a good idea to replace it. For example, if a company makesa product that people dislike yet still buy, then presumably they'dbuy a better alternative if you made one.[5]It would be great if there were a way to find promising intellectualthresholds. Is there a way to tell which questions have whole newfields beyond them? I doubt we could ever predict this with certainty,but the prize is so valuable that it would be useful to havepredictors that were even a little better than random, and there'shope of finding those. We can to some degree predict when a researchproblem isn't likely to lead to new discoveries: when it seemslegit but boring. Whereas the kind that do lead to new discoveriestend to seem very mystifying, but perhaps unimportant. (If theywere mystifying and obviously important, they'd be famous openquestions with lots of people already working on them.) So oneheuristic here is to be driven by curiosity rather than careerism— to give free rein to your curiosity instead of working on whatyou're supposed to.The prospect of superlinear returns for performance is an excitingone for the ambitious. And there's good news in this department:this territory is expanding in both directions. There are more typesof work in which you can get superlinear returns, and the returnsthemselves are growing.There are two reasons for this, though they're so closely intertwinedthat they're more like one and a half: progress in technology, andthe decreasing importance of organizations.Fifty years ago it used to be much more necessary to be part of anorganization to work on ambitious projects. It was the only way toget the resources you needed, the only way to have colleagues, andthe only way to get distribution. So in 1970 your prestige was inmost cases the prestige of the organization you belonged to. Andprestige was an accurate predictor, because if you weren't part ofan organization, you weren't likely to achieve much. There were ahandful of exceptions, most notably artists and writers, who workedalone using inexpensive tools and had their own brands. But eventhey were at the mercy of organizations for reaching audiences.[6]A world dominated by organizations damped variation in the returnsfor performance. But this world has eroded significantly just inmy lifetime. Now a lot more people can have the freedom that artistsand writers had in the 20th century. There are lots of ambitiousprojects that don't require much initial funding, and lots of newways to learn, make money, find colleagues, and reach audiences.There's still plenty of the old world left, but the rate of changehas been dramatic by historical standards. Especially consideringwhat's at stake. It's hard to imagine a more fundamental changethan one in the returns for performance.Without the damping effect of institutions, there will be morevariation in outcomes. Which doesn't imply everyone will be betteroff: people who do well will do even better, but those who do badlywill do worse. That's an important point to bear in mind. Exposingoneself to superlinear returns is not for everyone. Most peoplewill be better off as part of the pool. So who should shoot forsuperlinear returns? Ambitious people of two types: those who knowthey're so good that they'll be net ahead in a world with highervariation, and those, particularly the young, who can afford torisk trying it to find out.[7]The switch away from institutions won't simply be an exodus of theircurrent inhabitants. Many of the new winners will be people they'dnever have let in. So the resulting democratization of opportunitywill be both greater and more authentic than any tame intramuralversion the institutions themselves might have cooked up.Not everyone is happy about this great unlocking of ambition. Itthreatens some vested interests and contradicts some ideologies. [8]But if you're an ambitious individual it's good news for you.How should you take advantage of it?The most obvious way to take advantage of superlinear returns forperformance is by doing exceptionally good work. At the far end ofthe curve, incremental effort is a bargain. All the more so becausethere's less competition at the far end — and not just for theobvious reason that it's hard to do something exceptionally well,but also because people find the prospect so intimidating that feweven try. Which means it's not just a bargain to do exceptionalwork, but a bargain even to try to.There are many variables that affect how good your work is, and ifyou want to be an outlier you need to get nearly all of them right.For example, to do something exceptionally well, you have to beinterested in it. Mere diligence is not enough. So in a world withsuperlinear returns, it's even more valuable to know what you'reinterested in, and to find ways to work on it.[9]It will also beimportant to choose work that suits your circumstances. For example,if there's a kind of work that inherently requires a huge expenditureof time and energy, it will be increasingly valuable to do it whenyou're young and don't yet have children.There's a surprising amount of technique to doing great work.It's not just a matter of trying hard. I'm going to take a shotgiving a recipe in one paragraph.Choose work you have a natural aptitude for and a deep interest in.Develop a habit of working on your own projects; it doesn't matterwhat they are so long as you find them excitingly ambitious. Workas hard as you can without burning out, and this will eventuallybring you to one of the frontiers of knowledge. These look smoothfrom a distance, but up close they're full of gaps. Notice andexplore such gaps, and if you're lucky one will expand into a wholenew field. Take as much risk as you can afford; if you're not failingoccasionally you're probably being too conservative. Seek out thebest colleagues. Develop good taste and learn from the best examples.Be honest, especially with yourself. Exercise and eat and sleepwell and avoid the more dangerous drugs. When in doubt, follow yourcuriosity. It never lies, and it knows more than you do about what'sworth paying attention to.[10]And there is of course one other thing you need: to be lucky. Luckis always a factor, but it's even more of a factor when you'reworking on your own rather than as part of an organization. Andthough there are some valid aphorisms about luck being wherepreparedness meets opportunity and so on, there's also a componentof true chance that you can't do anything about. The solution isto take multiple shots. Which is another reason to start takingrisks early.The best example of a field with superlinear returns is probablyscience. It has exponential growth, in the form of learning, combinedwith thresholds at the extreme edge of performance — literally atthe limits of knowledge.The result has been a level of inequality in scientific discoverythat makes the wealth inequality of even the most stratified societiesseem mild by comparison. Newton's discoveries were arguably greaterthan all his contemporaries' combined.[11]This point may seem obvious, but it might be just as well to spellit out. Superlinear returns imply inequality. The steeper the returncurve, the greater the variation in outcomes.In fact, the correlation between superlinear returns and inequalityis so strong that it yields another heuristic for finding work ofthis type: look for fields where a few big winners outperformeveryone else. A kind of work where everyone does about the sameis unlikely to be one with superlinear returns.What are fields where a few big winners outperform everyone else?Here are some obvious ones: sports, politics, art, music, acting,directing, writing, math, science, starting companies, and investing.In sports the phenomenon is due to externally imposed thresholds;you only need to be a few percent faster to win every race. Inpolitics, power grows much as it did in the days of emperors. Andin some of the other fields (including politics) success is drivenlargely by fame, which has its own source of superlinear growth.But when we exclude sports and politics and the effects of fame, aremarkable pattern emerges: the remaining list is exactly the sameas the list of fields where you have to be independent-minded tosucceed — where your ideas have to be not just correct, but novelas well.[12]This is obviously the case in science. You can't publish paperssaying things that other people have already said. But it's justas true in investing, for example. It's only useful to believe thata company will do well if most other investors don't; if everyoneelse thinks the company will do well, then its stock price willalready reflect that, and there's no room to make money.What else can we learn from these fields? In all of them you haveto put in the initial effort. Superlinear returns seem small atfirst. At this rate, you find yourself thinking, I'll never getanywhere. But because the reward curve rises so steeply at the farend, it's worth taking extraordinary measures to get there.In the startup world, the name for this principle is "do thingsthat don't scale." If you pay a ridiculous amount of attention toyour tiny initial set of customers, ideally you'll kick off exponentialgrowth by word of mouth. But this same principle applies to anythingthat grows exponentially. Learning, for example. When you firststart learning something, you feel lost. But it's worth making theinitial effort to get a toehold, because the more you learn, theeasier it will get.There's another more subtle lesson in the list of fields withsuperlinear returns: not to equate work with a job. For most of the20th century the two were identical for nearly everyone, and as aresult we've inherited a custom that equates productivity withhaving a job. Even now to most people the phrase "your work" meanstheir job. But to a writer or artist or scientist it means whateverthey're currently studying or creating. For someone like that, theirwork is something they carry with them from job to job, if theyhave jobs at all. It may be done for an employer, but it's part oftheir portfolio.It's an intimidating prospect to enter a field where a few bigwinners outperform everyone else. Some people do this deliberately,but you don't need to. If you have sufficient natural ability andyou follow your curiosity sufficiently far, you'll end up in one.Your curiosity won't let you be interested in boring questions, andinteresting questions tend to create fields with superlinear returnsif they're not already part of one.The territory of superlinear returns is by no means static. Indeed,the most extreme returns come from expanding it. So while bothambition and curiosity can get you into this territory, curiositymay be the more powerful of the two. Ambition tends to make youclimb existing peaks, but if you stick close enough to an interestingenough question, it may grow into a mountain beneath you.NotesThere's a limit to how sharply you can distinguish between effort,performance, and return, because they're not sharply distinguishedin fact. What counts as return to one person might be performanceto another. But though the borders of these concepts are blurry,they're not meaningless. I've tried to write about them as preciselyas I could without crossing into error.[1]Evolution itself is probably the most pervasive example ofsuperlinear returns for performance. But this is hard for us toempathize with because we're not the recipients; we're the returns.[2]Knowledge did of course have a practical effect before theIndustrial Revolution. The development of agriculture changed humanlife completely. But this kind of change was the result of broad,gradual improvements in technique, not the discoveries of a fewexceptionally learned people.[3]It's not mathematically correct to describe a step function assuperlinear, but a step function starting from zero works like asuperlinear function when it describes the reward curve for effortby a rational actor. If it starts at zero then the part before thestep is below any linearly increasing return, and the part afterthe step must be above the necessary return at that point or no onewould bother.[4]Seeking competition could be a good heuristic in the sense thatsome people find it motivating. It's also somewhat of a guide topromising problems, because it's a sign that other people find thempromising. But it's a very imperfect sign: often there's a clamoringcrowd chasing some problem, and they all end up being trumped bysomeone quietly working on another one.[5]Not always, though. You have to be careful with this rule. Whensomething is popular despite being mediocre, there's often a hiddenreason why. Perhaps monopoly or regulation make it hard to compete.Perhaps customers have bad taste or have broken procedures fordeciding what to buy. There are huge swathes of mediocre thingsthat exist for such reasons.[6]In my twenties I wanted to be an artist and even went to artschool to study painting. Mostly because I liked art, but a nontrivialpart of my motivation came from the fact that artists seemed leastat the mercy of organizations.[7]In principle everyone is getting superlinear returns. Learningcompounds, and everyone learns in the course of their life. But inpractice few push this kind of everyday learning to the point wherethe return curve gets really steep.[8]It's unclear exactly what advocates of "equity" mean by it.They seem to disagree among themselves. But whatever they mean isprobably at odds with a world in which institutions have less powerto control outcomes, and a handful of outliers do much better thaneveryone else.It may seem like bad luck for this concept that it arose at justthe moment when the world was shifting in the opposite direction,but I don't think this was a coincidence. I think one reason itarose now is because its adherents feel threatened by rapidlyincreasing variation in performance.[9]Corollary: Parents who pressure their kids to work on somethingprestigious, like medicine, even though they have no interest init, will be hosing them even more than they have in the past.[10]The original version of this paragraph was the first draft of"How to Do Great Work." As soon as I wrote it I realized it was a more important topic than superlinearreturns, so I paused the present essay to expand this paragraph into itsown. Practically nothing remains of the original version, becauseafter I finished "How to Do Great Work" I rewrote it based on that.[11]Before the Industrial Revolution, people who got rich usuallydid it like emperors: capturing some resource made them more powerfuland enabled them to capture more. Now it can be done like a scientist,by discovering or building something uniquely valuable. Most peoplewho get rich use a mix of the old and the new ways, but in the mostadvanced economies the ratio has shifted dramatically toward discoveryjust in the last half century.[12]It's not surprising that conventional-minded people woulddislike inequality if independent-mindedness is one of the biggestdrivers of it. But it's not simply that they don't want anyone tohave what they can't. The conventional-minded literally can't imaginewhat it's like to have novel ideas. So the whole phenomenon of greatvariation in performance seems unnatural to them, and when theyencounter it they assume it must be due to cheating or to somemalign external influence.Thanks to Trevor Blackwell, Patrick Collison, Tyler Cowen,Jessica Livingston, Harj Taggar, and Garry Tan for reading draftsof this.
