A good bluntly realistic model of careers is that you start out being mentally quick and able to work long hours, and over time you gradually lose both of these traits but see your wages go up. Which is quite unfair. An equally valid model is that you start out knowing roughly nothing and nobody—a college graduate's friend group is mostly fellow recent graduates, and while they've learned a lot in school there are plenty of practical gaps—CS graduates tend to over-index on things like implementing a linked list versus being able to write a cogent git commit message, and finance/economics graduates can tell you all about supply and demand curves but can't necessarily update a discounted cash flow model based on new margin assumptions. That setup is fine—in most jobs, there's no particular time allocated to learning the theoretical parts. People will just assume that you know them and be annoyed if you reveal that you don't.

So over the course of a career, one means of being valuable is getting weaker—the 100-hour weeks are harder in your 30s and 40s, and mental reaction time slows down, too—but since workers' wages tend to peak in their 50s, empirically the compounding effects of knowledge and personal networks more than offset this.

That offsetting growth comes from two sources that sound like they're opposites but that actually have some deep similarities. One is that as people age, they accumulate more loose social ties, and those ties are relevant to what they do. Someone who's worked in investing for a long time will have a roster of friends whose opinions they respect on specific topics, like the mechanics of ETF construction, the best way for an American to tax-optimize investments in Europe, or how to handicap the odds when one of your portfolio companies faces a big lawsuit. In other fields, the specifics of expertise vary, but the general phenomenon holds—jobs have turnover, many jobs involve working with outside organizations, so over time the number of people who you vaguely know will keep on increasing.

But people also accumulate facts, and get better at pattern-matching. This is especially useful when an industry goes through a long cycle: in 2022, there were professional investors with a decade of tech investing experience who had literally never experienced a bear market that didn't resolve itself in a quarter or two. And, in fact, they'd experienced what looked like a literally apocalyptic bear market in 2020, and saw that the time gap between "the world is ending" and "stocks are now at a record high" was about five months. Someone a bit more experienced, who remembered either the long painful grind-down from 2000-2002, or that the low point for US equities during the financial crisis was months after the spectacular headlines—the market started moving up in 2009 when the flow of bad news slowed, not so much because there was good news.

This is a process that can be optimized, by recognizing inevitable losses and maximizing compounding gains. On the inevitable-losses side, any job where the critical skill is bursts of effort, whether occurring over a few days or a split-second, is a job that will tend to get harder with age. There can be a period where the ancillary skills offset the decline in raw ability—a 30-year-old trader is statistically past their peak reaction time, but knows more than they did at 25, and has probably gotten a few dumb mistakes out of the way. But eventually, reaction time becomes the limiting factor, and making a better decision that happens a little too late isn’t good enough.¹ Eventually, that doesn't work; there are some great jobs that are optimal for twentysomethings and require really extraordinary efforts to do with another decade or two of aging.

But that time period, when fluid intelligence is rising and then slowly declining, is also a time when it's possible to expand networks of personal connections and of knowledge. That latter process is partly building up a roster of facts, but it also means finding a library of analogies—the more situations you've been through or read about, the faster you can come up with a relevant analogy. And the more available analogies you have, the more precisely you can delineate what's the same and what's different.

Two actionable pieces of advice are about increasing the space of relevant connections:

1. Do case studies a bit outside of your exact job function—if you're a software engineer, learn something about product management; if you're in equities, learn a bit about how credit people think. You won't necessarily apply this directly, but it will inform the work that you actually do (sometimes an equity investor looking at a deep value situation can save a lot of time by noticing that the company's bonds trade at 50 cents on the dollar—there's a chance that the bonds are a better risk-adjusted deal and a potential indicator that the company's problems are deeper than they look.)

2. Similarly, try to have a network that skews a little bit away from what you work on. The more senior you get in your career, the less problems are framed as tasks. "Do X (which fixes Y)," becomes "Fix Y," with the specifics left as an exercise for the reader.

Your job title may never have "manager" in it, and almost certainly won't include something like "research librarian," but over time that's a growing share of the value that experienced workers add—there are many problems they can solve, and many more where they can find a good shortcut to a solution, either by remembering a way it's been solved before or knowing exactly who's suited to solving it. And the ability to do that only peaks when you either run out of social capital—remember, the more elite your job the more likely it is that you'll keep bumping into the same elites!—or start actively forgetting things. That takes a long time. And in the meantime, networks and knowledge compound.

Read More in The Diff

The Diff has talked about this kind of question around career paths in a few ways before:

• We’ve looked at the middle-income trap in tech ($).

• Some careers are unusually stable, but if they’re more predictable than average that risk has to go somewhere.

One particular way to both expand your knowledge network and your personal network is to write, but it’s mostly hard to do that for a living. (This piece was written a long time ago, and newsletter platforms have made it easier—but they’ve also made it more competitive! People will always accept an economic hit in exchange for a high-status job, so to make the economics work you need a way to translate that money back into status.)

Join the discussion!

1  In a more direct way, the inevitable memorization that comes with practice can be a good temporary substitute for thinking quickly. Practicing an intellectual skill is partly a process of subconscious memoization, replacing relatively expensive computation with cheaper recall.