Against the Lane – A letter to the next generation

The Provocation

There is a quote that almost everyone gets wrong.

“Jack of all trades, master of none.”

It arrived in English somewhere in the early seventeenth century and has been doing institutional service ever since. The phrase has a curious double life. The speaker often means it kindly, as shorthand for someone who is experienced, versatile, and capable across domains. The listener, particularly in a professional context, hears something else: unfocused, scattered, uncommitted. The gap between what is meant and what is heard is wide enough to end a career.

The longer version, which surfaces occasionally and almost never in everyday use, ends with a small but devastating coda:

“… but oftentimes better than a master of one.”

The strange thing about the phrase is that it keeps changing its mind. It arrived, four hundred years ago, as a compliment. By the eighteenth century someone had added “master of none,” and the compliment soured into a caution. In our own century a third clause appeared, “but oftentimes better than a master of one,” and the caution flipped back into praise.¹ A proverb does not normally argue with itself across four centuries. This one does, and the direction of the argument is the tell: each age rewrote the line to match what it believed about breadth and depth. The dismissive version, the one that stuck, belonged to an age that paid specialists. The optimistic ending belongs to ours, which has just begun to pay generalists again. The sentence got wiser, in other words, at precisely the moment we needed it to.

This is, depending on your temperament, either a cosmic joke or a perfect illustration of the argument that follows. We say we admire breadth, but in practice we promote the specialist. We celebrate the polymath in retrospect and reward the narrow expert in real time.

The following is about that contradiction: where it came from, why it persists, and why the cost of forgetting it has just risen sharply.

First, I will build a case from science and mathematics and my own corner of finance: that the ideas that matter often arrive in the seams between fields, and that finance, of all places, is the worst at noticing. Next, I will argue that what breadth is worth, relative to depth, has just changed, sharply, in the space of a few years. And finally, because that is the part I actually care about, I will finish with a letter to the next generation, including my children. You are welcome to read over their shoulder. The story will involve a roulette table in Las Vegas, a mathematician, a computer hidden in his shoe, and an earpiece that kept breaking. We will get to him.

The Wrong Lesson from Darwin

The most important ideas tend to arrive at the seam between fields, not deep inside any one of them. Start with the most famous example in science.

Charles Darwin’s colleagues thought he was wasting his time.

For more than two decades after returning from his voyage on HMS Beagle in 1836, the Royal Navy survey ship that had carried him around the world for five years, Darwin spent his days on what looked, to a serious Victorian naturalist, like productive idleness. He dissected barnacles for eight years. He grew orchids. He bred pigeons. On the Origin of Species (Darwin, 1859) finally appeared more than two decades after the Beagle dropped anchor. Until then, the field had largely given up waiting for him to produce something important. He was not finished even then: in 1881, the year before he died, he published the fruit of a lifetime’s observation of earthworms (Darwin, 1881), several hundred pages that remain an authoritative source on the subject.

The conventional reading is that Darwin was patient and admirably thorough. But patience is not quite the point. These were not digressions from his real work; they were how it got done. Each apparent detour added another vantage on the same question, until a pattern emerged that no single field could have produced.

Natural selection was not a theory any single discipline would have produced. It required the cumulative weight of observation across enough different organisms that the pattern beneath them became too obvious to miss. A specialist could have written a definitive study on barnacles, or a detailed treatise on earthworm mould, but never the unifying principle that bound them all. Darwin’s breadth was not a personal quirk; it was a method, and arguably the most productive one in the history of science.

David Epstein, in Range (Epstein, 2019), made the sharper point about where breadth actually pays. In “kind” learning environments (chess, classical music, golf), where feedback is clear and the rules stable, early specialisation works. In “wicked” environments (medicine, geopolitics, frontier science, finance), generalists routinely outperform specialists. Markets are the most wicked environment humans have invented.

Breadth is not a quirk of character. In complex fields, it is how you come to know anything at all.

None of this is an argument against depth. Depth is the price of admission; without it, breadth is just dilettantism. The claim is specific, and easy to mishear: you need depth and breadth, not one instead of the other. Breadth is the half we undervalue. We reward the specialist we can see, and miss the synthesis we cannot.

What Happens When Disciplines Converge

Darwin shows what breadth does inside a single mind. The next step is stranger: what happens when two entire fields, developed independently, turn out to have been describing the same behaviour.

The most striking moment of my training was seeing the Black-Scholes equation, after what a mathematician would call a few trivial transformations, sitting side by side with Fourier’s 1822 heat equation. It was, simply, what the polymath William Whewell called consilience: two completely different fields of study converging on the exact same underlying process.

Black-Scholes is the formula that prices financial options. It earned Scholes and Merton the Nobel Prize in 1997 (Fischer Black, who died in 1995, would almost certainly have shared it), spawned an industry of derivatives trading, and underpins much of how risk is managed today. Fourier’s heat equation, derived in 1822 (Fourier, 1822), describes how warmth diffuses through a solid object. One is a cornerstone of finance. The other is a cornerstone of thermodynamics.

With a change of variables, they are the same equation.

The mathematics describing how heat spreads through a copper rod is structurally identical to the mathematics describing how uncertainty diffuses through a financial instrument over time. The three men behind the formula, Black, Scholes and Merton, did not arrive at the most influential pricing formula in modern finance by going deeper into finance. They arrived at it by reaching sideways into the physics of heat dissipation.

Black and Scholes themselves did this reduction in their original 1973 paper (Black & Scholes, 1973), and said so plainly: “The differential equation (10) is the heat-transfer equation of physics, and its solution is given by Churchill (1963, p. 155).”

This is what E.O. Wilson took up in Consilience (Wilson, 1998), building on Whewell’s 1840 coinage (Whewell, 1840). (The same man had given us the word scientist seven years earlier.) The principle is precise: when evidence from genuinely different fields converges on the same truth, the answer becomes harder to dismiss. Knowledge advances by resonance, not addition. The link between seemingly disparate fields was not a metaphor but a correspondence, and it was invisible to anyone trained in only one of the two fields.

However, while the description of heat dissipation is an accurate fundamental physical law (under certain physical conditions), the conditions required to capture the behaviour of stock prices in options valuation, specifically, continuous pricing and a stable underlying stochastic process (the random walk the maths assumes prices follow), are admittedly less stable in the real world.

Range creates the conditions for crossing fields; consilience is what you find on the other side.

The Proof by Story

Convergence can sound abstract until you watch it run through one life. The clearest proof I know, and the man promised at the start, ties a basement workshop to a roulette table to Wall Street.

Claude Shannon was the kind of figure academic culture produces about once a generation and then quietly forgets it produced. At twenty-one, his master’s thesis (Shannon, 1937) showed how Boolean algebra could be used to design electrical circuits. This insight became the foundation of every digital device built since. In 1948, at Bell Labs, he published A Mathematical Theory of Communication (Shannon, 1948) and created the field of information theory. The paper treated language, music, telegraphy and noise as mathematically related problems. It could not have been written by a pure mathematician, a pure engineer or a pure linguist. It required the seam.

This is the man a young mathematician named Edward Thorp walked in to see, one afternoon in 1960, with a system for beating blackjack. A system he would later detail in his influential book, Beat the Dealer, which briefly led casinos to change the rules of the game (Thorp, 1962).

Five minutes in Shannon’s office

Thorp was an MIT instructor at the time and needed a member of the National Academy of Sciences to sponsor his paper for publication. Shannon’s secretary warned him, when he requested the meeting, that it would be brief. Shannon did not see people. Thorp would get five minutes, perhaps less, and would do well to come prepared (Thorp, 2017).

Five minutes became several hours. Shannon, the archetypal seam-crosser, recognised almost immediately that what Thorp had brought him was not a paper about gambling. It was a paper about probability, and beneath that, a problem in physics. Could the same approach work, Shannon wanted to know, for roulette?

Nine months in a basement

What followed was nine months of nights and weekends in Shannon’s basement workshop in Winchester, Massachusetts, the two men averaging twenty hours a week between Shannon’s day job and Thorp’s teaching. They built the world’s first wearable computer: a device the size of a cigarette pack, built around twelve transistors, that used the physics of planetary motion to predict where a roulette ball would land. One operator timed the wheel using a switch hidden in his shoe. The other received the prediction through a small earpiece that played one of eight musical tones, each corresponding to a section of the wheel.

In August 1961, accompanied by their wives, Shannon and Thorp drove to Las Vegas and tested it. It worked. Their calculated edge over the house was 44 per cent, a margin so absurdly large that no casino in history had ever faced anything like it.

They did not become rich. The wires connecting the toe-switch to the computer kept breaking. The earpiece occasionally failed. And the two academics, conscious of their reputations, were visibly nervous. Certainly not an asset at a roulette table. After a few sessions they abandoned the project. The world’s most sophisticated gambling system was defeated by a faulty connection and an excess of caution.

From Las Vegas to Wall Street

Thorp turned this same probabilistic toolkit toward the markets. By 1967, he had independently derived an options-pricing formula nearly identical to the one that would later earn Scholes and Merton the Nobel Prize (Thorp, 2017). At Princeton Newport Partners, his pioneering market-neutral hedge fund, he delivered twenty years of returns without a single losing quarter between 1969 and 1988. Shannon, too, applied his mind to capital; he is reported to have achieved annual returns of roughly twenty-eight per cent over three decades, quietly outperforming almost every professional fund of the era (Poundstone, 2005).

The willingness to look foolish in an adjacent field, for long enough, to find something useful.

The throughline from Bell Labs to Las Vegas to Wall Street is not coincidence, and it is not eccentricity. It is a method.

Hofstadter in One Breath

So far the case for breadth has been historical: this is how breakthroughs have happened. There is also a simpler, harder reason. In 1931, the mathematician Kurt Gödel proved something strange and permanent: no system of rules can prove, using only its own rules, that it holds together (Gödel, 1931). To know whether the foundations are sound, you have to step outside them and look. Douglas Hofstadter built a whole book on the idea; it comes down to one line (Hofstadter, 1979): you cannot see the whole from inside it.

The lesson applies almost too neatly to finance. A risk model built only out of financial inputs is trying to prove itself using only its own rules, exactly the move Gödel showed can never quite work. The feedback loops that make markets dangerous (prices change how people behave, which changes prices) only become visible from outside the model: from psychology, from ecology, from the study of crowds. The strange loop reveals itself only when you change your vantage point.

No one in finance saw this more clearly than George Soros, who built a career on “reflexivity”: the real economy shapes financial markets, which in turn reshape the real economy, round and round (Soros, 1987).

Why Finance Is the Worst Offender

Reflexivity is exactly the kind of outside-the-system insight finance is built to ignore. Which brings the argument to its worst offender, and, I am sorry to say, my own trade. The trouble starts with a mistake about what kind of knowledge finance is.

The Quantification Trap

Finance presents itself as a domain that rewards depth. It has a complex technical vocabulary, a substantial mathematical apparatus, large and growing datasets, and a credentialing infrastructure (the CFA, CAIA, the FRM, the MBA, even the PhD) that signals domain mastery. The appearance of quantifiability is seductive. It is also, for reasons the field has been slow to acknowledge, structurally dangerous.

It is easy to forget that science and philosophy were once the same thing. Newton’s great work was titled Philosophiæ Naturalis Principia Mathematica, the mathematical principles of natural philosophy. What we now call science was then a branch of philosophy, concerned not only with how the world worked but why. Descartes was as much a philosopher as a mathematician. The division came later.

Specialisation brought enormous gains, but it also bred a quieter error: the belief that being right about one thing makes you right about another. Newton is the cautionary tale. Few have ever read the heavens more precisely, yet he lost a fortune when the South Sea bubble burst in 1720 (Odlyzko, 2019). He is said to have remarked that he could calculate the movements of the stars but not the madness of men. The line is probably invented; the lesson is not. A market is not a machine to be solved. It is a crowd of people, moved by hope and fear as much as by arithmetic, and reading it takes more than mathematics.

A line usually pinned on Einstein, though it was really the sociologist William Bruce Cameron who wrote it (Cameron, 1963), names the trap exactly:

“Not everything that can be counted counts, and not everything that counts can be counted.”

Finance has spent a century perfecting the first half and quietly losing the second.

The CFA curriculum, in its current form, runs to roughly nine thousand pages of rigorous material across three levels. It is an extraordinary intellectual achievement. It is also, by my count, too silent on three subjects that do much to explain how markets actually behave: psychology, biology and history. This is not a criticism of the CFA, which does precisely what it sets out to do: produce technically excellent practitioners who speak a common language. The problem is what happens when the credential is mistaken for the territory. A practitioner trained exclusively in the language of finance is fluent in a vocabulary that describes only part of the system. The parts it does not describe are, with some consistency, the parts that may matter most when things go wrong.

The deeper issue is a confusion between two kinds of complexity. Markets are not complicated systems, in the technical sense. A jet engine is complicated: it has many parts, all of them tractable, and a sufficiently expert engineer can in principle understand the whole. Markets are complex. They are adaptive, non-linear and reflexive: the participants’ beliefs about the system change the system itself, which changes the beliefs. Complicated systems reward deep technical expertise. Complex systems reward something different: the ability to hold several models at once and the discipline to know which one applies when.

I have spent more of my life thinking about this than is probably healthy, partly because I studied economics in two different ways, at McGill in Montreal and at the University of Copenhagen, and the two universities sat at opposite ends of the same field.

At McGill, the economics department was housed within the Faculty of Arts. That placement was not an administrative accident. The dominant orientation was positive: how economic agents actually behave, how decisions are made under uncertainty, how knowledge is distributed across messy, limited, often confused or constrained human beings. Friedrich Hayek and Herbert Simon were on the syllabus. Bounded rationality was treated as a feature of reality, not a deviation from a more elegant model. I studied “Experimental Economics”, “Information and Uncertainty” and “Complex and Interactive Systems”.

At the University of Copenhagen, the M.Sc. in Economics sat within the Faculty of Science, and the physics envy was readily apparent. The curriculum borrowed differential equations and stochastic calculus. The dominant mode was normative: here is how rational agents should behave in well-specified systems. The models were beautiful. They were also describing a world populated by people I have never actually met.

Both traditions had genuine power. Neither was sufficient. And the gap between them, between description and prescription, between is and ought, turns out to be where most of the interesting questions in finance actually live.

I have watched, across more than two decades working in banking, the pension industry and foundations, the recurring institutional pattern this produces. The DCF model becomes the reality. The risk number becomes the risk. The benchmark becomes the objective. The map is taken for the territory, and the mistake is rewarded for as long as the territory is calm.

“When a measure becomes a target, it ceases to be a good measure.”

The popular formulation of Goodhart’s Law (Goodhart, 1975; Strathern, 1997)

Finance Underestimates Complexity

If the trap is that finance counts only what is easy to count, the obvious question is where the missing models live. Almost all of them sit just outside the syllabus.

From behavioural psychology, primarily through the work of Daniel Kahneman and Amos Tversky (Tversky & Kahneman, 1974), the field imported a vocabulary (loss aversion, anchoring, overconfidence, narrative bias) without ever quite importing the methodology. Practitioners learn to name the biases. Reliably correcting for them in committee, under pressure, with money at stake, is a different cognitive task entirely.

From evolutionary biology, the discipline could draw a more accurate language for what markets actually are: ecosystems, populated by competing strategies, in which any approach that works will attract capital until the capital itself extinguishes the advantage. Mean reversion is not a statistical curiosity; it is an ecological inevitability, the predator population exhausting its prey. The biologist’s metaphor makes this obvious. The language of alpha and beta, oddly, makes it harder to see. Andrew Lo, a rare field-crosser, built the bridge formally: his Adaptive Markets Hypothesis uses competition, adaptation and natural selection to reconcile behavioural finance with the efficient-markets view (Lo, 2017).

From history, particularly the empirical work of Carmen Reinhart and Kenneth Rogoff, the field could acquire a healthy respect for base rates and regime change. The phrase this time is different (Reinhart & Rogoff, 2009) has, with grim consistency, been the most expensive sentence in investment management, and the most popular one in the late stages of every bull market. Acemoglu and Robinson, in their seminal works Why Nations Fail (Acemoglu & Robinson, 2012) and The Narrow Corridor (Acemoglu & Robinson, 2019), add the longer lesson: institutions, not resources, decide which nations end up rich. And beneath history sits geography, the slowest variable of all. Jared Diamond’s Guns, Germs and Steel (Diamond, 1997) traces how continental shape and a handful of domesticable species handed some societies a head start that compounded over millennia; Tim Marshall’s Prisoners of Geography shows how mountains, rivers and access to the sea still set the board on which today’s politics is played. The deepest base rate of all is literally the territory.

The pattern across these fields is the same. The intellectual material exists. The incentive to integrate it does not.

Munger and the Distinction That Matters

It can be done, and one investor did it more deliberately than anyone. He is also the most misunderstood.

Charlie Munger is the most cited example of multidisciplinary investing in the modern canon, and the most frequently misread. The popular version has him reading widely, collecting mental models, and dispensing them in folksy aphorisms as Vice Chairman at Berkshire Hathaway’s annual meeting. All of it is true. All of it misses the point.

The distinction that matters, and which the rest of this essay rests on, is between two kinds of breadth.

Decorative breadth is the reading-widely-as-intellectual-accessory version. It is the executive who has read Kahneman, references System 1 (Thinking Fast) and System 2 (Thinking Slow) at conferences, and then builds portfolios exactly as he would have before (Kahneman, 2011). The vocabulary has changed. The reasoning has not. It is, on balance, mildly worse than no breadth at all, because it confers a sense of sophistication without any of the work.

Structural breadth is the version Munger actually practised. The models from psychology, biology, history and physics are not commentary on the analysis. They are the analysis. When Munger discussed the psychology of misjudgement, he was not adding colour to a discounted cash flow. He was describing the mechanism by which a price diverges from value, and the conditions under which it reverts. The model is load-bearing. Remove it and the analysis collapses.

The sequence matters as much as the breadth. Munger was not a generalist who wandered into investing. He was a serious investor, depth first, who built a latticework of models from other fields and used them to decide better within his own. Depth in something. Genuine curiosity in what lies next to it. The discipline to know which model applies when. Reverse the order and you do not get Munger; you get someone who has read widely and understood nothing.

The Foxes Were Always Right

One last piece of evidence, this time from the data. Philip Tetlock spent two decades studying forecasting accuracy across geopolitical events, economic indicators and political outcomes (Tetlock, 2015). Borrowing a distinction from the philosopher Isaiah Berlin (Berlin, 1953), he sorted his forecasters into hedgehogs (one big idea, applied everywhere) and foxes (many models, held lightly, updated often). Across thousands of forecasts, foxes won, consistently and substantially (Tetlock, 2005). The hedgehog’s confidence was not merely unwarranted; it predicted inaccuracy.

Asset allocation is a forecasting problem. So is risk management. So is scenario planning. If the best evidence we have says eclectic, update-willing thinkers beat single-theory specialists, then too many investment teams are built the wrong way round.

Hold this thought. It is about to become considerably more important.

The Institutional Trap

All of this raises an obvious objection. If breadth wins, and the evidence is this clear, why has an industry full of clever, competitive people not simply adopted it?

The answer is not stupidity. It is incentives. Narrow specialisation persists because it is locally rational at every level of the system, even where it is collectively costly. The trap is well designed. Nobody designed it.

In professional finance, the mechanism is the benchmark. A portfolio manager correctly positioned for a tail risk that arrives in three years will not, in most institutions, be given three years to be proved right. The evaluation horizon and the insight horizon are mismatched. The manager who would rather be roughly right than precisely wrong is fired before being proved roughly right. The one who is precisely wrong, in line with the benchmark or his peers, keeps the job. This is not a market failure. It is the system functioning precisely as built.

In credentialing, the mechanism is signalling. Two years reading deeply in philosophy of science, behavioural ecology and military history produce a richer mind and no certificate. Passing CFA Level III produces a narrower mind and a credential that opens doors. The system rewards the second because it is easier to verify. Verifiable beats valuable, in any institution that has to defend its hiring to people who were not in the room.

The deepest mechanism is principal-agent. Institutional investors answer to boards, trustees and regulators who must approve the decisions made on their behalf. A decision grounded in a discounted cash flow is more defensible to a committee than one grounded in complexity theory or evolutionary biology, even when the second is more likely to be right. Defensibility, not accuracy, is what the system optimises for.

The system is not broken. It is working exactly as designed. Which is precisely the problem.

The Premium Has Just Repriced

For most of financial history, the edge to earning superior returns came from information: knowing something first, or understanding something others did not. The merchant who learned of a war before his rivals, by carrier pigeon, by fast horse, by telegraph cable, in time by Bloomberg terminal, could act before the price moved. Gathering and assembling that information was costly. Getting it first was often decisive. Around this single asymmetry the whole architecture of professional finance was built. Research departments existed to find what others had not yet found; analysts, to interpret what others could not yet interpret. The specialist who knew one thing in greater depth than anyone else held an advantage that could last a career.

That edge has been eroding for years, and almost no one felt it go. In a rising market everyone makes money, so the specialist kept being rewarded and assumed his expertise was the reason, when increasingly it was the tide.

The shift is not that information has lost value. It is that it has become abundant, while judgment has become more important. The old edge was marginal: one more fact, gathered faster. The new edge is synthesis: filtering the noise, seeing the connections, and knowing what truly matters. Not the most information, but the best understanding of it.

That world, the one built on knowing first, is ending, and faster than most institutions will admit.

Two waves of compression

Every communication technology has done two things at once: carried richer information, and handed it to more people. Each step from pigeon to telegraph to internet both thickened the signal and widened the audience for it. What began as a message a merchant paid a fortune to receive first became, eventually, something everyone could see at once. Artificial intelligence is the latest step, and the most disruptive, because it compresses not just the delivery of information but the work of making sense of it.

The internet collapsed the cost of information. The proprietary advantage of investment banks and well-resourced funds, privileged access to filings, transcripts, industry data, expert networks, became, over roughly twenty-five years, a commodity available to anyone with a subscription. The asymmetry shrank, then largely disappeared. Specialists woke up to find that what they knew was no longer scarce.

Artificial intelligence is now collapsing a different layer: not the cost of information but more precisely the cost of answers. A capable language model can read an annual report in seconds, summarise a regulatory regime more comprehensively than most lawyers, and produce a competent first-pass valuation on request. The depth that took a specialist fifteen years to acquire is becoming available on demand, at a marginal cost approaching zero.

What stays scarce, and now separates one analyst from another, is judgement: the capacity to ask the right question, to see a pattern across genuinely different domains, and to synthesise what a model cannot yet assemble on its own. Knowing which question, of all the questions, is worth asking. Knowing what a number leaves out. Sensing what is reasonable, what is decent, what a society will actually accept. None of this appears in the training data, and none of it reduces to optimisation; it requires holding several incompatible frameworks at once and choosing between them, which is the one thing the model cannot do.

The mosaic and the tile

The frontier of specialist knowledge keeps receding, like the edge of an expanding universe, and the journey to it grows longer and dearer every year. Yet discovery does not always live at the border. More often it waits underfoot, in the stones we have not thought to turn, the angles we have not tried. Synthesis assembles tiles already visible into a mosaic that, once seen, looks as though it was hiding in plain sight all along.

This is not a future development. It is already visible in the structure of investment management. The portfolio manager has long outperformed the specialist analyst, not because she knows more about each company; the analyst usually knows it in a depth she cannot match. She wins because she knows how to size positions, manage correlations, and assemble a mosaic from pieces that behave differently under different conditions. The mosaic has always paid better than the single tile. AI has just made the tiles essentially free.

The implications for Tetlock’s foxes and hedgehogs are large. The hedgehog’s edge was depth in a single framework. When that depth costs less than a subscription to OpenAI or Anthropic, the hedgehog is competing with a machine that matches it instantly and for nothing. The fox is doing the one thing the machine cannot yet do: holding several frameworks loosely, switching between them, updating when the evidence turns. The gap between them is not narrowing. It is opening, sharply, in the fox’s favour.

The institutional trap was, for two centuries, an inefficiency the world could afford. It no longer can. The credential system is now producing experts in precisely the fields where AI has just become an expert too. The seam-crossers, meanwhile, face much less competition.

A Letter to the Next Generation

Everything to this point has been an argument. This is a letter.

If you have followed the argument, you already know its conclusion. The world has just changed the price of breadth. The specialist who built a career on knowing one thing very well now competes with a tool that knows it too, on demand, for nothing. The generalist who can synthesise across fields, ask the better question and see the pattern others have missed is, for the first time in a long time, holding the more valuable skill. I have spent twenty years watching this happen slowly in finance. The next twenty will see it happen almost everywhere.

Which raises the practical question, and the real reason for this essay: what do you tell your children about how to build a life in a world that has just been re-priced? I have two of my own, and I think about it often. So here is what I would say.

When I was younger, my parents told me to get a long education.

In their world, it was good advice. A master’s degree signalled specialist competence, and specialist competence bought a stable place in a stable hierarchy. The bargain was real, and for a generation it held. You traded years of your twenties for a credential that more or less guaranteed you a place in an organisation that more or less guaranteed it would still be there when you retired. Risk sat with the institution. Stability sat with the worker.

That bargain has been quietly rewritten. Not because education matters less; it matters more; but because what education is for has changed. A long education no longer buys a stable place in a stable hierarchy. It buys a foundation. What you build on it, and how willing you are to keep rebuilding, matters more than ever.

So here is the advice I find myself giving my children, and which I now think holds for all of us.

Try different things

The career ladder is becoming a lattice. Sideways moves, even backward ones, are no longer evidence of failure. They are often where the most useful learning happens, and increasingly how the most interesting careers are built. The person who has worked in three industries by forty has not been unfocused. She has been quietly compounding.

Be adaptable

The landscape will change, repeatedly, and more often than anyone expects. The specialist skills that matter at twenty-five may not be the skills that matter at forty-five. Reinvention is not a sign of failure. It is a sign of having paid attention.

Stay curious about fields that are not yours

The connective tissue between disciplines, the part that takes the longest to build and is hardest to credential, is becoming the most valuable real estate in the economy. The biologist who learns finance, the engineer who learns history, the philosopher who learns code: their skills do not yet have a category, which is exactly why they will be in demand. And the tool you may fear is the one that makes this reach possible. I believe AI will not replace the generalist; it will extend her knowledge domain, lowering the cost of a first step into an unfamiliar field so far that the only real barrier left is the curiosity to take it.

Be humble enough to keep updating

Tetlock’s superforecasters were not, on average, the smartest people in the room. They were the ones most willing to change their minds when the evidence changed. The willingness to update is not weakness; it is the most underrated form of intellectual courage. A line often attributed to Keynes, though probably not his, puts it best:

“When the facts change, I change my mind. What do you do, sir?”

Ask better questions

The premium has shifted, permanently, to the asking. The skill that will matter most in the next twenty years is not retrieving information, or even interpreting it. It is knowing which question, in the avalanche of possible questions, is the one worth asking.

Choose work that you find interesting

Pick the work that genuinely interests you. Not the work that signals well, or pays most in the first decade, or impresses the people who care about impressing each other. Find the questions you cannot stop turning over, and make a living adjacent to them. If the interest is real, the living usually follows, because the curiosity that pulls you into a field also pulls you into the ones beside it, and breadth, as this whole essay has argued, is now what pays. Money is the solvable problem.

The reverse is the trap. A secure salary with no curiosity behind it looks like safety and works like slow attrition, wearing you down year over year until security is all that is left, and it is not enough. A career can be comfortable and still be the wrong problem to have solved.

The compounding effect of working on something that interests you, over thirty, forty or more years, is not small. It is the difference between a career and a life.

A caveat I owe you, since this is advice. Breadth is a bet, not a guarantee. Ranging widely will not reliably make you the one who makes the discovery. But it loads the dice toward a life more interesting to live. Even when it pays nothing in the work, it pays in the pleasure of the work, in the rare satisfaction of discovering consilience, the same truth surfacing across fields that seemed to have nothing in common. That alone is reason enough.

A final word, since this is also a letter about stewardship. The old normative models assumed a single objective to maximise, a number to make bigger. They were not wrong, exactly. They were incomplete. The deeper question, what long-term actually means for a pension beneficiary thirty years from retirement, or a foundation meant to last in perpetuity, or a family wondering what to leave its grandchildren, is not a soft question decorating hard analysis. It is the question that determines everything else. Get it wrong and everything built on top inherits the mistake. Think clearly about what you are actually trying to achieve, not just what is easiest to measure.

The gap between how things are and how they should be is not a problem you solve once and file away. It is the permanent condition of anyone who takes both seriously. Get used to standing in it. The interesting work happens there.

Wisdom in the Footnotes

Return, finally, to the quote.

The truncated version, jack of all trades, master of none, won because brevity serves institutions, and institutions serve their own legibility. It is short, it is dismissive, and it confirms what the system already believed. The wiser, funnier, more honest full version is lost not by accident but by selection.

The people who changed their fields did not set out to be generalists. They set out to solve a problem their own field could not solve. Darwin among his barnacles. Shannon in his basement. Thorp at the roulette table. None of them were aiming at breadth. They were aiming at the problem, and the breadth was what the problem required.

• • •

The lane is comfortable. The seam is where the work is.

And wisdom, as it turns out, has always lived in the footnotes.

• • •

This article reflects my personal views.

For more on why early financial education matters, see: Why Most Family Offices Should Fail, And How to Build the Rare Exception.

For more on teaching your kids about investing, see: Teach Your Kids About Investing (Without Telling Them)

For wisdom in the footnotes (Fermat’s Last Theorem): see CFA vs CAIA: What Studying Both Markets Reveals About Complementarity

References

---

Notes

1. The history is worth spelling out, because the irony runs deep: the phrase’s ancestor was an insult, and the original target was Shakespeare himself. In 1592 the playwright Robert Greene attacked the young Shakespeare in his pamphlet Greene’s Groats-Worth of Wit as “an absolute Johannes factotum,” a Latin term for a jack of all trades, meaning a dabbler with no craft of his own. The English form “Jack of all trades” appears soon after, in Geffray Minshull’s Essayes and Characters of a Prison (written 1612, published 1618), and at that point carried no insult at all: it simply meant a versatile, useful person. The dismissive tail came over the following century. Thomas Fuller’s Gnomologia (1732) recorded the sour form “Jack of all Trades is of no Trade,” and by Charles Lucas’s Pharmacomastix (1741) the now-standard “master of none” had attached itself, curdling the compliment into a faint sneer. The redemptive coda, “but oftentimes better than a master of one,” is more recent still: lexical histories of the phrase find no documented instance of the full, optimistic version before the twenty-first century. It is not a lost antique but a modern addition, retrofitted onto an old proverb to flip its meaning back toward praise. A saying about the value of generalists, assembled by many hands across four centuries, has exactly the provenance it deserves. As for the Shakespeare attribution that still circulates: he never wrote it, and the one time the phrase’s ancestor crossed his path, it was thrown at his head. ↩︎