A child today can, from a phone, reach more information in an afternoon than most royal libraries held. Every proof, every lecture, every argument anyone has ever bothered to publish is one search away, free, for anyone with a connection. By any honest measure, humanity has never been closer to having handed every person on earth the sum of what it knows.

And yet the tally of towering minds has not obviously kept pace. Several serious writers on the history of science and the arts have pointed at the same uncomfortable shape in the data: measured against population, against literacy, against the sheer number of people now trying, the rate at which genuinely field-defining figures appear looks flatter — in some tellings, lower — since roughly the middle of the last century than it was before mass education, before the internet, before any of it. The claim is hard to prove cleanly; genius is not a quantity anyone can count with confidence, and every generation underrates its own. But it is hard to dismiss cleanly either, and it is worth sitting with rather than waving away.

This book was set off by an essay making exactly that case: the writer and neuroscientist Erik Hoel's "Why We Stopped Making Einsteins," which argues that the missing ingredient is not information — we have drowned ourselves in that — but a particular kind of relationship: sustained, personal instruction from someone further along, aimed precisely at the edge of what one specific learner can almost do. What follows is not a retelling of his essay, which is worth your time in full. It is my own attempt to think the claim through, test it against what else is known, and turn it into something a reader can actually use without needing a private tutor of their own.

Information got cheap. The relationship that turns it into ability did not.

Here is the shape of the argument, before the chapters unpack it. A great deal of research on what makes schools better — smaller classes, nicer buildings, more funding — moves outcomes only a little. One intervention moves them enormously: individual tutoring, calibrated in real time to exactly what a specific learner has and hasn't grasped. That effect has been sitting in the research literature since the 1980s, largely unused, because building it for everyone looked — and still looks — unaffordable. Meanwhile, if you go looking through the childhoods of an unusual number of historically outsized minds, the same structural feature keeps turning up, in eras and cultures that agreed on almost nothing else: not wealth exactly, not raw ability exactly, but early, close, personal instruction from an adult who paid enough attention to calibrate to the child in front of them.

None of this is a plan to reproduce the nineteenth-century tutor's study, which was funded by an inequality nobody should want back. It is an argument that the specific mechanism — calibration, proximity, a relationship rather than a broadcast — can be rebuilt in smaller, fairer, more modern forms, once you know that is the thing worth rebuilding. That is the book's actual promise: not genius on demand, but a clear enough account of the mechanism that you can go find or build a version of it for whoever you're responsible for teaching, including yourself.

The book runs in three short parts. First the puzzle and the number that points at an answer. Then the pattern, in the research and in the historical record, of what that answer actually looked like in practice. Then the honest cost of it, and what a reader without a household staff can still do with the idea. Each chapter closes with a small card — the principle, the strategy, concrete steps, and a way to know whether it's working — so nothing here stays merely interesting. A quiet meter fills at the edge of the page as you go, saved privately in this browser, waiting whenever you return.

Try an experiment. Name, without looking anything up, five scientists or mathematicians alive today whose reputation will plausibly outlast the century — the way Einstein's has, or Darwin's, or Curie's. Most people who try this stall at one or two, and a fair number stall at zero. Now try the same exercise for any twenty-year stretch between 1880 and 1950. It gets easier, not harder, despite that era having a fraction of today's population, a fraction of its universities, and none of its internet.

That mismatch is the puzzle this book opens with, and it deserves to be stated carefully rather than dramatically. Nobody can count genius the way you count a population, and every era is bad at recognising its own — reputations take decades to settle, and we are still inside ours. Several writers working on this question have tried anyway, building rough proxies from prize counts, citation patterns, and the density of field-defining breakthroughs per capita, and the shape that keeps emerging is a rise through the nineteenth and early twentieth centuries followed by a plateau, or in some tellings a decline, from around the middle of the twentieth century onward — even as the population doing science, and the tools available to it, both grew enormously.

The easy explanation is that the low-hanging fruit got picked — that early physics and mathematics were simply easier to revolutionise than what's left. That is real and it matters; a first-year student today learns in a semester what took the field's founders a lifetime to work out. But it cannot be the whole answer, because "harder problems" should mean *more* total resources thrown at each one, and generally does, and still the rate of era-defining output has not obviously kept pace with either the money or the headcount now aimed at it.

We gave everyone the library. We did not obviously give anyone the librarian.

That distinction — information versus the relationship that turns information into ability — is where the next chapter starts, with a number that has been sitting in the education research literature since 1984, mostly unused.

In 1984, the educational psychologist Benjamin Bloom published the results of a research program that had spent years comparing three ways of teaching the same material: ordinary classroom instruction, classroom instruction plus structured mastery checks, and one-to-one tutoring calibrated to the individual student. The tutored students didn't edge out the others. They finished, on average, about two standard deviations ahead — meaning the typical tutored student outperformed roughly ninety-eight percent of students taught conventionally. Bloom called the gap "the 2 sigma problem," and the problem, as he framed it, was not whether tutoring worked. It plainly did, better than almost anything else measured before or since. The problem was that nobody could see how to afford it for everyone.

It is worth sitting with how large that effect actually is, because the phrase "two standard deviations" slides past people who don't spend their time with statistics. It is the difference between an average student and one who tests better than nearly all of their peers — not because they are a different kind of learner, but because someone recalibrated the instruction to them, continuously, in real time. A tutor notices the exact half-formed misunderstanding a worksheet cannot see, closes it before it compounds, and adjusts the next question to sit exactly at the edge of what the student can just barely do. A classroom, built to serve thirty learners moving at different speeds with one teacher's attention divided thirty ways, structurally cannot do this — not from any failure of the teacher, but from the arithmetic of the format.

What happened to Bloom's finding next is the more interesting part. It did not get ignored, exactly. It became a fixture of education-school syllabi, quoted in the introduction of a hundred papers proposing something else — smaller classes, better software, adaptive platforms — each one an attempt to approximate the tutoring effect at a price a system could actually pay. Some of that work helped. None of it has closed the gap Bloom measured. The finding survived; the response to it mostly did not survive contact with the constraint that started the whole search — cost, at the scale a national school system operates on.

The tutoring effect was never a mystery. It was a cost nobody could spread across enough students to make it fair.

Which raises the next question honestly: what did that calibrated relationship actually look like, in the lives of people who had it? The next chapter goes looking for the pattern.

Marcus Aurelius opens his private notebook — the collection later published as the Meditations — not with a philosophy, but with a list of debts. Page after page, he names the specific people who shaped him and what, precisely, each one taught him: this uncle taught him patience, that tutor taught him to endure hardship without complaint, another showed him how to write without ornament. By the time modern historians count them, upward of a dozen distinct teachers appear in that list, brought to a boy who would eventually run an empire, each one apparently given real, sustained access to him rather than a single lecture in passing.

The pattern is not confined to the ancient world or to imperial households. John Stuart Mill's own autobiography — one of the most detailed accounts any historical figure has left of his own education — describes being taught Greek by his father starting at around age three, with lessons woven into daily walks and conversation rather than confined to a classroom hour, continuing in this close, one-to-one form for well over a decade. Bertrand Russell was largely kept out of formal schooling as a boy and instead taught by a sequence of private tutors, at a pace and depth no classroom of the period would have attempted. And in a smaller, less famous version of the same shape, a young Einstein's early exposure to advanced mathematics and philosophy came substantially through a family friend, a medical student named Max Talmud, who visited weekly for years and simply talked to him, book after book, well past what school alone was offering.

None of these are proof of anything on their own — a handful of biographies is not a controlled study, and it would be easy to cherry-pick four names and call them a law. But the pattern is worth taking seriously precisely because it recurs across eras, cultures, and social classes that otherwise had almost nothing in common, and because it matches, independently, exactly the mechanism Bloom's number pointed at a century and a half later: not more material, but a specific person, present often enough and attentive enough to notice where a given mind was actually stuck, and to meet it there.

What these lives share is not money. It is that somebody was paying close, repeated, individual attention — for years, not for an afternoon.

Which makes the next question sharper. If the mechanism is available in principle to almost anyone with time rather than wealth, why did it become rare? The next chapter takes that on directly.

Tutoring did not disappear. It is, if anything, a larger industry today than it has ever been — test-prep centres, homework-help apps, after-school programs, all doing brisk business. What changed is its job. For most of the households that use it now, tutoring exists to bring a struggling or merely average student up to a passing standard: closer to the classroom's floor, not further past its ceiling. The version aimed at pushing an already-capable learner as far as they could possibly go — the version this book has been describing — has become, for most people, either unavailable or simply unimagined as a thing worth asking for.

This is not a story with an obvious villain. Mass, compulsory, roughly standardised schooling solved a real problem that pure one-to-one tutoring never had to solve: how do you educate an entire population, not just the children of families who can either pay for or personally provide individual instruction? Judged against that goal — universal literacy, a baseline of shared knowledge, a system that doesn't quietly abandon whichever children happen to be born to inattentive or unlettered parents — mass schooling has been one of the more genuinely equalising inventions of the last two centuries. The tutoring model it displaced was never egalitarian to begin with; it was, in its historical form, a privilege of exactly the households that already had every other advantage.

Which leaves an uncomfortable, honest tension sitting at the center of this book, and it deserves to be stated rather than smoothed over: the mechanism that most reliably develops an individual mind furthest is also, in its classic form, one that a fair society has good reason to be wary of reviving at scale. A school system has to serve the median student and the struggling one at least as much as the exceptional one, and a society that funnelled its tutoring resources only toward children already ahead would be choosing to widen exactly the gap it should be trying to close.

We didn't abolish the tutor. We just started reserving one for people who were falling behind, and stopped imagining one for anybody trying to get exceptional.

The last chapter is where the book earns its keep: not a plan to fix a national school system, but a way to actually build a version of this for whoever you are responsible for teaching — including, quite possibly, yourself.

Take the mechanism apart one more time, because everything usable in this final chapter follows from it. What a tutor actually supplied was three things at once: proximity — enough contact to notice the specific place a mind was stuck; calibration — the judgment to aim the next question exactly at the edge of current ability, neither too easy nor too far; and duration — years, not a semester. None of the three, examined separately, requires a private household employee. They require a standing arrangement with somebody, and the discipline to keep it running.

Start with proximity, because it's the one people most often skip. Whatever skill you're developing, identify the specific person or small group you can reach on a fixed, recurring basis — a senior colleague who'll do a weekly fifteen-minute review, a near-peer slightly ahead of you in the same pursuit, a study partner with a genuinely regular slot. "Near-peer" mentoring — someone a few steps ahead rather than a formal expert far above — turns out to carry more of the tutoring effect than intuition suggests, because a near-peer remembers exactly where the current edge is, having only recently crossed it themselves.

Calibration is the part that actually has to be earned, and it comes almost entirely from one habit: asking for feedback on something specific and real, not general encouragement. "How am I doing?" produces nothing usable. "Here is the actual thing I made — what is the single smallest thing wrong with it?" produces exactly the kind of correction a tutor would have given, because it forces the other person to look at your real current edge instead of offering a vague verdict on your overall ability.

Duration is where most good intentions actually die, and it is the one factor that has nothing to do with cleverness and everything to do with a calendar. A single brilliant mentoring conversation changes almost nothing on its own; the effect Bloom measured came from years of continuous calibration, not from one especially good session. Which is why the practical version of this book's whole argument compresses to something almost embarrassingly plain: find the person, ask a specific question, put it on a repeating calendar invite, and do not let it lapse.

The tutor was never the point. Proximity, calibration, and duration were the point. The tutor was just how they used to arrive.

That is the whole book, really, turned finally to face you. Somewhere in your life is a skill worth this — yours or someone else's. The mechanism that used to require a household is, it turns out, mostly a matter of who you ask, how specifically you ask it, and whether you keep the appointment.