Learning To See Again...
That we know more than we can tell, and the knowing is the last thing that’s ours
“Attention is the rarest and purest form of generosity."
- Simone Weil
“The real voyage of discovery consists not in seeking new landscapes, but in having new eyes."
- Proust
John Ruskin taught people to draw, and then told them the drawing was beside the point. What he was really after was their eyes.
I believe the sight is a more important thing than the drawing
he wrote in 1857. Put a pencil in a person’s hand and something stranger than a skill takes hold, the way they see begins to change, the way sight floods into a blind man who has lived his whole life without it.
His reasoning was physical. Set a pencil against paper and you have to slow down in front of something you’d otherwise have glanced at and filed away — to find the real edge of the leaf, the true angle of a shadow, the exact place a line bends. The hand interrogates the eye.
Do it long enough and the looking outlives the lesson. You keep seeing that way after the pencil is down, and you can’t switch it off. The marks on the page are residue. The thing you walk away with is the attention. And attention, Ruskin understood, is not a fixed endowment. It is trained, or it atrophies.
Hold onto that word: trained.
It’s about to decide what your work is worth.
Everything you can do is splitting into two piles: the part a machine can copy, and the part it can’t. The whole question of the next ten years is which pile you’re standing in.
Dan Loeb’s Lesson
More than a century and a half later, Dan Loeb is making the same discovery from the deck of a hedge fund and it is costing him money.
Loeb spent the back half of his career learning to pay up for quality. The lesson of Quality Investing and The Outsiderswas to find the durable business, the moat, the high return on capital, the hard-won edge and own it for years instead of trading it. He got good at seeing it.
Then last year became the worst stretch of his career for exactly those names. The apparently high-quality companies, he found, “very rapidly became less so” once AI arrived.
Why? Here’s the mechanism, because it’s easy to wave at and worth getting right. A quality stock is expensive for one reason: the market is paying today for the belief that its high returns will last. The premium isn’t for this year’s profits. It’s a bet on durability, on how many years the moat holds before competition grinds those returns back to ordinary.
AI didn’t have to touch the profits. It only had to make the market less sure the moat would last. Shorten the believed runway and the premium deflates, because the premium was the runway.
Quality names carry the biggest durability premium of all, so they re-rate first and hardest, long before a single quarter misses. The business can be fine and the stock still falls.
For some of the companies Loeb owned, it was worse than a re-rating. The moat literally was the codifiable thing, proprietary data, an information-services niche, expertise that had been written down and AI doesn’t just cast doubt on that kind of moat. It eats it.
Loeb hadn’t misjudged the businesses’ quality. He’d graded them right under the old rules, real moats, high returns on capital but the rules changed. What he’d misjudged was durability: which of those moats were built on something tacit and unscrapeable, and which were just expertise in explicit form, sitting there waiting to be copied. He’d treated a few of the second kind as the first.
Everything that can be written down, codified, scraped, transferred is losing its premium at once. The hard part, the part the rest of this essay is about is telling, in the moment, which is which.
What Was Ever Worth Anything?
Which brings back the question Ruskin was really asking in 1857. If the artifact is no longer the point, if the sketch, the codified expertise, the written-down moat is suddenly cheap, then what was ever worth anything?
His answer was the eye. The trained attention behind the hand. The thing that doesn’t transfer, because it was never written down.
Loeb is not an outlier, and the debasement is not a metaphor. It has already changed the shape of the market he works in.
Ken Griffin saw it too, and it rattled him. For years he ran Citadel as an AI sceptic, certain the technology would never reach his analysts’ judgment. Then he watched agents do in days the judgment work his PhDs had taken man-years to produce, inside his own four walls and admitted there was no position left to retreat to. He went home, by his own account, depressed.
Two Types of Knowledge
Start with a distinction the chemist-philosopher Michael Polanyi spent his late life on, in a single sentence:
We know more than we can tell
Knowledge comes in two kinds.
The explicit kind can be written down and handed over, a recipe, a manual, an earnings model.
The tacit kind can’t: the surgeon’s hands, a trader’s feel for the tape, the thing you can show but never quite say.
AI is built to swallow the first kind. As it feeds, the analysts at 13D Research argue, it debases explicit knowledge the way a flood of currency debases money not by destroying it, but by making it cheap and everywhere at once.
What that breeds in a market is a monoculture. Every fund now trains its models on the same transcripts, the same feeds, the same price histories. Underneath the branding, the engines converge.
We actually discussed this from a different angle in December 2025, in the piece titled why is modern culture stagnating.
Plant a whole field with one identical crop and it’s beautiful in fair weather and defenceless in a blight. A market running the same tools on the same data is the same story. In the calm of mid-2025 — conditions nobody would call a storm — quant funds lost 4.2% in a single crowding unwind. Every strategy fell at once, because underneath they were the same strategy.
But..the smart money has figured something out, and you can tell by watching where the most sophisticated machine-users are sending their own money now.
D.E. Shaw, a firm that helped invent quantitative investing, raised five billion dollars for what Bloomberg called its first hedge fund run by humans — purely discretionary.
Capital is fleeing the monoculture. And it is fleeing toward the one thing the machine can’t mass-produce. It is fleeing toward the eye.
Using Both Types of Knowledge
So what is the eye, once we put the pencil down?
Loeb draws the line straight through his own business.
Public markets, he expects, will only get more automated. But he can’t see a machine on a creditors’ committee at two in the morning, reading a room and a balance sheet at the same time. The high-touch stuff stays human because it’s relational and contingent. It won’t transfer, because it was never written down.
But what he tells his own analysts about the other half of the work is to use AI relentlessly, because it gives back whatever you put into it. The machine is an amplifier. It swallows the part that can be written down and hands your judgment back larger.
We discussed this with Gavin Baker’s analogy of the samurai and the machine gun last week.
A lifetime of hard-won pattern recognition, cut down in the field by a peasant with no judgment and a faster gun and the only position left standing being the composite, the samurai who picks up the gun.
The eye and the engine is the same truth from the other side.
The engine is the gun: tireless, literal, blindingly fast, and blind. The eye is the samurai: the thing that decides where to point it.
The part I left out last time is that the samurai’s eye is not something he carries into the field intact. It is trained, and it can go dull.
What Is The Eye Made Of?
Two things finish the picture of what the eye is made of.
The first is almost embarrassing to put in a markets essay, and Loeb says it anyway: kindness. Be kind to people who can’t possibly help you, he says, precisely because you can’t predict where it pays.
Kindness makes empathy, empathy makes connection, and connection is the raw material of every relationship that turns out to matter. The least automatable asset he can name is generosity offered with no expected return because a machine has no use for a gift it can’t price.
The second is subtraction.
Loeb keeps a warning Eric Schmidt gave at a dinner in 2013: you’ll want to treat this burst of change as a blip that settles back to normal. Don’t. It only speeds up from here. Schmidt has been right at every checkpoint since.
Set that speed against a fixed human ceiling — our brains barely evolved to handle social media, Loeb notes, and AI is a much larger dose of the same — and the bottleneck stops being the technology. It becomes the nervous system trying to drink from a hydrant.
That flips the scarce skill. It’s no longer the ability to keep up. It’s the discipline of subtraction: deciding what not to look at, so the few things that matter get all of your attention.
Seeing clearly is only half of it. The trained eye also chooses, ruthlessly, where to look.
Even God Would Get Fired
Hayden Capital put a number on it. Across roughly 6,500 US stocks from 1985 to 2024, the median company fell 85% from its peak at some point, and took two and a half years to recover. That’s the base rate for the average stock. Not the disaster. The average.
Wes Gray sharpened it into an experiment with the perfect name: “Even God Would Get Fired.”
Build a portfolio with literal perfect foresight, the actual fifty best stocks of the next five years, known in advance, and you’d still have ridden it through a 76% fall, with five separate drawdowns of more than 30% along the way.
Flawless selection doesn’t spare you the wall. It guarantees it.
So if you can’t avoid the drawdown, the edge can’t be avoidance. It has to be diagnosis. And this is where judgment earns its keep.
A hall-of-fame stock and an average one fall about the same distance. They differ in the nature of the wound, not its depth. Is it cyclical, something the business heals from, or secular, a permanent change in how the customer feels about the product? The average stock never comes back. The great one does.
And in the moment, falling, the only thing that tells them apart is whether the damage is to the price or to the business. No model can make that call for you, because the data that would prove it arrives only after it’s too late to act. It’s a judgment. It’s Ruskin’s discipline, aimed at a chart that’s still dropping.
It’s also, exactly, the call Loeb got wrong on his dissolving moats, a secular wound he’d booked as a cyclical one.
Stranger still, the monoculture pays the patient, something we discussed in Staying.
Loeb’s frame for whether AI changes markets is Ecclesiastes by way of Reminiscences of a Stock Operator:
There’s nothing new under the sun. The machines don’t remove human emotion. They re-encode it as risk limits and stop-losses that force selling on the way down. Rational for the fund. Irrational for a long-term owner.
The more capital runs on the same forced-selling rules, the more reliably those rules manufacture the very mispricings a steady hand can buy. The herd’s flinch is the patient eye’s harvest. The thing the machines are built to suppress the willingness to take the pain of a loss in the short run, turns out to be the edge.
Sitting still is a faculty. It is trained, or it atrophies.
Of course, the eye has its own way of failing, and it has nothing to do with going blind.
A long run of being right breeds a quiet overconfidence, the survivor who mistakes his lucky stretch for the way the world works, and stops looking because he’s sure he already knows what he’ll see. An eye stays sharp only as long as you keep using it. Stop, and it closes.
The Machine’s Trap
But the machine makes looking optional. Every glance can be outsourced, every judgment deferred, every gap filled by something fast and frictionless and a faculty you stop using is a faculty that fades. Abundance has always been the enemy of the eye. It’s why Ruskin made his pupils slow down. So how do you train an eye when nothing forces you to anymore?
Nobody alive is better placed to answer than Rick Rubin, who has spent fifty years building the most valuable and least explainable asset in music: taste.
Ask him why a take is good and he can’t tell you. He says he knows nothing, runs on pure intuition, feels his way to the answer. From the outside it looks like magic. Up close it’s the residue of an absurd amount of looking.
Rubin treats himself as a researcher. Hear a band he loves and he hunts down whoever inspired them, then whoever inspired them, down and down the lineage. Find a coffee he likes and he’ll taste a hundred more, and read everything anyone has written about the machines, before he settles. The intuition he can’t explain is just years of deliberate comparison, compressed into a verdict he can reach in a second.
Then there is the edit.
His rule is that to arrive at less you have to do more: if you want to end at seventy percent, don’t shave off the last thirty cut all the way to forty, then build back only what the work can’t live without. What survives is the essence. It is the same instinct as the trader telling a secular wound from a cyclical one, or a parent working out which child needs what, the discipline of knowing what to throw away.
None of it is fast, and none of it is free. He shows up to the studio on mornings he would rather be on the beach, and waits sometimes for nothing because the magic only arrives if you’re there when it comes. Which is Ruskin’s point, a century and a half on, in a different room. The pencil. The hours. The looking again.
This is why the premium is not democratic. The value that survives gathers around whoever still does what Rubin does whoever keeps looking once looking became optional.
The eye was always trainable. It is just getting rarer, because fewer and fewer of us are made to train it.
This will only get worse in the future, because the “answers” will feel free and frictionless.
Learning To See Again
Which is where we end, back in 1857, with a man teaching drawing to people who’d mostly never become artists, and not minding in the least.
Ruskin understood the thing the market is now repricing in real time, and the thing the young are saying in anger. The artifact was never the point. The sketch, the model, the codified moat, the written-down expertise, all of it is residue, and all of it is going cheap.
What appreciates is the eye. The trained attention behind the hand. The judgment that decides what counts, diagnoses the wound while it’s still bleeding, sits still when the machines flinch, and refuses the survivor’s flattering story.
It was never something you could consume your way into. You earn it the way it has always been earned, through slow, patient, hands-on work, by looking again at the thing you’d have glanced past.
And here’s what I didn’t expect to find at the end of a markets essay.
The faculty I’ve been calling the eye is the one I wrote about in April from the other side of my own house, standing at a kitchen sink, working out that the whole of parenting might come down to whether you turn the water off when your child walks into the room.
The eye that prices a business and the attention that witnesses a person are the same muscle, trained the same way, by slow and patient looking, and dulled the same way, by the abundance that keeps the feed faster than the page and the phone louder than the child.
The machine will take everything that can be told. What it can’t take is the thing Polanyi pointed at, and Ruskin trained, and a furious 23-year-old defended without knowing she was quoting either of them:
That we know more than we can tell, and the knowing is the last thing that’s ours.
The blind man receives his sight. The world arrives, all at once, transformed.
But only if you put the pencil to the paper. And only if you keep it there.
AI will take everything that can be told. The knowing is the last thing that's ours
The eye is getting rarer because fewer and fewer of us are made to train it. Sharing this is one small way to make one more.
A note on sources:
This essay is a work of synthesis, and most of its raw observations belong to other people. I want to be clear about who.
The spark was a single issue of 13D Research’s What I Learned This Week (21 May 2026): the Ruskin passage read as a discipline of attention (ch. 4); the “knowledge monoculture,” the Polanyi tacit/explicit distinction, and the debasement frame, along with the D.E. Shaw and Millennium tells (ch. 3); and the data behind the Gen Z backlash (ch. 8). The intuition that these belong inside one frame is what I’ve tried to add. The threads are theirs.
Dan Loeb’s observations — quality moats dissolving, what stays human, “nothing new under the sun,” kindness as the last moat, essentialism and the Schmidt acceleration — are from his conversation with Patrick O’Shaughnessy on Invest Like the Best (2026). The samurai-and-the-machine-gun frame is Gavin Baker’s, from his own appearance on the same show.
The drawdown data and the cyclical-versus-secular diagnostic are from Fred Liu’s Hayden Capital Q1 2026 letter, drawing on Michael Mauboussin’s “Drawdowns and Recoveries” and Wes Gray’s “Even God Would Get Fired.”
The reflections on taste, the “ruthless edit,” and making things for an audience of one are Rick Rubin’s, from his conversation with David Senra on the Founders podcast (2026) and his book The Creative Act.
Where I’ve quoted, I’ve kept it short and named the speaker. Everything I got right here, I owe to the people above. Everything I got wrong is mine.
Further reading
13D Research — What I Learned This Week, 21 May 2026 (chs. 3, 4, 8)
Dan Loeb and Gavin Baker — Invest Like the Best with Patrick O’Shaughnessy, 2026
Fred Liu — Hayden Capital Q1 2026 Letter (citing Michael Mauboussin, “Drawdowns and Recoveries”; Wes Gray, “Even God Would Get Fired”)
Rick Rubin — The Creative Act: A Way of Being (2023); in conversation on the Founders podcast (2026)
Michael Polanyi — The Tacit Dimension (1966)
John Ruskin — The Elements of Drawing (1857)





Unique content. The closest that I have come to this is Dominic Frisby. Very thought provoking.