In the last essay I wrote about the PSHE framework and the climb from being handed a clear brief to being handed nothing but a space and the job of finding the problem inside it. I left out the part where that climb gets genuinely hard. It isn't the seniority. It's that the higher you go, the more the room fills up with smart-sounding questions, all competing for the first thing you say next — and picking the wrong one to answer first is how a good decision quietly turns into a bad one, dressed in excellent analysis.
I've started calling the fix for this an eigen question, and I should say up front: I'm borrowing the word, not the maths. An eigenvector, if it's been a while, is the one direction a transformation doesn't rotate — everything else in the space gets spun and stretched, but that one direction just scales, up or down, and keeps pointing the same way. I'm not claiming decisions are literally linear algebra. I'm claiming the shape of the idea is exactly right: in most hard decisions there's a cloud of questions that shift depending on which way you look at the problem, and then, if you look hard enough, one question that doesn't shift at all — whose answer stays the load-bearing one no matter how the argument gets reframed around it.
In linear algebra, most vectors change direction when you apply a transformation — an eigenvector doesn't; it only stretches or shrinks. Applied loosely to a decision, most sub-questions shift as new facts arrive — the eigen question is the one whose answer stays the load-bearing one no matter how the argument gets reframed. It's a borrowed metaphor, not a theorem — the value is in the discipline of looking for it, not in the maths being literal.
Here's what that looks like when it isn't abstract.
The time machine
Picture the room. A whiteboard covered in a roadmap, six months of engineering already spent, a launch date that's already been said out loud to the board. Someone from leadership puts it plainly: "we have to sell the time machine." And the room does what rooms do — it reaches immediately for the busy questions. What do we price it at. Who's the customer. Which channel. What does the competitor do in response. All good questions. All, in this case, completely beside the point, because none of them is the eigen question sitting underneath the whole thing, which somebody eventually has to be willing to say out loud: is this actually good for humanity — or at least, not harmful to it?
If the honest answer is no, watch what happens to the room. The pricing model is still correct. The customer segmentation is still sharp. The channel strategy still makes sense. And it doesn't matter, not even a little, because the whole decision just collapsed to zero. That's the tell of a real eigen question — not that it's the hardest question in the room, but that a "no" to it makes every other answer in the room stop mattering at once.
Six more rooms
The time machine is a deliberately large example, and I think that's exactly why it's useful — it makes the shape of the idea obvious. But the same shape shows up in rooms that are much smaller and much more ordinary, which is really the whole point. Judgment isn't a special faculty reserved for boardrooms. It's a muscle, and like any muscle it gets built mostly in the unglamorous reps.
The boardroom
Leadership says it plainly: "we have to sell the time machine." Six months of engineering are already sunk into it, and a launch date has been announced to the board.
Where attention goes first
- •What do we price it at?
- •Who's the target customer?
- •Which channel do we launch through?
- •What does the competitor response look like?
The eigen question
Is this actually good for humanity — or at least, not harmful to it?
If the honest answer is no, the pricing model, the launch plan and the competitor analysis are all still true — and none of them matter. The whole decision collapses to zero the moment this one answer is no.
Notice the range — a boardroom and a cancelled train run on the exact same muscle. Judgment isn't a special skill reserved for the big rooms; it's a habit that's easier to build in the small ones.
The one I keep coming back to is the cancelled train, because it's the least dignified example and the most honest one. Almost nothing about being fifteen minutes late is actually urgent — the urgency is a feeling generated by losing control of the morning, not a fact about the morning itself — and yet I've watched myself, and plenty of capable people, spend real cognitive energy solving problems the situation never actually had. If you can't find the eigen question on a station platform, I'm not sure the boardroom version arrives from nowhere on the day it matters.
Practice where it's cheap
There's a second half to this, and it's the part I think gets skipped most often: good judgment on a big, expensive decision is very rarely someone's first attempt at judgment. It's usually the twentieth or the two-hundredth, rehearsed somewhere the cost of being wrong was small enough to actually learn from.
Pilots
A commercial pilot spends dozens of hours in a full-motion simulator — engine failures, storms, instrument loss — long before ever touching a real aircraft with passengers on it.
The simulator doesn't just teach the mechanics. It teaches what panic feels like at altitude, cheaply, so the first time it happens for real isn't also the first time they've felt it.
The common thread: none of these fields believe judgment arrives fully formed. They all built a cheap, repeatable place to get it wrong first.
This is, not coincidentally, close to a thread I've pulled on before. I wrote in The Oldest Game about how old this instinct is — that a huge amount of what looks like play, going back thousands of years, is actually a species rehearsing high-stakes judgment somewhere the stakes have been turned down on purpose. And I wrote in Seeing Early, Thinking Sideways about Shell's scenario planners, who ran exactly this kind of rehearsal before the 1970s oil shock and were, as a result, the ones already prepared when it actually landed. None of these fields believe judgment shows up pre-installed. They all built a room where the first mistake is allowed to be cheap — and then they made sure the real version wasn't anyone's first time.
I take that seriously enough that it shapes how I try to operate. Before a decision that actually matters, I'd rather have already been wrong about a smaller, similar one somewhere the wrongness cost almost nothing — a side project, a low-stakes call at work, a genuinely small bet — than let the first real attempt at that particular judgment be the one with money or people attached to it. It's not a guarantee. It's just better odds, and better odds, compounded over a career, is most of what good judgment actually is.
What this has to do with leadership
I think this is the real content of the word "leadership," stripped of the usual decoration. Not the loudest voice in the room, not the biggest title. The person who, when the room is full of smart, busy, competing questions, can find the one that actually decides it — and has the nerve to ask it out loud before the room has finished admiring all the other ones. That's a trainable skill, not a personality type, and the training ground is everywhere: a disagreement with your manager, a junior who wants the easy answer, your own kid, your own commute. Every one of those is a free rehearsal for the boardroom, if you're willing to treat it as one.
I don't think this is the last word on judgment — it's closer to the first proper one, and I suspect there are several more episodes in this series waiting to be written, because the eigen question changes shape depending on how many people the decision touches, and I've only really covered the personal end of that range here. For now, the practice is simple to state and genuinely hard to do: next time a room fills up with questions, resist the pull of the busy ones long enough to ask which single answer, if it came back no, would make all the others stop mattering. That's usually the one worth asking first.
Sources & further reading: The eigenvector/eigenvalue concept is standard linear algebra, applied here as a deliberate metaphor rather than a literal claim. On one-way vs. two-way door decisions, Jeff Bezos's 2015 shareholder letter is the primary source for that framing. On rehearsing judgment in simulation, standard aviation and surgical-training literature on simulator-based practice; Annie Duke, Thinking in Bets (2018), on treating decisions under uncertainty as a rehearsable skill rather than a test of intelligence; and, from earlier in this series of thought, Carlota Perez and Shell's own scenario-planning history as discussed in Seeing Early, Thinking Sideways. The Oldest Game (Deep Roots, part three) is the earlier essay on play as humanity's original rehearsal technology.