Think of something you’re not doing.
Not a goal you haven’t started — that’s a different problem. Something you’ve decided against, or postponed indefinitely, because of a specific obstacle.
Now answer one question about that obstacle:
Who enforces it?
Name the person. Or the rule, with the document it’s written in. Or the consequence, with the instance where you saw it happen to someone.
If you can do that, fine — you’ve got a real constraint and you should plan around it.
A surprising amount of the time, you can’t. There’s a wall there, you’ve been standing at it for years, and when you go looking for what’s holding it up you find nothing at all.
The Claim
A significant proportion of perceived obstacles have no enforcement mechanism. They persist because untested constraints generate no disconfirming evidence: a person who does not attempt the action never discovers that nothing prevents it, so the constraint remains unexamined while accumulating apparent solidity through age. Phantom obstacles fall into recognisable categories — permission being awaited from someone who is not withholding it, a rule inherited from an environment that no longer applies, a consequence assumed but never verified, an audience presumed to be observing who is not, and a past decision still being honoured under conditions that have since expired. They are distinguishable from real constraints by a single test: whether a specific enforcing person, document, or observed instance can be named. Real constraints survive the test. Phantom ones dissolve at the point of examination.
Five categories. Most people find at least two of theirs on the list.
One: The Permission
You’re waiting for someone to say it’s fine.
Not consciously — you’d deny it if asked. But there’s a sense that the thing isn’t yet authorised, that it would be presumptuous, that at some point somebody senior will indicate that you’re now the sort of person who does this.
Here’s the problem: nobody is running that process.
There is no committee. No one has your file open. The person you’re implicitly waiting on has not thought about your situation in some time, and if you asked them directly they’d be mildly puzzled that it required their input.
This is particularly common in people who did well in structured systems — school, professional training, hierarchical organisations — where authorisation genuinely was a thing that arrived from above on a schedule. You learned the shape of that world accurately, and then it stopped applying, and nobody announced it.
The tell: if you can’t name who would give the permission, you’re not waiting for permission. You’re waiting for a feeling, and feelings don’t arrive on request.
Two: The Rule Nobody’s Enforcing
Second category, and it’s the largest.
Rules from environments you’ve left, still being followed.
You don’t do things in a certain order because that’s how it was done at a company you left in 2017. You don’t ask for certain things because your first manager made it clear they weren’t asked for. You don’t put yourself forward in a particular way because it wasn’t done in your family.
Every one of those was, at some point, a real constraint with a real enforcer.
The enforcer is gone. The rule stayed.
That’s the mechanism: rules outlive the conditions that produced them, because there’s no process for decommissioning a rule. Nobody sends a notice saying the thing you learned at twenty-four no longer applies. You just keep following it, and it keeps producing no consequences, and the absence of consequences reads as the rule working rather than as the rule being irrelevant.
The test is the same one. Who enforces this, here, now? If the answer is a company you no longer work at, you’re being managed by a former employer for free.
Three: The Unverified Consequence
Third, and this one’s cheap to check, which is why it’s slightly embarrassing.
You’ve assumed the answer would be no.
You haven’t asked. You’ve modelled the asking, run it forward, arrived at a refusal, and treated the modelled refusal as data.
Then you’ve made decisions on it for years.
The thing about assumed refusals is that they’re free to generate and impossible to falsify. You never find out you were wrong, because you never created the conditions where you’d find out.
And modelled refusals are systematically pessimistic, for a straightforward reason: you’re modelling using your own sense of what you’re worth asking for, which is exactly the variable most likely to be miscalibrated in this situation.
The test here isn’t “who enforces it.” It’s simpler. Have you asked?
If not, you don’t have a constraint. You have a guess with a lot of mileage on it.
Four: The Audience
Fourth, and this one is quietly enormous.
You’re not doing the thing because of how it would look.
To whom, exactly?
Because the honest answer, most of the time, is a composite. Not a specific person — an assembled panel of former colleagues, people from school, a version of your parents, someone from a previous industry, and a general sense of people.
That panel does not exist and has never convened. And the real individuals in it are, without exception, thinking about their own situations rather than yours. They will notice what you do for approximately the length of time it takes to notice it.
The strange asymmetry here: you can verify this instantly by asking yourself how much time you spent last month thinking about the career decisions of someone you worked with in 2019.
None. You didn’t think about them at all.
That’s how much they’re thinking about you.
Five: The Expired Decision
Fifth, and this is the subtlest.
You decided something once, under specific conditions, and you’re still honouring the decision after the conditions ended.
I don’t do that kind of work. Decided when you were burnt out by a particular version of it, eleven years ago.
We don’t move. Decided when the children were small, and they aren’t now.
I’m not going back to studying. Decided during a period when you had no money and no time, both of which have changed.
These are the hardest to catch because they were good decisions. They were correct, they were made deliberately, and they served you well.
Decisions don’t come with expiry dates. There’s no mechanism that flags a standing choice for review when its inputs change, so a well-made decision from 2015 keeps running on 2026 conditions, silently, forever.
Test: what conditions were true when you decided this, and are they true now?
Why They Feel Solid
Worth explaining, because the obvious objection is that these things feel completely real.
They do, and the reason is the same loop that keeps installed limits alive.
A constraint you never test generates no evidence against itself. And a constraint that goes unchallenged for a decade doesn’t feel like an untested assumption — it feels like an established fact, because in your experience it has been consistently, invariably true.
The avoidance produces the certainty. Then the certainty justifies the avoidance.
That’s why the age of an obstacle is not evidence of its reality. A phantom obstacle gets more convincing with time, not less, because every year adds another twelve months in which it wasn’t contradicted.
What This Is Not
It is not a claim that most obstacles are imaginary. Real constraints exist in enormous numbers — money, time, obligations to other people, the actual state of the market, the fact that you have a mortgage. The claim is narrower: a specific and identifiable subset has no enforcement, and that subset is invisible precisely because it’s never examined.
It is not an instruction to ignore constraints. Testing an obstacle is not the same as pretending it isn’t there. Some of your tests will confirm the wall is real, and that’s a good outcome — a confirmed constraint is something you can plan around, while an untested one just makes you steer.
It is not about confidence. This is frequently framed as a self-belief problem and I think that’s wrong. Highly confident people carry phantom obstacles too; they just carry different ones. The issue is unexamined assumption, and examination is a procedure rather than a personality trait.
The Cost
Here’s the price, and it’s the reason people don’t run this audit.
If the wall isn’t there, you stopped voluntarily.
While the obstacle stands, your position is entirely defensible. You’d have done it, but the circumstances didn’t allow. There was a rule. It wasn’t possible. Anyone would understand.
Test it, find nothing, and that account collapses retroactively across every year you stood there.
Not “I couldn’t.” I didn’t check.
That’s genuinely uncomfortable and it’s a completely rational thing to avoid, which is why most people who read something like this agree with it entirely and test nothing.
Second cost, smaller: some of the tests will fail. You’ll ask, and the answer will be no, and then you’ll have a definite refusal instead of a comfortable ambiguity. Ambiguity is more pleasant to hold, and it’s worth less than nothing.
Common Questions
How do you tell if an obstacle is real?
Name who enforces it. A specific person, a document, or an instance where you saw the consequence happen. Real constraints survive that question. Phantom ones dissolve, because there’s nothing behind them to name.
Why do imaginary obstacles feel so solid?
Because they’re never tested, so they never generate disconfirming evidence, and an assumption that has gone uncontradicted for a decade feels like an established fact. Age makes them more convincing rather than less.
Am I waiting for permission I don’t need?
If you can’t name who would give it, then yes. Nobody is running an authorisation process for your life. This is especially common in people who did well in structured environments where permission genuinely did arrive from above.
Does anyone actually care what I do?
Briefly, and considerably less than you’d assume. The reliable check is to ask how much time you spent last month thinking about the career decisions of a former colleague. That’s roughly how much they’re thinking about yours.
What if I test it and the obstacle turns out to be real?
That’s a good outcome. A confirmed constraint can be planned around, resourced, or worked with. An untested one just makes you steer without ever knowing why.
How do I catch an expired decision?
Ask what conditions were true when you made it and whether they’re true now. Standing decisions have no review mechanism, so a correct choice from years ago keeps executing on current conditions indefinitely.
Where to Start
One obstacle. This week. Test it.
Pick from the list, and pick the cheapest one to check — not the most important. The most important one is high stakes and you’ll postpone it. The cheapest one produces a result in days.
Then do the actual test, which is almost always one of three things:
Ask the question you’ve been assuming the answer to. One email, one conversation. You’ll get a real answer, which is worth more than a decade of modelled ones regardless of what it is.
Name the enforcer out loud. Say the sentence: “I don’t do this because ___ would ___.” If you can’t complete it with a specific person and a specific consequence, you’ve got your result without needing to do anything else.
Check the conditions. For an expired decision, write down what was true when you made it, then write down what’s true now, and put them side by side.
One test. That’s it.
Because the thing about a wall you’ve never touched is that touching it is the entire experiment, and it takes considerably less time than the eleven years you’ve spent walking around it.
I’ve spent twenty-five years working on how ordinary people develop capabilities that look impossible from the outside, and a surprising proportion of what looks like a breakthrough from the outside is somebody discovering that a constraint they’d honoured for a decade had no one behind it.
Not overcoming the obstacle.
Finding out there wasn’t one.