A note on what this article is. This is a piece about a mechanism — why changes that are structured as projects tend to reverse when the project ends. Weight is the most familiar example of the pattern, which is why the question gets asked in those terms, but bodies and health are genuinely complicated, involve real biological and medical factors, and are not what I’m qualified to write about. There are no targets, numbers, plans, or protocols here, and nothing in this piece is advice about food or exercise. If your relationship with food or your body is a source of ongoing distress, that’s worth talking to a doctor or a qualified professional about, and it’s a more important question than the one this article answers.
With that said, here’s the mechanism, because it applies far beyond its most famous instance.
Something strange happens at the finish line.
A person sustains a demanding change for months. Genuine effort, genuine consistency, genuine result. Nobody could call it a failure of commitment — the evidence of commitment is right there, in the months.
They reach the target.
And within a period that’s usually shorter than the effort took, they’re back where they started.
The standard explanation is that they lacked the discipline to maintain it. That explanation is available, it’s socially accepted, and it doesn’t survive the first question you’d ask of it: where did the discipline go?
It was demonstrably present for months. Discipline that vanishes on the day of success isn’t a character trait behaving inconsistently. It’s a structure being removed.
What’s Actually Happening
Changes structured as projects tend to reverse on completion, for a reason built into their structure rather than into the person attempting them. A project has a defined endpoint, and the endpoint provides the organising structure that sustains the behaviour: a target to measure against, a reason to decline competing demands, and a criterion for what matters. Reaching the target removes all three simultaneously. Because the behaviour was being maintained by that structure rather than generated by the person’s self-concept, its support disappears at the moment of success. The previous behaviour then returns, not as a decision but as a default requiring no effort. This produces the characteristic pattern in which reversion follows achievement rather than failure, and it is commonly misattributed to a loss of discipline occurring at precisely the moment discipline had most recently been demonstrated.
Three parts.
One: The Ending Was Built In on Day One
This is the load-bearing point.
When a change is framed as a project, its termination is specified before it begins. That’s what a project is — a defined effort with a defined endpoint.
So on the first day, the last day already exists. The whole structure is oriented towards a moment when it stops.
And the mind is not confused about this. It knows the arrangement is temporary. It has been told, explicitly, by the framing, that this is a period of doing something rather than a change in what you do.
Which is why the reversion doesn’t feel like failure when it happens. It feels like completion. The project finished. Projects finish. Nothing has gone wrong.
Contrast that with something you’ve done consistently for a decade without a target. Ask yourself when it ends and the question won’t parse, because there’s no endpoint in the structure. It isn’t a project. It’s just what you do.
A change with a finish line has quitting scheduled into it. That’s not a risk. It’s the design.
Two: Reaching the Target Removes the Architecture
Second, and it’s the same mechanism that makes achievement feel flat.
While you’re working toward something, the target does an enormous amount of unnoticed work. It tells you what matters. It gives you grounds for declining things. It supplies a criterion for every small decision, so that most decisions aren’t decisions at all — they’re just consistent with the target.
Remove it and you don’t only lose a goal. You lose the decision-making apparatus that was running off it.
Now every choice that used to resolve automatically becomes a live question again, one at a time, with no reference point. And live questions get answered by the default, because defaults are what answer questions in the absence of anything else.
This is the same structural finding as the successful plateau: the problem was holding the life together, and solving it takes the architecture down with it.
The reversion, then, isn’t a collapse of will. It’s the predictable consequence of removing the thing that was making the behaviour easy.
Three: The Language Gives It Away
Third, and this is the diagnostic.
Listen to how people describe a change while they’re in it.
I’m on a plan. I’m being good at the moment. I’m doing this until the summer. I’m allowed one on Fridays.
Every one of those frames the behaviour as external, temporary, and imposed. Something being done to the person, for a period, with rules.
Now listen to how someone describes something that’s actually part of them. They don’t describe it as anything. They can’t easily answer questions about how they stay motivated, because there’s no motivation involved — the question doesn’t connect to their experience.
That difference is the entire predictor.
If the person can describe the effort, the effort is still required. And anything requiring effort ends when the effort is disrupted, which is what happens at week three and again at the finish line and again at every interruption in between.
The person who doesn’t revert isn’t maintaining anything. There’s nothing left to maintain, because the behaviour stopped being a thing they do and became a description of who they are.
What Makes a Change Hold
The mechanism implies the fix, and it’s not a better plan.
Remove the endpoint. A change with a finish line schedules its own reversal. One without a defined ending has nothing to complete, so there’s no moment at which the structure is removed. This sounds like a small semantic difference and it’s the largest single variable in whether something holds.
Change the description before the behaviour. Not “I’m doing X for now” but a statement about a kind of person who does X, indefinitely, without it being remarkable. Then derive the behaviour from that rather than pursuing a target. That direction — identity first, behaviour second — is the whole argument of stealing your future self, and it’s the inverse of how almost everyone attempts change.
Expect the environment to pull. Old settings and old company reliably reactivate previous versions, independently of anything you decide. That’s why you become the old version around the old people, and it’s a large part of why reversion so often begins during a trip, a holiday, or a return home rather than during ordinary weeks.
None of that is faster than the project version. It’s slower, and it has no satisfying finish. That’s precisely why it holds.
What This Is Not
It is not a claim that willpower explains body weight. It doesn’t, and I’d be talking outside my competence if I suggested otherwise. Body weight involves biology, medication, genetics, sleep, illness, and circumstances, and the mechanism described here is one factor among many rather than the explanation. What this article describes is a pattern in how behavioural change is structured — no more than that.
It is not a criticism of anyone who has reverted. The pattern described here is a property of project-shaped change, not of the people attempting it. Someone who sustained months of effort and then reverted demonstrated the discipline and was let down by the structure.
It is not an argument against having goals. Goals are excellent at producing a period of effort. What they’re poor at is producing a permanent state, because a goal that’s been reached stops functioning. Use them for what they do, and don’t expect the other thing from them.
The Cost
Here’s the price, and it’s the reason project-shaped change remains so popular despite reverting so reliably.
You have to give up the finish line, and the finish line was the motivating structure.
The target is what makes the effort bearable. It’s the thing you’re counting down to, the reason today’s difficulty is temporary, the image you hold when it’s hard.
Take it away and you’re being asked to do something indefinitely, with no arrival, no completion, and no moment where you get to stop and be congratulated.
That’s a much worse offer, emotionally. It’s also the only version that holds, which is an unsatisfying pair of facts to hold together.
Second cost: the identity version is slower and produces nothing visible for a long stretch. Project-shaped change produces a measurable result and a date. Identity change produces a gradual, unremarkable, uncelebrated shift that nobody notices until it’s been true for a year.
Most people would rather have the visible version, twice, than the invisible version once.
Common Questions
Why do people revert after reaching a goal rather than after failing?
Because the goal was supplying the structure that held the behaviour up. Reaching it removes the target, the criterion for decisions, and the grounds for declining competing demands, all at once. The behaviour loses its support at the moment of success.
Where does the discipline go?
Nowhere — it was never the mechanism. What ends is the structure that made the discipline productive. Someone who sustained months of effort has demonstrated discipline; what they were missing was an arrangement that didn’t terminate.
What makes a change permanent instead of temporary?
The absence of an endpoint, and a self-concept that generates the behaviour rather than a target that requires it. If the change is described as something you’re doing for now, its ending is already specified.
Why does the language matter?
Because how someone describes a change reveals whether effort is still required. Being able to explain how you stay motivated indicates the behaviour is still being maintained. Where it’s become identity, the question doesn’t connect to anything.
Is this the same as the three-week collapse?
Related, and it’s the same underlying mechanism at a different point. Week three is when temporary supports run out mid-effort. This is when the structure is removed by success. Both are behaviour losing its scaffolding rather than a person losing their will.
Does this mean goals are useless?
No — they reliably produce a period of effort, which is genuinely valuable. They’re just poor at producing permanent states, because a reached goal stops functioning. Use them for the first thing and build the second thing differently.
Where to Start
Ten minutes, and it’s a language exercise rather than a plan.
Take a change you’re currently attempting or have attempted before, and write down how you’d describe it to someone at dinner.
Then look at your own sentence for three things:
Is there an end date in it? Explicit or implied. “Until the summer,” “for now,” “this year.”
Does it describe an effort or a person? “I’m doing X” versus “I’m someone who X.”
Could you answer “how do you stay motivated?” If you have a good answer, the effort is still required, which means the arrangement is still temporary.
Then rewrite the sentence once, as a description of a kind of person, with no end date and nothing remarkable about it.
That rewritten sentence won’t change anything on its own. What it does is show you the gap between how you’ve framed the change and the form changes have to take to survive their own success.
I’ve spent twenty-five years working on how ordinary people develop capabilities that look impossible from the outside, and the ones who hold something for decades almost never describe themselves as having been disciplined about it.
They describe it as what they do, which is the same thing said by someone who never scheduled an ending.