Parkinson's Law
Work expands to fill the time available for its completion. So set the time first, and let it decide the scope.
The Model
Parkinson was studying the British Admiralty and found something absurd. Between 1914 and 1928 the number of ships fell by two thirds and the number of officers manning them fell by a third, while the number of administrators rose by nearly eighty percent.
There was less to administer every year, and more administrators every year.
His conclusion was that work is not a fixed quantity that takes as long as it takes. It is elastic. It stretches to occupy whatever container it is given, and in an organisation the container keeps being enlarged by the people inside it.
He named two mechanisms. Officials multiply subordinates rather than rivals, because subordinates increase your standing and rivals reduce it. And officials make work for one another, so each addition generates enough correspondence to justify the next.
The operating consequence is short. A deadline is not a prediction of how long something will take. It is a design parameter that decides how long it takes.
Why operators get this wrong
Founders hear the law and immediately compress the wrong axis.
"We'll ship the redesign in two weeks" gets announced with the same forty-item scope attached. What follows is a two-week sprint, a four-week overrun, a set of half-finished items and a team that has learned your deadlines are theatre.
The law compresses elastic work — the polish, the fourth revision, the deliberation, the meeting about the meeting. It does not compress the irreducible work, and if you do not know which is which, you are not applying the law, you are just shouting at a calendar.
The correct move is always to fix the date and cut the scope. Fixing the date and holding the scope is how the law gets a bad name in a company that then stops using it.
The second failure is that nobody applies the law to money or headcount, where it operates identically. A budget expands to consume its allocation. A team expands to fill its org chart.
The six-person department that genuinely needs six people usually needed three and hired to a round number that appeared in a plan eighteen months ago.
Applied
Body
Give a training session ninety minutes and it takes ninety. Give it forty-five and the same work fits, because the elastic parts collapse first.
Watch what actually disappears under a hard stop. The phone comes out less, and rest periods stop drifting from two minutes to five while a conversation finishes.
The warm-up stops being social. The accessory list self-prunes to the two things that matter, without any decision being made about it.
Density rises and nothing of value is lost. That is the signature of elastic work being squeezed out rather than real work being cut.
Set the constraint externally so it cannot negotiate with you. A call at quarter to eight, a school run, somebody waiting in the car park.
A self-imposed limit you can extend is not a limit.
Eating has the same property. A defined eating window has no metabolic magic in it — what it does is remove the elastic hours during which unplanned food happens.
Business
Set the calendar before the scope, on purpose, and say out loud that this is what you are doing.
Instead of asking how long the new pricing page will take, commit to publishing it on Friday at five and let that decide what is in it. The version that ships Friday is usually eighty-five percent of the version that would have shipped in five weeks, and it starts collecting real data thirty days earlier.
Those thirty days of data are worth more than the missing fifteen percent, every time, and the missing fifteen percent is usually the part the market would have told you to change anyway.
Meetings are the purest demonstration available. A recurring sixty-minute meeting will use sixty minutes for the rest of its natural life. Cut it to twenty-five and the agenda survives intact — the status theatre and the two people who talk to be seen talking do not.
Then apply Parkinson's harder question before you approve a new role. Is this work real, or did it appear because there was capacity available to generate it? Half the time the honest answer is the second one, and hiring makes the condition permanent.
AI Leverage
The law now runs in a direction Parkinson never had to consider. Output used to be capped by the friction of producing it, and that friction was doing quiet, useful work as a limit.
The friction is gone. So work expands to fill the capacity available, and capacity is now effectively unbounded.
A team that produced four articles a month can produce forty. So it does, because it can, and it publishes forty pieces nobody reads while the one decision that would have changed the quarter has still not been made.
The container has to be set deliberately now, because nothing else sets it. Decide the deliverable and the deadline before opening the tool, not after seeing what the tool can do.
The useful framing: the model's job is to compress the time a fixed scope takes. It is not to inflate the scope because generation became cheap. Those two uses feel identical from the inside and produce opposite results in a quarter.
The Drill
Two parts, both this week.
First, take the task that has been on your list longest — the one that has migrated across four weekly plans and is starting to feel like a personality trait.
Book fifty minutes. Phone in another room, and a named artefact due at the end rather than an activity: "pricing copy published or in review," not "work on pricing."
Ship whatever exists at minute fifty. Whatever exists at minute fifty is the honest scope of that task, and you have been carrying a fictional version of it for a month.
Second, take your longest recurring meeting and halve it for four weeks. Do not announce a trial. Just change the invite.
Nobody will ask for the time back. That is the finding, and it will change how you read your calendar for the rest of the year.
Stoic parallel
Memento mori is Parkinson's Law applied to a life, and it was in use eighteen centuries before the Admiralty made the mechanism visible.
The Stoics kept death in view constantly — not as morbidity, but as the container. Marcus wrote about doing each act as though it were the last thing you would do, and about how a man who has been living for decades has often not lived, only existed for a long time.
Remove the boundary and a life becomes elastic in exactly the way an afternoon does. There is always another year for the thing that matters, so the year fills with the things that do not.
Seneca made the accounting explicit. We guard our property and our money jealously and give our time away to anyone who asks, though time is the one possession it is honourable to be greedy with.
A deadline is a small death, deliberately placed. It is also, in practice, the only reason anything ever gets finished.
One model per week.
Applied to training, business, and AI leverage. No fluff.
Related models
- Pareto PrincipleResults are not spread evenly across inputs. Measure the distribution before you decide where the effort goes.
- LeverageOutput per unit of input. Change the multiplier, not the effort behind it.
- Opportunity CostsThe real price of anything is the best thing you gave up to get it, and it never appears on an invoice.
Origin: Cyril Northcote Parkinson, from his 1955 essay in The Economist on the British Civil Service