Deadlines, and what they are actually for

Work expands to fill the time available, and a deadline is the main mechanism preventing it. How well the mechanism works depends on how the deadline was set.

The evidence on deadline types

Research comparing deadline structures has found consistent differences.

Evenly spaced externally imposed deadlines produce the best performance. Self-imposed deadlines improve on having none at all. A single deadline at the end produces the worst outcomes.

Notably, when people set their own deadlines they generally do not space them optimally — they cluster them later than would serve them best, while still doing better than with no deadlines.

Which suggests that people have some awareness of their own tendency to delay and under-correct for it.

Why the final rush is expensive

Work concentrated at the end has predictable costs.

Less time for revision, since the first draft is the only draft.

No slack for the unexpected, which means any problem becomes a missed deadline.

Reduced quality of thinking, because time pressure degrades judgement in measurable ways.

And no opportunity for feedback, since nothing exists to show anyone until it is too late to act on their response.

That last point is frequently the largest cost and the least noticed.

Designing better deadlines

Interim deadlines with an external component. The key finding is that external beats self-imposed, so making an interim deadline external is where the gain is.

Sending a draft to someone on a fixed date converts a private intention into a commitment with a witness.

Deadlines on artefacts rather than on progress. "Outline by Tuesday" is checkable; "make good progress by Tuesday" is not.

Space them evenly, which is what people do not do naturally.

Make the first one soon. The first deadline determines whether the project starts, and a distant first deadline guarantees a delayed start.

Parkinson's law, examined

The observation that work expands to fill the time available is an aphorism rather than a research finding, and it has some empirical support in studies of task completion under different time allocations.

The practical implication is that generous deadlines do not produce better work in proportion, and that shorter deadlines frequently produce comparable output.

The limit is real: below some threshold, compression degrades quality genuinely rather than removing slack.

The useful heuristic is that the first substantial reduction in a generous timeline usually costs little, and successive reductions cost progressively more.

The planning fallacy interaction

Which complicates all of this.

People underestimate durations systematically, so a deadline set by estimating the work required is generally too tight.

Which produces the common pattern of aggressive deadlines missed routinely, and a culture in which deadlines are understood not to mean anything.

That is worse than generous deadlines met, because a deadline that carries no expectation of being met provides none of the benefit.

The correction is the outside view — how long did this take last time — applied when setting rather than when reporting.

Deadlines that do not work

Arbitrary ones. A date with no consequence and no reason is quickly recognised as decorative.

Deadlines nobody will check. The external element is doing the work, and an external deadline that is not actually externally observed is a self-imposed one.

Too many at once. Simultaneous deadlines force triage rather than prioritisation, and the triage is done under pressure.

Deadlines on work that is not defined. A date attached to an ambiguous task produces anxiety rather than action, because it is not clear what would satisfy it.

The personal version

For work with no external deadline, which is where most important individual projects live.

Manufacture the external component. Book the conversation, promise the draft, enter the event, tell the person expecting it.

The cost of doing this is a small social obligation. The alternative is a project with no forcing function, which is the condition under which most personal projects quietly stop.

The self-imposed version

Deadlines you set yourself work considerably less well than external ones, which is unsurprising and worth planning around.

What improves them is making them observable — a date given to someone who will notice, a meeting booked to present the result, a commitment made in front of people.

The mechanism is not shame so much as reality. A date nobody else knows about can be revised silently, and silently revisable dates are not deadlines.