Why deadlines work, and what they cost
The short answer
A deadline converts an open-ended thing into an ordered one, and ordering by deadline is not merely a habit — it is provably optimal for meeting deadlines on a single processor. That result is earliest deadline first, established by Liu and Layland in 1973.
That is a real guarantee and it comes with a real condition: it is optimal for that objective. It does not minimise average waiting time, it does not respect importance, and it will happily starve anything without a date attached.
So deadlines work. The question is what they cost, and the cost is specific enough to plan around.
Why the mechanism is real
A thing with no deadline has no position in any ordering. It competes with everything and loses to anything that is asking now, because "asking now" is the only comparison a queue can make without being told otherwise.
Giving something a date does three things at once. It creates a comparison — this before that. It makes the item loud, and loudness is what gets work selected. And it bounds the scope, because work expands to fill the time available and a date is what stops the expansion.
That is why an arbitrary deadline often works even when everyone knows it is arbitrary. The mechanism is the ordering, not the consequence.
What they cost
Everything without a date starves. This is the central cost and it is not a side effect — it is what an ordering rule does. Work with no deadline never reaches the front, because deadline-bearing work keeps arriving ahead of it. The formal name is starvation, and it is the reason important work never gets done.
Most genuinely important work has no natural deadline. Nobody sets a date for thinking, for the difficult conversation, or for the thing that will matter in a year. So a purely deadline-driven system systematically defers exactly the category of work it can least afford to.
Urgency and importance stop being distinguishable. Once everything that matters has been given a date in order to get done, the date stops carrying information. Every item is now urgent, which is the same as none of them being urgent, and the distinction collapses.
Optimality is conditional and the condition is easy to lose. Earliest-deadline-first is optimal on one processor when the work is schedulable at all. If more is due than can be done, the guarantee is void — it does not degrade gracefully, and the failure mode is missing several things rather than one.
Using them well
Put deadlines on things to sequence them, not to threaten yourself. The ordering is the useful part. A date carrying a punishment mostly produces padded estimates and quiet reprioritising.
Give the dateless work a standing slot instead of a fake date. This is the correction for starvation and it is what operating systems do — aging, where waiting work rises in priority until it must run. A fixed first block for the important-but-undated thing is aging implemented by hand, and it is more honest than inventing a deadline nobody believes.
Do not give everything a date. Once every item has one, you have removed the signal and kept the pressure, which is the worst of both.
Check what is actually due against what can actually be done. If the sum exceeds capacity, no ordering rule helps and the useful action is telling someone, early. That is not a scheduling decision.
The honest summary
Deadlines are a scheduling discipline with a proof behind it and a known failure mode. They are very good at the thing they are good at, and the cost falls entirely on work nobody has dated — which tends to be the work that matters most.
That is not an argument against using them. It is an argument for knowing which objective you have chosen, which is the only real question in prioritisation.
Where to go next
- How to use the urgent/important matrix — keeping the two axes apart
- Why your important work never gets done — starvation, and the standard fix
- Is there a best prioritisation method? — why optimality is always conditional
Using this in Decide One
Decide One asks for a duration on every line and optionally a when. The duration is what makes a day's total checkable against the hours in it, which is the check a deadline on its own never provides.
A line you did not finish is unfinished. It is not marked against you, and a day you did not open is not counted.
Open Decide One — free to use, with no account. Nothing you write leaves your device.
Decide One is a priority instrument. Bring the day into view, choose a method, give the first thing real time.