Schedule padding

Why does the same flight take half an hour longer in the timetable than it did thirty years ago?

Aircraft have not got slower. The airframes on many routes cruise at much the same speed they did decades ago, and some are faster. Yet the published time from gate to gate has crept steadily upward, sometimes by half an hour on a route of a few hours. The obvious explanations — congestion, longer taxi queues, air traffic routings — are real and account for part of it. But they do not account for the fact that these flights now routinely arrive early. What is a timetable actually promising?

Start with what a schedule is for. It is not a prediction of a single flight; it is a commitment about a whole network. An aircraft does not fly once and go home. It flies a rotation — perhaps four or five legs in a day — and each departure waits on the previous arrival. Ask, then, what happens when leg one runs twenty minutes late and every subsequent block time is set to the typical duration.

The answer is that the twenty minutes does not stay put. It arrives at the gate, walks through the turnaround, and departs again on leg two. Nothing absorbs it, so it persists to legs three and four, growing as it collects fresh delays. And it does not stay in that aircraft either: passengers miss connections, bags miss their aircraft, and crews approach the hard legal limits on duty hours — at which point the flight is not late but cancelled, stranding people and an aircraft in the wrong city overnight. A schedule with no slack does not fail gracefully. It fails all at once, and expensively.

So what would you do? You would build the recovery time in. Some of it goes into the turnaround, so a late inbound can still make its outbound. Some goes into the block time itself, so a flight that meets ordinary headwinds and an ordinary taxi queue still lands inside its published slot. The buffer is not a lie about how long the flight takes; it is a statement about how often the promise is meant to hold. A timetable built on the median journey is late roughly half the time. One built on the eightieth or ninetieth percentile is not.

Now push on the incentives, because this is where the story stops being purely engineering. In several jurisdictions punctuality is measured against the airline's own published schedule — in the United States, an arrival within fifteen minutes of the advertised time counts as on time. Notice what that permits. An airline can improve its measured punctuality without flying any faster, arriving any sooner, or fixing anything at all, simply by publishing a longer time. This is not speculation; the lengthening of block times and its relationship to on-time statistics has been examined repeatedly by regulators and researchers, and airlines have been candid about scheduling to protect the number.

So the same act — adding buffer — serves two ends at once, one legitimate and one closer to gaming. Which makes it genuinely hard to audit: from the outside, an airline padding to survive congestion and an airline padding to flatter its statistics look identical.

And the buffer is not free. Aircraft and crews are among the most expensive assets a carrier owns, and both are earning only while flying. Every minute of slack is capacity not sold, an airport slot occupied by an aircraft doing nothing, and a passenger planning their day around a time the airline expects to beat. Pad too little and the network cascades. Pad too much and you have bought reliability with capacity, and told your customers something that is not quite true.

Three refinements. First, padding and congestion are not rivals. Taxi times at busy hubs really have grown, routings really are less direct, and part of the lengthening is honest measurement of a slower world. The two causes are additive and hard to separate, which is exactly why the argument persists.

Second, buffer works because delay is asymmetric. A flight can lose an hour easily and cannot gain one — there is a hard floor on how fast the aircraft goes and how quickly a gate can be turned. A buffer therefore only ever has to soak up one direction, which is what makes a modest amount of it so effective.

Third, this is not a peculiarity of aviation. Railways publish recovery allowances in their timetables for identical reasons, and any tightly coupled network with variable stages does the same thing under a different name. The trade is always the one above: buffer converts capacity into reliability, and the honest question is never whether to pad but how much, where, and whether the published figure still means what a reader thinks it means.

• How hub-and-spoke banking concentrates connections and therefore concentrates delay propagation.
• Why measuring punctuality against a self-published schedule creates the incentive it does, and what alternatives regulators have tried.

Every term the collection defines is gathered in the glossary.