Boarding order
A Socratic walk-through of boarding order — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why does boarding an aircraft neatly from the back forward end up slower than letting people on at random?
Boarding from the back forward feels obviously right. Nobody has to walk past anybody already seated; the cabin fills like a glass filling with water; the procedure has an orderliness that seems as though it must translate into speed.
It does not. Simulations, and field tests with real people and real bags, put strict back-to-front among the slower methods — slower than simply opening the door and letting people on in whatever order they turn up. That is worth explaining, because the intuition behind back-to-front is not stupid. It is optimising the wrong quantity.
Reasoning it through
REASONING #Ask first what actually consumes time during boarding. It is not walking. A person covers the length of a narrow-body aisle in well under a minute, and if walking were the constraint, boarding would take a couple of minutes.
The time goes on two kinds of blockage. Aisle interference: a passenger stops in the aisle to put a bag in the overhead bin, and everyone behind them stops too. Seat interference: a passenger has a window seat, the aisle and middle seats in that row are already occupied, and those people must stand and step out.
Now the key observation. Stowing a bag takes some fixed time — call it half a minute or so — and it is the dominant cost. But crucially, several people can stow bags at the same time, provided they are far enough apart along the aisle that they do not block each other. Boarding time is therefore roughly the total stowing work divided by the number of people who can do it simultaneously.
Write that down and the puzzle dissolves. If N passengers each need time t to stow, and k of them can be stowing at once, boarding takes about N·t/k. The procedure's job is to make k large.
Back-to-front does exactly the opposite. It deliberately gathers the currently-boarding passengers into one contiguous block of rows — which is to say, into one short stretch of aisle. Within that stretch, essentially one person can stow at a time while the rest of their group queues behind. Back-to-front maximises the neatness of the queue and minimises k.
Random boarding, by contrast, scatters the arriving passengers along the whole length of the cabin. At any moment several are stowing in different places, none blocking the others. Nobody designed this to be parallel; it is parallel by accident, because the seat assignments are spread out. That accidental parallelism beats the deliberate ordering.
Push the reasoning one step further and you get the design that actually wins. If spreading people out is good, spread them out on purpose: board passengers in alternating rows, so consecutive boarders are two rows apart and can stow simultaneously, and within that, board window seats before middles before aisles, so seat interference never occurs either. That method has been tested against the alternatives with volunteers and luggage, and it is substantially faster than both back-to-front and random.
Notice what the reasoning did. It replaced "keep people from walking past each other" with "keep people from waiting behind each other", and those turn out to be almost opposite objectives.
The analogy
THE ANALOGY #Think of a long counter with one shared walkway behind it and a queue of people who each need to spend half a minute putting something on a shelf.
Send them in strictly by position — everyone for the far end first — and they arrive as a clump. The walkway at that end is full, one person reaches their shelf while the rest stand behind, and the other ninety per cent of the counter sits idle.
Send them in scattered by position and the whole counter is in use at once. The total shelving work is identical; the elapsed time is not, because the second arrangement uses the length of the counter and the first insists on using one end of it at a time.
The counter's shelves are interchangeable, whereas an aircraft's overhead bins fill up, so late boarders may find no space near their seat and have to walk back — a failure mode with no counterpart at the counter, and one that carry-on charges have made considerably worse.
Clarifying the model
THE MODEL #Airlines are not optimising boarding time alone, which is why the fast method is rare. Boarding order is also a product. Priority boarding is sold, and status is recognised through it; families need to sit together; and the alternating-row method requires passengers to arrive in an exact sequence, which is difficult with groups who will not be separated. A method that is fastest in a trial with cooperative volunteers can be slower in practice if a fifth of the cabin cannot be placed where it belongs. The persistence of slower methods is not simple error — it is what happens when a process has several objectives and speed is only one.
The carry-on effect has grown and it is a real confound. As checked-bag fees pushed luggage into the cabin, stowing time per passenger rose and bin space became scarce. That change makes every method slower, and it particularly punishes late boarders. Some of the apparent gains from priority boarding are really gains from bin access, which is a distributional matter rather than an efficiency one.
"Random is better than back-to-front" is a narrower claim than it sounds. It holds against strict back-to-front with small groups. Coarse back-to-front with large blocks behaves more like random, because a big block is spread over many rows and recovers some parallelism. The general principle is about spacing, not about a rivalry between two named procedures.
The falsification test. If parallel stowing is the mechanism, then boarding time should fall as the average aisle-distance between consecutively boarding passengers rises, and a method that seats window passengers first should reduce time further by eliminating seat interference. If a strictly ordered back-to-front group boarded as fast as a scattered group with the same number of bags, the account here would be wrong and something else — walking distance, or aisle congestion in general — would have to be doing the work.
A picture of it
THE PICTURE #How to readStart at the top terminal and follow one passenger. The two diamonds are the two blockage types, and the loop back from Wait in aisle is the whole problem — it runs once per person queued ahead, so a procedure that clumps passengers runs it many times. The dashed edge marks which quantity the boarding order actually controls: not the stowing step, which is fixed, but how often the waiting loop fires. Note that the walking distance never appears, because it is not where the time goes.
What became clearer
WHAT CLEARED #Back-to-front optimises the appearance of order and the avoidance of passing, when the quantity that matters is how many people can stow simultaneously. Once boarding is seen as a parallel process rather than a queue, the ranking inverts: scattering passengers along the aisle is what buys speed, and the neat filling-a-glass procedure is the one that guarantees only one person works at a time.
Where to go next
ONWARD #- Why bin scarcity turned priority boarding into a distributional question rather than an efficiency one.
- How single-aisle and twin-aisle cabins differ, and why the second aisle changes the arithmetic.
- Why turnaround time matters more to low-cost carriers, and what else they compress to get it.