pushing tin

The delightful thing about geeks is that, when they become annoyed, they become obsessed with solving a problem. Julianne at Cosmic Variance points out an example: Jason Steffen's computer modeling of an ideal passenger loading scheme for airlines (available on arXiv).

Steffen's first insight is that the commonplace notion of loading from back to front rather than front to back or randomly doesn't address the main bottleneck of passenger-loading: waiting for people to stow their baggage. Indeed, in his model, back-to-front is the second worst loading method examined--only marginally better than the front-to-back worst-case scenario. The key observation is that in each case, only one passenger is able to stow baggage at a time. The first passenger in line begins stowing baggage immediately, but the second person in line has to wait for her to finish before he can begin stowing his bags--and as a result he blocks the third passenger from stowing her bags. Even random orders performed better--by half--because several passengers were able to stow bags at once.

The ideal depends on aircraft layout, but in the case of rows of 6 passengers, the best scenario is to board in six stages, each ordered back to front: 1) even rows of wing seats, 2) odd row wings, 3) even center, 4) odd center, 5) even aisle, and 6) odd aisle. This can result in a six-fold decrease in loading time.

Realities such as families traveling together, recalcitrant passengers, and psychological effects (front-to-back or back-to-front are more psychologically satisfying than random loading because despite their real disadvantages, it seems) are, of course, all ignored.