In airports one often finds oneself standing in a central line that splits into many smaller lines.  For example, a single long line for security screening splits near its end into many small lines in front of each screening machine.

How should individuals in the central line be allocated to the many small lines? One obvious solution is to let individuals in the single long line choose the smaller line in which to continue.  Some people might choose badly, or be stressed by having to make a choice, but overall this approach requires little supervision, and competition between choosers should make lines roughly equally bothersome.  All else equal, if line A was consistently faster or more pleasant than line B, line A would attract more people until the added length on average cancelled A’s advantage.

Many airports assign employees to the task of sorting people into lines.  In some places the line sorter always tells people where to go, while in other places the line sorter usually lets people choose, and only sometimes intervenes to tell people where to go.  As far as I can tell, line sorters are never purely advisory; either then let you choose without offering advice, or they use their authority to command compliance.

On a recent trip, I had several occasions to observe myself being sorted from a single central line into a particular line, and then to track how long it would have taken me had I been free to choose a line for myself.  It seems to me that on average the lines I would have chosen completed faster than the lines to which I was assigned.  Now I didn’t collect formal data, and I could just be responding here to random luck, but I suspect not, because a theory occurred to me that would explain this pattern.

My theory: lines vary in speed due to variations in personnel, but line sorters prefer lines to take similar amounts of space.  Some employees just take more time when, for example, working a scanning machine.  When people notice that lines vary in speed, they naturally prefer the faster line, which will make that line longer in terms of the space it uses.  Line sorters then “correct” this imbalance by sending people to the spatially-shorter but temporally-longer line.  Sorted folks then regret being sorted, knowing that on average their trip would be faster if they could choose for themselves.

Now if there is limited space for final lines, it might make sense to assign someone to limit how fast people can move from the main line into final lines. I suspect, however, that informal custom would usually work fine – when the best final line got full, the main line would usually just stop and to wait for room in that best line.  If space were limited in the main line it could then make sense to assign someone to make sure all available space in final lines was used. But this should be pretty rare – it would usually be far cheaper to just have more space for the main line.

I suspect that what is really going on here is that orgs who manage lines are embarrassed by variations in personnel productivity.  It seems unseemly to egalitarian ex-foragers to let line-walkers publicly endorse some personnel as faster.  They’d rather pay to try to “hide” this variation, even though most folks in line will probably notice it anyway.

Another explanation is that variations in line length seem “chaotic,” suggesting that the line-managing org does not fully control the situation.  So “security theater” requires the illusion of control, via equal length lines.  This would better explain systematic differences between security and non-security lines. If true, this seems an interesting contrast to the “autonomy” literature, such as in medicine, where many say we should let people make decisions for themselves even when this hurts them a bit on average, so that they gain the benefits of feeling in control of their lives.

Added 1p: If some org personnel might conspire with folks in line, the org might reasonably want a random matching of folks with final lines. This doesn’t explain cases where folks can usually pick their line, but are assigned to lines when line lengths are especially mismatched.