Week 5 Lab: Lab 5 — One Line or Six

Run the bank experiment with real humans, and find out why the fairer system is the one people complain about on sight.

22 March 2027

Before the lab#

Read the week 5 notes. Wear something you can stand up in for an hour, because you are the customers.

In the lab#

We run the 1970 bank experiment on ourselves. Twenty-four students, six “tellers”, one arrival script read off a stopwatch so both runs see identical arrivals.

Run one: six lines. Pick a teller when you arrive. You may not switch — no jockeying, which makes this the generous version of parallel queues, not the realistic one. Each customer records their own wait.

Run two: one line. Same arrivals, same service times, single queue feeding all six.

Then the numbers. Put every individual wait on the board as a dot plot, both runs side by side. Two things fall out, reliably, every year:

The single line wins on the mean, and by more than most of the room predicted beforehand — we take predictions first, in writing, precisely so that you cannot later remember having expected it. The gain is not speed of service. It is that run two never has a teller standing idle while somebody waits, and run one does, constantly.

The spreads are further apart still. Run one has someone who waited four times the median. Run two does not have that person, and cannot. Ask yourself which of the two differences you would notice as a customer, and which one a manager would be able to see in their reporting.

The subjective round. Finally, before you see the data, everyone writes down which run felt worse. A reliable minority say the single line, because thirty people look like a longer wait than five do, and because in six lines you get the pleasure of believing you chose well. The gap between the felt ranking and the measured one is the practical problem with deploying good queue design, and it is not solved by being right.

Afterwards#

Your ordering-rule assignment is due at the end of next week. If you are stuck for a system, the dot plots on that board are twenty-four separate arguments for looking harder at somewhere you already go.