Two and a half years ago noted economist Steven Landsburg wrote that the economics department of the Univ. of Rochester spent a week struggling to explain why people don't walk up escalators.

My colleague, Steve Margolis, helped out the poor, befuddled Rochesterites with this penetrating analysis. He has now done some field research in Washington, D.C. and he reports as follows:

I've noticed in the Metro stations that people do, in fact, walk up and down escalators. This fits with my explanation of Landsburg's observation that people don't walk on escalators. People who use the Metro typically confront eight escalator rides a day--one into the station, one down to the trains, one up from the trains, one up to the street, and then repeat that to go home. Pretty soon, they become comfortable with escalators. For them, the transition problem goes away. For this one physical task they're practically acrobats. They walk on escalators and it saves them a lot of time.

Not everyone in the Metro walks up escalators, but many do. They seem to be more likely to walk up long escalators than short ones. Any explanation that has to do with the time saved per step versus the effort expended won't explain that, since both of those are proportional to the length of the ride. But initiation and termination are costs that are invariant with respect to the length of the ride, and therefore relatively less important in long rides.

If my explanation is right, we should be able to observe an amended version of Landsburg's claim. In places where many of the users get little practice on escalators, such as airports and suburban department stores, people don't walk on escalators.