The first stall already ran the experiment.

Last time, a simple model of two sellers on a beach put the second stall next to the first. It assumed the trader already knows the spot is good. He does not. Nobody hands him a footfall count. He has small savings and one attempt at a pitch, and the only evidence in front of him is the stall already trading there.

That stall has been at the foot of the LRT stairs for six months and is still there. Without meaning to, it has tested three things. Enough people want nasi lemak here. The crowd on this spot stops and buys, instead of walking past on the way to the train. And the enforcement officers leave this spot alone, or the licence holds.

Opening next door inherits all three answers at once. The alternative is to spend a month of stock proving them again on an untested spot.

People call this copying: one trader sees another doing well and follows without thinking. The second trader is reading the first stall's results. Economists call this rational herding. Bikhchandani, Hirshleifer and Welch modelled it in 1992: when your own information is thin, moving with a choice you have watched work is the sound bet.

The anthropologist Clifford Geertz, writing about the bazaar in Sefrou, Morocco, said the trader's main problem is finding out what the options are. Where nobody publishes prices or footfall, the stall next door is the nearest thing to information.

Malaysia's local authorities did something similar, a trader told The Sun: they mostly put MyKiosk stalls, the government's roadside kiosks, where informal trading was already happening.