There are two bars near my home in Studio City, CA, one called Clear and the other called Sapphire. They are located within one block of each other, and both aim at approximately the same demographic (single 20- to 30-somethings). When I first visited these bars, just over a year ago, Clear was the undoubted champ. On Friday and Saturday nights, people would line up and wait for an hour or more to get in; and even on weeknights, Clear could still get a respectable crowd. Sapphire, on the other hand, subsisted on Clear’s overflow – the people who got fed up with waiting at the door on weekends. On weeknights, Sapphire was almost deserted.
After frequenting Clear and (occasionally) Sapphire for two or three months, I went on a several-month hiatus. Now I’ve returned to the scene, and a transformation has taken place. Sapphire now dominates, with Clear picking up the scraps. On a weeknight, when Sapphire is about half-full, Clear is a ghost town. On weekends, Sapphire fills up first, with the overflow going to Clear. Sapphire does not usually have a line, though, so Clear’s overflow crowd can be substantial.
What happened? What we have here is a textbook case of network externalities. Network externalities exist when consumers’ satisfaction from a good or service increases with the number of other customers using it. In the case of bars and nightclubs, customers may care about ambience and service, but they care even more about meeting and mingling with other people – especially members of the opposite sex. As a result, customers choosing between two near-equivalent bars will tend to choose the one with the larger crowd, at least up to a point. The result is herding: one bar will attract large numbers, while the other remains largely empty. If people did not care about the actions of other customers, we would expect a more even distribution of customers across locations.
Network externalities set the stage for multiple equilibria. Just as left-side-driving and right-side-driving are both viable equilibria of the sides-of-the-road game, most-people-go-to-Clear and most-people-go-to-Sapphire are both viable equilibria of the choose-your-bar game. Which equilibrium actually occurs can result entirely from random factors (slightly more people happened to visit Clear early on, resulting in a snowball effect leading to Clear’s dominance), or it can result from historical factors (Clear might have opened a short while before Sapphire did).
But what explains the switch from Clear to Sapphire? I surmise that as Studio City became known as something of a Valley hotspot, more and more people found themselves having to wait too long to get into Clear. Eventually, Sapphire was getting enough overflow traffic that it reached the tipping point (another common feature of network externality situations). People realized that, at least on weekends, there would be enough people at both bars to make them worthwhile. Sapphire’s reputation as a place to meet-and-mingle caught up with Clear’s, and the expectations transferred over to the weekdays as well. For some relatively short period of time, people might even have randomized between the two locations, because either place could have had the better crowd on any given night. But that kind of equilibrium is unstable, just as randomizing over left-side-driving and right-side-driving is unstable. Very small differences in choices can tip the balance in favor of one of two extremes. In this case, a small handful of committed Sapphire patrons might have snowballed into the current Sapphire equilibrium.
UPDATE: Read my further analysis here.