1990s telecommunications bubble

The 1990s telecommunications bubble was a costly information cascade: a widely repeated forecast that Internet traffic would keep growing at 1,000 percent annually drove companies to build enormous amounts of network capacity, even after growth slowed. The episode illustrates how imitation can turn a shared assumption into a collective decision that is difficult to challenge—and how the failure to circulate and aggregate independent judgments can leave organizations with information they do not use.

The forecast becomes conventional wisdom

In the early days of the Internet, traffic was growing at 1,000 percent a year. Beginning around 1996, that rate slowed dramatically, as growth rates would be expected to do, but the change went unnoticed. The 1,000-percent figure had already become conventional wisdom, and the forecast of continuing explosive growth shaped what companies thought they needed to do.

The expectation translated into investment on a vast scale. Telecommunications companies spent tens, and eventually hundreds, of billions of dollars building capacity intended to handle the traffic they anticipated. Not investing appeared tantamount to business suicide. Even people who doubted whether the predicted volume would materialize faced a powerful counterpressure: everyone around them insisted that it would.

An information cascade

The bubble shows how a cascade can make decisions depend less on independent assessment than on other people’s choices. When a forecast becomes conventional wisdom, a person may follow the prevailing view despite private doubts. Each company’s willingness to invest then appears to confirm that the forecast is sound, reinforcing the pressure on others to follow. In this case, the crowd’s apparent agreement concealed the weakening growth rate rather than correcting the original assumption.

The account contrasts this process with collective decisions made by people who hold diverse opinions, reach independent conclusions, and rely on private information. Those conditions make it more likely that a group’s judgment will be good. In a cascade, by contrast, influential actors or those whose positions put them first in a network can shape the direction of decisions; later choices are no longer independent and may be profoundly influenced, even determined, by earlier ones.

The costs of unchecked belief

The conventional wisdom was seriously questioned only after the bubble burst, when most telecommunications companies were bankrupt or on the verge of bankruptcy. The scale of the infrastructure spending and the eventual financial distress turned a growth forecast into a cautionary example: widespread confidence did not establish that the underlying expectation was correct.

The episode also makes visible a broader organizational risk. Decentralized groups may contain specialized knowledge and independent perspectives, but that information is not useful to the whole unless it can be shared and aggregated. The telecommunications companies’ investment decisions, as described here, reveal the danger of a crowd in which people respond to the same prevailing belief without adequately testing it against changing evidence. The result was not simply that individuals held a mistaken forecast, but that collective action amplified it into a costly build-out.