In the mid-1980s, when even computer scientists had difficulty imagining a broad use for the fledgling Internet, a cluster of information science pioneers began troubleshooting how to categorize everything.
It was a small group of people — just 30 or 40, remembers Craig Partridge, a young scientist at the time who became involved in the discussions. Many of the conversations about the future of the Internet happened not in formal meetings but in casual memos or hallway conversations.
At first, scientists thought top domain names would reflect individual institutions: the Massachusetts Institute of Technology would be found at .mit, for example, rather than at .edu. Eventually, a discussion group, led by the late Jon Postel at the University of Southern California, decided that geographic location was not as important as general classification: What mattered about MIT was that it was a university.
These decisions ultimately defined a major organizing principle of the Internet: Web sites are categorized in the broadest possible terms. If users can remember what comes before the dot, guessing what comes after it is fairly simple: If not .com, then .net; if not .net then .biz.
"Originally it was going to be .cor for corporate," instead of .com, remembers Partridge, now a chief scientist at Raytheon BBN Technologies (he helped design, among other things, how email is routed), but that idea was scrapped for reasons he can't remember. In the mid-1980s, there was a meeting to nail down the final list of top-level domain names. "Dot-com, dot-edu, dot-org," Partridge begins to tick them off. "Dot-net, dot-ARPA — and that may have been the full list. It was definitely no more than 10." It was as though only five or eight area codes were invented for the telephone.