Before Internet routing became invisible infrastructure, one very practical problem was simply figuring out how to get from one Unix machine to another through a chain of dial-up connections.
The Stargate Internet Museum’s page explains the problem neatly. Steve Bellovin had written pathalias, a program that could create routing information if it had a map of the network.
The problem was that no complete map existed.
UUCP networks were built from relationships between individual systems. One machine knew the machines it called directly. Those machines knew others. Mail and files could move hop by hop, but finding a good route across the larger network required knowledge that no single operator automatically possessed.
In January 1984, a birds-of-a-feather session at the USENIX conference in Washington, D.C. recruited more than thirty volunteers to build and maintain a map.
The resulting effort became known as the UUCP Project.
The network had human maintainers
The project divided responsibility among regional volunteers who supplied and maintained connectivity information. The maps were distributed through Usenet, eventually in the comp.mail.maps newsgroup, where other systems could collect them and generate routes.
Initial funding came from USENIX. When that funding ended, the work continued on a volunteer basis.
The museum page names Mark Horton as the person who ran the project, with major contributions from Mel Pleasant, Tim Thompson, Berry Kercheval, Steve Morenberg, Karen Summers-Horton, and a much larger collection of regional coordinators.
That structure is worth remembering because modern networking makes topology feel automatic.
In the UUCP world, the network map was partly a social artifact. Humans knew which systems existed, which links worked, who was responsible for which region, and how changes should be published so everyone else could calculate a path.
It was infrastructure maintained through cooperation and text files.
The Stargate Museum continues from this page into surviving 1984 map entries, which makes the exhibit more than a summary. It points directly at the kind of data that made the system work.
CacheRat’s 1,967 Ancient Web Domains research list includes pages like this because network history is easy to flatten into a straight line from ARPANET to today’s Internet.
The real history contains a lot more phone calls, volunteers, hand-maintained maps, and machines asking their neighbors for directions.