In the summer of 1996 I received a grant from the Intel corporation through Grinnell College to intern at the National Public Telecomputing Network (NPTN), an organization of CNs referred to as "Free-Nets" (SM). This internship occurred at an interesting time -- NPTN was already having financial difficulties such that several management positions were open, but we did not yet know that it would be bankrupt by September. I worked primarily with the Rural Information Network (RIN) program, configuring, shipping and supporting electronic community centers for small towns. These systems did not seem ideally suited for many of the communities for which they were intended, and I found that very little research had been done into what features made a Free-Net succeed or fail. I resolved to at least try to find that out.

The Cleveland Free-Net (CFN) in Cleveland, Ohio, is the oldest and largest Free-Net, using FreePort software under the UNIX operating system. The Great Lakes Free-Net (GLFN), located in Battle Creek, Michigan, is middle-aged and middle-sized, one of the first RINs and the very first Free-Net to use a graphical user interface (FirstClass) and a Macintosh computer as its server. Finally, the Worth County - Sylvester, Georgia Free-Net (WC-SGFN) is less than two years old and very tiny, and its hardware and software (a Macintosh computer running NovaServer) are nearly identical to those in the new RINs I was preparing. Between the three of them, I felt I had a complete sample of the hardware and software setups used by Free-Nets, as well as a cross-section by age and size.

Most of my actual research, aside from first-hand observation, took the form of a survey of the Free-Nets' users. I chose subjects on the basis of posts they made to the systems' message forums, so my sample was neither random nor representative, but I tried to take demographic information into account while interpreting the responses in a strictly qualitative fashion. Survey questions -- in the form of statements with which the subjects could agree or disagree -- ranged from the abstract (A sense of community is necessary for maintaining a safe and pleasant neighborhood) to the specific (The Cleveland Free-Net would be better overall if it had a graphical interface) to the irrelevant (Cleveland's mass-transit system is adequate). Between 0.01% and 0.53% of each Free-Net's usership was surveyed.

The paper goes into more specific detail about the strengths and weaknesses of each Free-Net than I could summarize in this abstract, but the general findings were as follows. Computer use is not in itself antisocial; all three Free-Nets improved their communities. Access to the system must be universal if the entire community is to benefit. Emphasis must be on local content, or the systems will come into unflattering comparison with commercial Internet service providers. The management style of a CN is as much an effect as a cause of its success. CNs may follow a "life cycle" of growth and decay similar to that of cities, but my evidence is not conclusive on this point.

I believe strongly that further research into existing community networks should be a priority of any organization seeking to follow NPTN's lead in establishing new CNs. The age of pioneering is largely over, and while this means that reinventing the proverbial wheel in each community is a waste of resources, it also means that we have no excuse for giving communities square wheels. If the community networking movement is to continue into the future, it must learn from the successes and failures of its past.