There are many possible ways to measure the relative "importance" or "size" of LISTSERV sites: number of lists, number of subscribers, number of messages delivered, etc. In most cases, the number of messages delivered is used to rank mailing list managers, because it is the most accurate measurement of the amount of activity on the server. But, depending on what you are looking for, other variables may be more appropriate. There are currently three LISTSERV "Top 20" lists, ranking the various servers according to:

• Number of messages delivered by the server

• Number of lists hosted by the server (not available for May 1995)

• Number of subscribers (not available for May 1995)

What do these numbers mean exactly?

The statistics are gathered daily by a monitor at L-Soft which polls all the LISTSERV sites that participate in the "open" LISTSERV network over the worldwide Internet. Corporate servers operating in a "closed" environment are not included.

The statistics are organized in monthly reports. The message volume figures are a flat average of the accumulated traffic for the entire month. Thus, a server that delivers 300,000 messages per month will show a daily average of 10,000 messages. In practice, there is a lot less traffic during the weekend, and even during business days there can be significant fluctuations (~30%). The list statistics (number of lists, number of subscribers, largest lists in the network) are measured as of the last day in the month.

For the current month, the message volume figures are averaged based on the number of days that have elapsed. In practice, the figures are not very meaningful in the first few days, especially if the month starts on a weekend. In that case it is best to refer to the previous month for more accurate figures.

A fourth "Top 20" list is available for the more technically oriented: Number of messages generated by the server. This corresponds to the number of mail messages created by the server while processing postings to its mailing lists. LISTSERV delivers its mail using an intelligent mail delivery function called DISTRIBUTE, which improves mail delivery times and reduces bandwidth usage by spreading the delivery chore over scores of participating servers. Each of these servers processes the recipients for its service area (typically its organization or campus). In most cases, DISTRIBUTE simply alters the manner in which the mail is delivered, and the number of messages forwarded to other sites for optimized delivery is roughly equal to the number of messages received by other sites for the same purpose. In some cases, however, this is not the case, either because the server elected to serve a much wider service area than its own organization, or because it is explicitly forwarding all its deliveries to another machine (for instance, many mainframe-based servers forward their mail to a workstation or PC for delivery).