Respondents most typically get the documents from a mailing list. 838 (53%) reported receiving documents directly from the MIT-system, 403 (25%) through another Internet mailing list, 55 (3%) through a local distribution network based mailing list, and 23 (1%) from a non-internet-mailing list. These numbers account for 82%of total respondents, more than 90%of respondents getting the documents. The figures mask some duplication since respondents could list more than one source, but it is reasonable to assume few people would subscribe to several mailing-lists to receive the same information. (On the other hand, most of the people who receive the documents through commercial email or a phone-network are included in these numbers.)
Respondents were much less likely to get the documents by retrieving them from servers.
One might assume that most of the duplication of sources is by people who are not on mailing lists, but retrieve the documents from various servers. People on mailing lists would have little need to do retrieval, except when they want a particular documents that might not be on the sublists to which they subscribe. It is more speculative to identify these retrievers with the 20%of customers who reported that they read the documents at most once a week (compared to the the 80%who look at least several times a week. These possibilities will be tested through more fine grained inspection of the data.
The profile of the respondents in regard to sources is not representative of the entire group of recipients. People on mailing lists delivering the documents were more likely to get a survey than people who retrieve documents from servers, commercial forums and other public facilities for single shot use. So each report of retrieving represents more people than a report of receiving documents from a mailing list. This relation should be strongest for reports of retrieval from bulletin boards since their users often lack access to email.