Messages entering an educational information distribution network may be ranked for an ordered introduction into the network to maximize the timeliness of message arrivals over the set of users. Electronic mail, EDI documents, and broadcast news may be ranked by the users who choose to examine those messages of interest or benefit to themselves. Individuals choose to examine messages based on the message's features; a decision theoretic user interface may likewise examine features and determine which unexamined messages to present to the user. This interface may learn or estimate parameter values from user supplied relevance feedback. This knowledge may be sent to a gateway or bridge that similarly ranks messages based on their feature values and the knowledge the gateway has gained from the user supplied relevance feedback. Means to estimate parameters and the expected performance of the gateway ranking system can be computed algebraically. Future work on this approach to ranking of messages before they enter the network will involve further analyses of the efficacy of ranking. (17 references) (Author/DB)