New communication applications on large scale networks such as the Internet will be driven by mobile computing and information access. We have defined a network programming model called Mobile Assistant Programming (MAP) that supports the development and execution of applications for efficient information access on the WWW. Mobile assistants are high-level interpreted programs that can move between nodes, create clones and report results. Their execution is asynchronous and persistent to allow client disconnections and survival of node and data link failures. Using an implementation of MAP based on the World-Wide Web framework and the Scheme programming language, we have developed an application that searches for relevant HTML documents on a set of WWW servers. Our experience with the MAP and the WWW application shows that significant performance improvement can be achieved by moving computation closer to data: both the elapsed time and network traffic are reduced.