In Lander's introduction to Agent technologies she uses the terms "agency" and "agent". The manner in which this is done implies the terms are interchangeable though logic suggests they are not. While both terms' definitions are said to be incomplete and debated, little clarification is offered.

      Later on, Lander takes a broad view of agency. The characteristics she notes seem to be related to an agent. An agency seems more related to a function or a group of individuals providing some function than to a single component. Yet, based on Lander's statements, one can only conclude she means the terms agency and agent to be synonymous.

      I was surprised that XML was not mentioned as a communication language in addition to KQML, KIF, etc. Although XML is more general in purpose, it seems to serve the goals of the other languages mentioned. XML was around in 1996 (see www.oasis-open.org/cover/xml.html for details) but maybe it wasn't widely known about in 1997 outside of web and database worlds.

      The description of the coordination mechanism "Narrowing the search space" sounds very much like Lander's and Lesser's work in "Customizing Distributed Search among Agents with Heterogeneous Knowledge" but it is cited under "Explicit control representation". Since Lander was an author of the cited work, I can only assume she knows where best to categorize it, but I found the explanations unsatisfying.

      Overall, this was an interesting overview of agent based systems. It did not provide a great deal of technical information though. It seems best used to peek one's interest in multiagent systems and provide pointers to more information rather than providing a useful background of the subject. Since the title indicates the article will speak of issues in these systems, the article does seem to be true to its intent. The list of questions, or issues, the article presents are interesting but of limited use without some background knowledge of multiagent system, which the article does not provide. This is inconsistent with the beginning of the article which seems to try to be an introduction to multiagent systems.