In examiming the agent literature, Tom has found two essential new concerns.
One, a qualitative difference with software engineering concerns in the
past, namely, agents combine a large number of underlying technologies, and
are hence very complex artifacts. The second is a real difference in
relation to past software engineering concerns, i.e., the use of
introspection in the implementation of an agent. The use of
anthropomorphisms in agent technology may be dangerous because it might lead
to an illusion of understanding of concepts which does not actually exist.
For example, I have never found a proper definition of the concept of
autonomy, at least one that can be used for the design and analysis of
agents. So, for me the most interesting rsearch challenge in relation to
agent technology is the management of the complexiity of multiple
technologies referred to above. The use of coordination technologies might
be an aid to deal with this complexity.