BKAGTCHB.RVW   980315

"Agent Technology Handbook", Dimitris N. Chorafas, 1998,
0-07-011923-6, C$63.95
%A   Dimitris N. Chorafas
%C   300 Water Street, Whitby, Ontario   L1N 9B6
%D   1998
%G   0-07-011923-6
%I   McGraw-Hill Ryerson/Osborne
%O   C$63.95 800-565-5758 fax: 905-430-5020 louisea@McGrawHill.ca
%P   382 p.
%T   "Agent Technology Handbook"

In the computer virus research community there is a long-standing
debate about the existence of "good" viruses.  Some maintain that no
self-reproducing program can be other than evil, while others maintain
that it is possible to create viral types of programs that can perform
useful functions and harness the power of multiple processors, and
processes, in order to solve difficult problems.  To date the debate
is indeterminate, and recent discussions seem to turn on the question
of whether a viral program can perform any better than, say, network
software launched and controlled from a central source.  This
deliberation is germane to agent software because the same type of
question arises.  Can it be shown that agent software, by its nature,
has useful advantages, and, if agents are programs that are useful,
what is the distinctive of agent software?  To date, expositions on
agent technologies have not dealt well with these queries.

Part one is a conceptual background to the topic.  For any useful
definition, agents must have a certain autonomy and act independently
of, although on behalf of, the user.  Chapter one looks at a range of
relevant work in the field of artificial intelligence, necessary for
that autonomous action.  "Facts and Opportunities for Knowledge
Workers" is supposed to be the theme for chapter two.  Knowledge
workers appear to be defined by a need for multimedia, virtual
reality, banking, or micro-manufacturing, and the chapter returns to
more AI work.  Chapter three, entitled "Thinking Machines," turns
again to artificial intelligence research, concentrating on those
areas that might be considered to be knowledge based or expert
systems.  Overviews of KQML (Knowledge Query and Manipulation
Language) and Telescript are given in chapter four.

Part two is intended to provide examples of the things that agent
technology could accomplish.  A variety of experiments and projects in
telecommunications are looked at in chapter five.  The discussion of
computer security in chapter six deals primarily with failures of
security, of which there are no shortage of examples.  Chapter seven
defines software-hardware codesign for us, but the miscellaneous
activities described in the piece bear little resemblance to the
definition and almost none to each other.  Network management is,
interestingly, one of the most often cited uses for beneficial viral
software as well, but chapter eight, once again, contains more stories
of failure of standard systems than success with agent software.

Finally, part three proposes to show how to implement agent
technology.  Much of chapter nine is a rambling, although interesting,
philosophical discourse on the nature of creative thought as much as
anything else.  A similar lack of focus pervades chapter ten, engaging
as the material may be.  Chapter eleven presents a number of difficult
problems for automation, many of which have been presented previously
in the book.  Some of the underlying structures for problem analysis
are described in chapter twelve.  Chapter thirteen presents another
meander, including the importance of information to strategic
planning, the year 2000 problem, and Web programming languages.

I am not sure what to make of the fact that there are almost thirty
pages of acknowledgements before the book starts.  On the one hand, it
is nice to know that the author is grateful to those that helped him. 
On the other, it is rather jarring to note that, in spite of extremely
large headings and the inclusion of many addresses, there must be
almost one person to thank for every page in the book.

At the conclusion of the book, the questions outlined above remain. 
Agent technology persists as an undefined and unproven field.

copyright Robert M. Slade, 1998   BKAGTCHB.RVW   980315