[MUD-DEV] TECH : MUD game on a multi agent system

[Robert Zubek]

Cyril MudDev wrote:

> I want to make a MUD based on a multi agent system (which complies
> with the FIPA specification). Does anyone know of a similar
> project?  (a MUD based on the multi-agent technology) Does someone
> know wich problems I may encounter?

Hmm, no, I don't know of any MUD agents based on FIPA. There may be
one or two 'personal assistants' out there based on KQML - take a
look at

  http://www-ksl.stanford.edu/knowledge-sharing/agents.html

Actually, while protocols such as FIPA or KQML have standard
specifications, they may be suboptimal for use in MUD development.

For one, they're both tailored to particular domains, with different
communicational requirements from online worlds. FIPA appears to be
optimized for financial transactions (buying/selling web agents,
etc) and KQML for exchange of propositional knowledge (such as
assertions stored in a truth maintenance system). Also, both
protocols seem to be based on a logic-based view of communicative
acts (take a look at KQML specs for some examples), which imposes
additional constraints. I suspect the kind of specificity these
protocols afford doesn't make them well-suited for modeling
communication in game worlds - it seems unlikely that either
protocol would make it easy to express the kinds of information that
are important in a game.

Secondly, and more importantly, those specifications are in a
certain sense 'weakly grounded', in that they spell out the syntax
of the communication protocol, but say little about the semantics of
usage. This is highly problematic - it's as if one tried to use
English knowing only grammar rules, and not what sentences actually
mean or how to participate in a conversation. This results in
implementational 'babelization' as every group implements their own
vision of the semantics of the protocol, and the visions don't
always coincide. See http://wwwis.win.tue.nl/acl99/

So in the end, it's not clear how helpful these protocols would be
in game development - they specify mainly the syntax and not the
semantics of communication (the latter is the really difficult but
also really interesting part, IMVHO :), and even the syntax is
probably ill-suited for application in MUDs. But I only hope I'm not
starting off from incorrect assumptions about your project - can I
ask if you have any particular reasons for using a 'standard' query
protocol such as FIPA or KQML, rather than your own custom-tailored
one?

Rob

--
Robert Zubek
rob at cs.northwestern.edu
http://www.cs.northwestern.edu/~rob


_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev