Re: First annual report, Swarm Development Group

[Marcus G. Daniels]

>>>>> "SR" == M Lang / S Railsback <address@hidden> writes:

SR> It's not clear what we're to infer about this from your report- it
SR> sounds like money is scarce. On the other hand, there seems to be
SR> little effort to recruit memberships. Are memberships still
SR> important to SDG? 

I'd really like to say that the money raised from individual
memberships give me an excuse to pay attention to questions from
those members, and that members get some value from that.  
What seems to happen, funnily enough, is that the members tend to figure
things out by themselves and even do helpful things like submit bug
fixes.  It was never really anticipated that the money from individual
memberships would cover more than a proceedings and a CD-ROM, and a
bit of support.

Money for salaries must come from sources other than individual
memberships.  We had hoped to get a a half-dozen government or
corporate group members, and judging from the constituency of the
Swarm mailing lists, it still seems like this ought to be within the
realm of possibility.

Right now, the SDG is running on the last stage of money from a grant.

So, if there are groups out there that just think they can get software
and support for free indefinitely, let me assure that is not the case.
In a few months, we either be forced to turn inward to fundable
research grants, partner with companies that do agent-based models,
or quit entirely.

Also please keep in mind the mission of the SDG is to support the
development of software for the general betterment of the agent-based
modeling community.  The SDG exists to find good solutions to specific
problems, the problems of agent-based modelers.  Traditionally we have
done that by adding features the Swarm simulator.  Currently we are
contributing to a compatible next-generation system, IMA.

When Swarm first started, there were two rough ideas about how the
project might evolve.  One was that some group with a commercial
agenda (presumably VC-funded) would come along and implement something
like Swarm in a high-energy and very polished way.  The concept would
be proven and then everyone would then move on to other research.

While there have been several folks with commercial ambitions along
the way, their main characteristic has been their ambitions, not their
deep resources and ability.  For a while, Chris Langton and Glen
Ropella looked at such an enterprise, but they eventually found it did
not match with their personal and scientific goals.

The other future that was imagined for Swarm was to build a
self-sustaining user community, something along the lines of Debian
GNU/Linux.  (Debian is not a vendor, per se.  First and foremost,
Debian is the collaboration of hundreds of volunteers.)  In this
model, commercial users would be free to do pretty much whatever they
wanted with the software, but this latitude would come at the cost of
responsibility of using the resources of their organization to fix and
improve the software.  The idea being that the most productive thing,
in general, would be to cooperate with other people using the
software.

We now find ourselves in a different future.  Swarm contributes to
scientific progress and commercial success in the domains of the its
use, but most of the organizations involved apparently don't see it in their
interests to do their part to keep the project moving forward.
(Lang, Railsback & Assoc. not included!)

As you point out, there are a number of companies that have
investments in Swarm models.  It would be nice if there was a way we
could work together to a mutually-beneficial outcome.  For the SDG, a
good outcome is that we advance the cause of tools for agent-based
modeling in some clear, useful way. 

Regarding small projects: as I am the only employee of the SDG,
taking time to work on a particular model or infrastructure that only
makes sense in the context of a particular model is time that doesn't
serve the basic cause of the organization, or even its survival.
So, for me to work on model or model substrate technology, it pretty much
has to come down to either substantial money for the SDG, or that the
model has some pedagogical value as a good freely-redistributable demo, etc.

One of the large tasks after the next release is to develop a
web-based GUI for Swarm.  I'm just starting on this (today, actually)
in regards to the IMT plug-in for Netscape 6 -- probes are in-memory
document fragments that can be visualized using either style metadata
(CSS), or features from a widget library.  Anyway, to really hash out a nice
GUI, it would be beneficial to develop it in collaboration with
someone working on a largish application or two.

Another big task, one that I won't pursue unless there is dedicated
funding, is Swarm integration with ArcInfo 8.  Via COM, it is
possible to represent agents in ArcInfo 8 geodatabase landscapes and
visualize them.  This would involve adapting a variant of the XPCOM
support in Swarm for Microsoft COM, and then building some models
either with the IMT infrastructure (adding a GIS plug-in) or, IMO
less-desirably, using Microsoft supported languages like Visual Basic,
Java, or C#.

The SDG directors have been throwing around some other development
tracks that might interest potential collaborators or partners. 
I'll post those as soon as these ideas are nailed-down.

                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.