Re: Back to the Future of Swarm

[glen e. p. ropella]

sthomme> Just to give the denizens of the list another data point:

sthomme> what I would have used if Swarm had not become available was
sthomme> Modsim II from CACI, Inc. That's now Modsim III. It is a
sthomme> Pascal-like language with built-in constructs for simulation
sthomme> plus the requisite support libraries. It compiles down to C++
sthomme> as its 'assembly' language, making it easy to link in C++ and
sthomme> C code. They support MS VC++ under NT, and native compilers +
sthomme> GNU on certain Unix platforms. (But not under Linux.) It is
sthomme> by now a fairly mature product; the graphics abilities are
sthomme> superior to Swarm's, but I'm not sure the scheduling support
sthomme> is. [I haven't done a feature-by-feature comparison lately.]

sthomme> The company will let an academic user have a (perpetual!)
sthomme> single user license for $1500; a departmental 20-user license
sthomme> is $3000. After one year, the (optional) support contract is
sthomme> $3-400 per year.

This is one of the reasons I don't regard ModSim as very 
useful at all.  It's not worth the price.  This is the
same type of situation Apple found itself in (and has
recently come around).  Yes, it's high quality.  Yes,
it's easy to use. And, yes, it's aesthetically pleasing.
But, most of us are cheap b*st*rds and aren't willing to
pay $120 for a good pair of shoes, much less $1500 for a
simulation package.

And this is not restricted to academics and hackers.  There
are several companies who have a seriously difficult time
justifying the purchase of such a rich-man's package.  Even
if the workers want such a beast, it's a monumental task 
trying to convince finance that you can't get by with
something much cheaper written in Visual Basic or Excel. [grin]

sthomme> Chris Langton suggested at Swarmfest that they'd had feelers
sthomme> from companies willing to lay out serious bucks for specific
sthomme> simulations. If that's true, perhaps the above numbers are
sthomme> relevant?  If there's enough demand for specific application
sthomme> development, perhaps some of those bucks could support the
sthomme> development of the base Swarm system ...

I think they probably are relevant numbers.  But, as Scott said,
the important difference between the idea of a company formed
by the Swarm team and something like CACI is that we would be
selling a service, not a piece of software or a site license.

sthomme> Of course, what you *don't* get going with a commercial
sthomme> product like Modsim is: (a) any say-so over product
sthomme> development (unless you are a big commercial licensee); (b)
sthomme> the opportunity to exchange deep thoughts and/or witty barbs
sthomme> with guys like Glen and Manor and Roger; (c) a diverse and
sthomme> interesting user group like this one!

I think the reason something like ModSim could never have
as dynamic a user community as ours is because we're not
marketing technology (and, yes, we are marketing Swarm....
just in a different way).  We're marketing information.
We're supplying the tool for free and then trying to teach
people how to use that tool.  Right now, we have a very 
broad audience and customer...the public.  This means that
our resources are spread too thin over a large population.
The company (if formed) would be capable of tightening 
the information beam, thereby providing a more steady and
predictable product (the information embedded in the "teaching").

And, the objections to this are absolutely correct, the
tightening of the beam focussed on commercial customers
would eliminate alot of coverage of the broader public 
audience.  But, what a company could ensure is that some
measure of diffusion occurred *through* maintenance of 
the tool.  After all, high quality software reduces the 
amount of information it takes to use that software....
maybe not by all that much...but it does reduce it somewhat.

(Barry and I had a long drawn out argument about the purpose
of software and what "high quality", "well-written," "engineered,"
etc. software *means*.  But, I still think my above statement
is correct. [grin])

glen