Re: [Swarm-Modelling] Good reviews with the info the greenhorn needs?

[Maarten Sierhuis]

Amen Glen!

This reminds me of a recent debate we had here at NASA, about similar agent-based tools we have in our organization. Here is a quote from a wise person (not me) in our organization:

"At a higher strategic level -- drawing on some advice I got once from our good friend

[name deleted] -- let's keep our focus on *problems* and not on *methods*. We should not

be drawn into an argument about what [Tool X], or [Tool Y], or [Tool Z] can or cannot do in

the abstract. Rather we should demonstrate that [Tool X] can solve the actual problems

that [name deleted] and his team are presented with. If [Tool X] has solved these problems on time

and on budget, it becomes moot whether or not [Tool Y] or [Tool Z] might be able to solve

some of them.

That said, we always want to be armed with knowledge about these other systems, in

case we have to field reasonable or unreasonable claims about their capabilities. That's

why I found the Kieras survey useful (reattached here) even though I would probably quibble

with most of the details."

Doei ... MXS

_______________________________________________________________________

 Dr. ing. Maarten Sierhuis                 USRA/RIACS

 Senior Scientist                               Mail Stop B269-1

 Human-Centered Computing            NASA Ames Research Center

                                                      Moffett Field, CA 94035

 e-mail: address@hidden              

 Phone: (650) 604-4917                                                                                                                       

 Fax: (650) 604-4036

 http://homepage.mac.com/msierhuis

 http://www.agentisolutions.com

 http://ic.arc.nasa.gov/projects/brahms/index.html

 http://www.riacs.edu

_______________________________________________________________________

This communication is intended for the use of the addressee only and may contain information that is privileged or confidential.  If you are not the addressee, you are hereby notified that any dissemination, distribution or use of this communication is prohibited.  If you received this communication in error, please destroy it, all copies and any attachments and notify the sender as soon as possible. Any comments, statements or opinions expressed in this communication do not necessarily reflect those of NASA or USRA/RIACS, its subsidiaries and affiliates.

On Nov 23, 2006, at 12:00 PM, glen e. p. ropella wrote:

Rob Bowers wrote:

Are you saying that e.g. NetLogo, AgentSheets, and Swarm all have the

same uses, if only the user finds them?

No.  I'm only pointing out that "usefulness" is user- and task-

specific.  It is quite difficult to address these issues in the

abstract.  What's always required is to examine the concrete, detailed

needs of any given user and task before a reasonably accurate

recommendation can be made.

Hence, Gary's advice about consulting someone who's been through it a

few times is the most practical advice.  It's much more efficient than

reading reviews _when_ the core requirements for the tools are unclear

or in contention.

purposes. What I think will especially help the new user is something to

read to allow him to judge the suitabillity of a toolkit himself.

And before investing in learning it!

I agree completely.  And I don't disparage the reviews that are out

there, especially side-by-side validation studies.  I think Mike North

did some work awhile back comparing Repast and Swarm directly in the

context of a concrete model.

But, what I'm trying to point out is that this sort of abstract

evaluation will never substitute for tacit knowledge that comes from

actually _trying_ to build a model in any given tool.  And ultimately,

that tacit knowledge can be very different from the abstract knowledge

one infers from a written review or biased comparison.

Is everyone to learn the use of

every toolkit, and the programming languages they're based on? This will

forbid its use to many people who see value in simulation, but happen

not to be programmers. (Don't get me wrong - learning to program was my

best move in 10 years).

Unfortunately, yes.  Computer modeling is (and will remain for the near

future) an inherently _technical_ task.  The only thing that really

changes between tools is the language.  One tool uses C.  Another uses

diagrams and paradigms (e.g. Ptolemy II).  Yet another focuses on XML.

But, in the end, as Marcus pointed out, they all come down to the

completely unambiguous technological context of Turing complete

programming languages.  That bottom, hard context always finds a way to

bubble up into the model regardless of what layers of abstraction the

tool kit provides.

The best we can do with current technologies is to _facilitate_ the user

by maximizing ease of use.  And we do that by keeping the tools as clean

as possible, providing as many examples as possible, delineating useful

patterns and idioms where we can, and providing opinions when asked.

Now if every new, non-programmer user will begin with analogous

questions, I suspect that somewhere they will have been brought

together. The question is where...

This gets back to a perfect task for the SDG.  The big task is to

provide guidance on the uses of ABM (regardless of which tool is used).

 The _first_ step in that task is to clearly delineate where Swarm is

efficient and effective and where it's not.  The next step would be to

state the original and evolved requirements for Swarm.  And the next

step is to place those requirements in the space of requirements of

other packages like Mason, Arena, Simulink, etc.

-- 

glen e. p. ropella, 971-219-3846, http://tempusdictum.com

And therefore the victories won by a master of war gain him neither

reputation for wisdom nor merit for valour.   -- Sun Tzu, "The Art of War"

_______________________________________________

Modelling mailing list

address@hidden

http://www.swarm.org/mailman/listinfo/modelling