Re: Open source? -> Need to read source to model?

[Benedikt Stefansson]

Darren Schreiber wrote:

>
> Some of my next ambitions include importing empirical data into my model,
> automating the collection of statistics and results, and toying with
> genetic algorithms.  R, HDF5, and LISP are all acronyms suggesting that
> I'll need to expend a lot of energy to get the results I seek, and the
> prospect of learning additional software/interfaces is just not exciting me.
>

I think Darren's last comment really says it all w.r.t. the discussion that has
gone on in this thread.

Some people (scientists) find it exciting to learn how to program and use
different types of software. The majority of  people (scientists) do not find
this exciting at all. They don't want to learn programming or have to deal with
the nitty gritty issues involved in developing and using simulation programs.

It has been said before in this thread that the optimal arrangement allows for
a domain specialist working together with a programmer in building models. But
as Miles pointed out domain specialists also need to understand the process
they are trying to model and how that process can be represented in a computer
model. The specification of the model and the implementation of a simulation
can not be split into two neat halves.

In effect the collaborators need to speak the same language and understand
which parts of the problem can fit into the modeling framework. Which really
means that the scientist has to get his hands dirty with all the nitty gritty
stuff he was trying to avoid learning. *)

I've recently been involved in training users to use a proprietary Agent Based
Modeling system, which unfortunately will not be released to the public. It was
interesting to see that the system, which could be described as an evolved
'MAML' like interface (see
http://www.syslab.ceu.hu/maml/maml.html) nevertheless seemed confusing to
trainees that had no prior programming experience. Using the system required
the trainees to understand what the system was doing in the background and what
its inherent constraints were - in other words, they could not use the tool
productively until they had become familiar with the wizard behind the curtain.

This is also my experience from trying to implement the 'domain specialist
+ programmer' paradigm: It doesn't work that well. Eventually, either the
domain specialist has to roll up his sleeves and dig into code or the
programmer has to become a de facto domain specialist.

It seems to me that this is the nature of the beast, although I'd sure hope to
be proven wrong sometime in the not to distant future when someone comes up
with a way to program computers without programming.

Benedikt

*) Just think how this works in the classical paradigm of mathematical
modeling; clever modelers master the mathematical foundation first and then
study a particular problem domain. The best theoreticians are the ones that
realize how to capture the essence of the problem given the inherent
limitations of their tool kit.

--
Benedikt Stefansson      | address@hidden
CASA, Inc.               | Ph : (505) 988-8807 x101
Santa Fe, NM 87501       | Fax: (505) 988-3440





                  ==================================
   Swarm-Modelling is for discussion of Simulation and Modelling techniques
   esp. using Swarm.  For list administration needs (esp. [un]subscribing),
   please send a message to <address@hidden> with "help" in the
   body of the message.
                  ==================================