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

[Marcus G. Daniels]

Rob Bowers wrote:
> What the new user needs to know, from the beginning, is whether a 
> toolkit will be able to do what they need it to do. Let it be 
> difficult to learn (says I), or computationally inefficient. But after 
> putting so much into learning it, let it have been the right toolkit 
> to have learned.
Any Turing compute machine on a modern computer can get the job done.  A 
popular language here is Java or C.
Both are powerful and you can do anything with them.   As has been 
mentioned, you won't get conceptual support for modeling from these 
languages, although you can it from libraries implemented for them.  
What "modeling support" _means_ is a little unclear: You'll probably get 
100 answers if you ask 100 people, or perhaps fewer, more vague 
answers.   Vague answers like "because there is a community of users", 
or "because it does GUI stuff for you".  

You've cast the question in such a way that "learn to use a popular 
programming language" is pretty much the obvious answer.  But I don't 
think that is the right answer, necessarily.  Better for many, I 
suspect, to throw some simulations ad models away.  E.g. start with 
something like NetLogo and use it until it stops being useful.   It's 
not so much about the technology as it is about the intent of the 
toolkit projects.  Are they optimizing for ease of use, for flexibility, 
for concise and correct models, for  performance, etc.?  One might claim 
that all of these are optimized by some toolkit, but I'm very skeptical 
about that!  The way I see to optimize for them all is to build on a 
better programming languages, in order to benefit from economies of 
scale outside this small community...