Re: Coding your own

[Miles Parker]

>At 08:22 AM 5/8/00 -0700, Steve:
>>So: I'm trying to decide how much to say about this issue in a paper I'm
>>preparing to submit. Feedback?
>
><GR> Heh, I can't help decide how much of this you should put in a paper;
>but, I can say that I agree wholeheartedly.  However, the important
>caveat is that the modeler can't be ignorant of the code anymore
>than the coder can be ignorant of the model.  The *primary* aspect
>that needs to be present is to have experts in at least two domains,
>programming and modeling (in their respective macro-domains, obviously,
>like "programming simulations" and "modeling fish populations").  Then
>each of these experts has to have more than a passing interest in
>the other domain and a willingness to play what-if games with their
>partners.

Also totally agree, and want to amplify a bit what Glen said. I doubt things 
would work that well if one just took a kind of 'service bureau' approach to 
obtaining programming resources. It seems to me you would need to find someone 
who is at least willing to think about the model itself, and who hopefully has 
some real interest in it. [On the other side, the modeler has to have at least 
a passing interest in the coding process and issues themselves, as we've 
discussed in the past.] 

I think it would be a mistake [and forgive me, I'm prob. just stating the 
obvious here] to think of interaction with programmer/analysts or SE's as 
simply a one way 'go off and do this' kind of thing. Good programmers are 
trained and inclined to think through precisely the kind of issues that come up 
in model design carefully and creatively, and really good collaboration can 
result from it. I, for one, regard this as a good opportunity to break down the 
boundaries between functions and disciplines, so that's my bias. [And OT, from 
a somewhat self-interested point of view I reject the blue-collarization of the 
programming/IT field that seems common in many areas, particularly corporate 
ITS.] That said, there are programmers who just want to get a complete spec and 
go off and implement it, wrong or right, but I think thats a real (and 
unwelcome) rarity, unless you're implementing low-level hardware drivers or 
something.

At the same time I wholly agree with the point that people whould primarilly do 
what they do best. I also think its beneficial for people involved in modelling 
to have at least some expereince with programming to get a loose feel for the 
issues involved, even if that programming experience simply consists of 
experimenting with StarLogo or writing spreadsheet simluations.

-Miles




Miles T. Parker
Software Engineer
The Brookings Institution  1775 Mass. Ave. NW  Washington, DC  20036
http://www.brook.edu/es/dynamics/models/ascape
mailto:address@hidden  voice 202.797.6136  fax 202.797.6319



                  ==================================
   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.
                  ==================================