Re: Paper on Control Toolbox?

[heberf]

David Doolan wrote:

> The conference paper would be of service to those of us attempting 
> to convince academic advisors that software development is indeed
> a worthy academic aspiration, i.e., one can get pubs from it.
> Some are lucky to be in fields where this is taken for granted.
> Others, such as myself, are limited to crufty homegrown fortran
> solutions, or very expensive, feature poor windows implementations
> of those crufty fortran codes.  (I can provide  very specific 
> examples of this upon demand.)  As it stands at the current time,
> such a paper could not be published in my field.  Even worse,
> I get no credit towards a dissertation from the engineering that is
> going into the implementation of a numerical modeling code.  

I know just the type of product you are referring to.
 
[snip snip]
> My personal interest is not so much in the actual publication as in 
> the legitimization of what I am currently doing on the somewhat on the
> sly, which would in turn improve the quality of my code.

This is important for all of us.  In my field the replication of results is 
becoming more important.  It used to be that people would never share their 
data 
or their code.  But now more and more researchers are posting their data and 
their code.  The problem is that if you post S+, SAS, SPSS, or Gauss code not 
everyone can run it.  I know a guy who got special permission to put NPSOL on 
his ftp site just so others could use his routines (which relied on NPSOL).  
This kind of exception is just that, an exception.  Here is where free software 
can make a big contribution.  When you or I publish papers we can post the 
m-files, oct-files, and data and a pointer to the Octave site where people can 
download the whole thing and replicate our results.  There are two benefits to 
this. (1) We ligitimize our research by making it replicable.  This increases 
the chances that people will implement our methods, extend our code, and cite 
our work. (2) We create more octave users who can help contribute to the code 
base.

I've found the best way to do this is on the cover page of a working paper you 
put the ftp site where the data can be found.  

Heber Farnsworth

 
> Thanks for listening,
> 
> Dave D
> 
> reply to address@hidden
>