Re: slightly off-topic: support open source for publically-funded research

[glen e. p. ropella]

I agree completely.  However, you've picked the least important
point of the argument as your rhetorical fulcrum.  Publishing
the source code for quirky or novel logic *is* the single best
way to communicate that logic.  However, that is important *only*
when that logic is the focus of the communication.

If the point of the research is to *use* that logic to achieve
some other result, then publishing the source code is simply
a red herring and it encourages Bad Behavior.  In that case,
it is better to wrap the functionality in an abstracted
specification and de-emphasize the source code (though don't
hide it).

Re: the more complete data for modeling the author... Ideally,
... in a world where many readers admit their responsibility...
you're absolutely right again.  However, this isn't an ideal
world.  In the world we live in, both authors and readers are
prone to mis-stating or mis-interpreting any given point.
Authors can get lost in the details of the source code and
readers can get distracted by the details of the code.  Hence,
the source code should only be emphasized *if* (and only if)
it is of primary interest to the research.  If it is not of
primary interest to the research (and it's not in most publicly
funded research), then the author and reader should spend
significant efforts abstracting away from the source code,
itself, and try to understand what that code is USED for....
which, in turn, means non-source specifications and solid
experimental procedure.

At 05:03 PM 11/16/2001 -0700, you wrote:
> My claim is that novel, complex, abstractions can be hard to grasp
> independent of their form (in English, ODEs, or computer code, etc).
> These kinds of abstractions are not unusual when doing agent-based
> modeling and so open source is useful because the common case of
> research use involves the exercise and modification of computer code.
>
> In other words, I think alternative representations have marginal
> value since they will ALSO be hard to grasp.  Worse, since research is
> shifting sand by definition, there isn't a lot of available energy to
> do a good job of iterating the translation, and thus it will probably
> be done in an inadequate way when averaged over many instances of
> different research efforts.  So, assuming Murphy's law, open source is
> important because it is a useful as a contingency.
>
> In less complex cases (which should be sought), I believe it is
> preferable to avoid code as a communication scheme.  Frankly, I'm
> inclined to bound the application of ABM to first-cut evaluation of
> phenomena for just this reason...
>
> As far as software re-use goes, my experience is that having source
> code is invaluable for understanding what software does and even is
> intended to do (given the context).  It isn't necessary to become
> intimately familiar with a whole system, just dig down enough into the
> particular pieces that are of interest at the time.  I find this kind
> of thing, combined with available documentation, helps to me to model
> what the author was thinking at the time of its writing.

=><==><==><==><==><==><==><==><==><==><==><==><==><==><==><==><=
The competent hawk hides its claws.       =><=        Hail Eris!
glen e. p. ropella   =><=   H:831.335.4950  =><=  C:650.776.4616



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