REPLY Models, Problems and Systems (SD7020)

[SDMAIL Jack Homer]

Posted by  "Jack Homer" <jhomer at comcast.net>

Re: modeling problems rather than systems.
I have been a little dismayed to see extended discussion of yet another 
question that was answered in our field decades ago; see, for example, 
Industrial Dynamics (JWF, 1961), section 5.1, page 60:

"In practice there will be no such thing as THE model of a social 
system, any more than there is THE model of an aircraft...In designing a 
dynamic simulation model of a company or economy, the factors that must 
be included arise directly from the questions that are to be answered.  
In the absence of an all-inclusive model, which we are unlikely to 
achieve, there may well be different models for different classes of 
questions about a particular system.  And a particular model will be 
altered and extended as each new question is explored."

Some of the discussion here has considered "system" to be broader than 
"problem", and has therefore wondered whether the "model a problem not a 
system" dictum may be too limiting.  In practice, however, much of the 
value in the dictum has been to give us permission to model variables 
BEYOND those normally considered in a narrow conception of what the 
system is.  The "system" is often considered in a physical sense 
(perhaps coming from a traditional operations research perspective) to 
comprise only the production division and not the sales division, or 
only the company and not its suppliers and customers, or only the 
hospital and not the community that surrounds it.  We are properly urged 
to model problems (or, as they are sometimes known, issues or questions) 
rather than systems, not only because systems are potentially without 
limit in breadth and detail (where do we draw the boundary? when do we 
stop disaggregating?), but also because the common view of systems is 
actually TOO constraining with respect to boundaries, leading to the 
kind of silo thinking and suboptimization that has been forever the bane 
of operations research.

We should always emphasize with novices to our field the importance of 
modeling problems/issues/questions rather than systems, and should not 
blur the lines of this important principle.

- Jack Homer
Posted by  "Jack Homer" <jhomer at comcast.net>
posting date  Fri, 9 May 2008 09:15:50 -0400