REPLY Models, Problems and Systems (SD7007)

[SDMAIL George.McConnell]

Posted by  George.McConnell at selex-comms.com

If I may put in my 2p-worth on the subject of "model the problem not the 
system"...

I have, I guess two relevant points to make on this.  The first is that 
whatever is modelled is "a system" because we have defined it as such 
therefore the dictum is rendered rather meaningless in my view.

The second is that I think that the 'spirit' of the dictum (as against 
the actual words) is something that it is necessary to take on board 
when carrying out any modelling (not just SD).  Too often I am asked to 
"build a model" - the person asking me to do this has rarely considered 
what this is to be a model of - they expect a model of "the system" 
without really considering what particular "system" is of interest in 
their problem.  My response of "what do you want a model of?" usually 
results in a significant period of "umming and ahhing" indicative of the 
fact that the reality is that they haven't a clue.

I've seen many models of "the system" consume much resource and produce 
little worthwhile output - they are usually large models that are close 
to as difficult to understand as "the system" itself.  As Kim says, in 
many cases we actually want a model that will allow us to manage the 
system - I would argue that the model is still not a model of "the 
system", but of "the system that manages the system" (I hope that 
doesn't come across as mere play on words) and this latter is actually 
"the problem" that we are managing.

Sorry if that is not clear, it is one of those subjects where the 
language gets in the way of itself!

regards
          *George*
Posted by  George.McConnell at selex-comms.com
posting date  Thu, 8 May 2008 14:05:08 +0100