REPLY The Death of System Dynamics? (SD6215)

[System Dynamics Mailing List]

Posted by  "Schuette, Wade" <wschuett at jhsph.edu>

 
Without excluding other excellent avenues, I second the general thought of 
John Gunkler that "expert modelers presenting fully developed models" are 
not the only way to influence policy.
 
In my own experience,  complex artifacts that follow long meetings are 
meaningful only to the participants and almost useless to convey information 
to those who weren't present and participating in the creation of the 
understanding reflected in the artifacts.   No one has time or ability to do 
a "cold start."
 
Others have presented eloquent arguments focusing on the creation process 
itself, on SD-guided discussions with a wide diversity of participants to 
tease out, simultaneously,  understanding, information, buy-in, and ownership 
of the resulting shared vision.   In fact, separating these tasks into isolated 
development of a rigorous model seems to be followed often with unsuccessful and 
awkward attempts to "sell" the results to stakeholders and policy-executives, 
making the entire experience ineffective.     The operating point is "optimized" 
to get the rigor in,  which, unfortunately, leaves the managers, executives, and 
the public out.
 
I hear comments from business leaders of this nature: "Never trust a complex 
model."
 
So, I'd suggest that higher total impact would be achieved by focusing more 
attention on this first step, the "power-white-board",   the initial 
meetings in which emotion, politics, data, and ideas are slowly teased out 
and focused into a problem representation and simultaneous social 
transformational residue of agreement with each other (and somewhat with the 
model.)   The fact that the representation may be "executable" is almost a bonus.    
As with any statistical analysis, less complex and less-complete models may be 
far more credible than "more accurate" models with multiple adjusments to the 
data that just seem like "cooking the data" to some.
 
In that sense, I fear the academic and expert community long ago overshot the 
"sweet spot" and the simplest model that seems "complete" to them is already far 
beyond the complexity that can be considered by the average person.  In fact, to 
the average politician, a simplistic model that comes with community buy-in and 
an acceptable political compromise would be far more acceptable than a "good" 
model that half the community could not understand and rejected out of hand. 
The perfect is enemy of the good.
 
In the case of SD, we have the "good" being enemy of "the acceptable."   That's 
the "sweet spot" that needs to be lubricated with tools and facilitators, it seems 
to me,  to get the camel's nose into the tent in the first place.   
 
Along that axis, the right question is - can we moved this room full of people 
to any model of complexity level X or less that can represent this situation with 
maximum social buy-in for this audience?   I recognize that this can be viewed as 
an alarming compromise, or ethical cop-out,  but I suggest it's a realistic 
recognition of one actual problem domain which is extremely non-academic in nature 
and has very different ground rules and value systems.
 
Along those lines, I share an observation made by Gary Olson, a professor of "technology-
mediated collaboration" here at the University of Michigan's School of Information.  
He noted that many, if not most, of the policy-making meetings he attended never once 
touched a white-board, and some lacked an agenda or subsequent minutes. 
 
The failure point, in other words, is way further back than refined system-identification 
techniques, back in the area where people, even academics, ever learned (or didn't learn) 
how to sit together in a room and draw on the entire room's expertise and generate 
collective wisdom.  So, to some extent, in this metaphor, we're trying to teach 
advanced physics to a room full of people who flunked algebra and aren't very confident 
of the whole idea of addition.  
 
For K-12 students, remediation can be addressed. For existing policy-makers and in-service 
education, I fear that what we really need is a "SD-LITE" stripped down tool, stripped 
down even more than Vensim, to get people used to thinking "this way" at all. 
 
For other audiences that do have the trust or the background, proceed as you were, of course.
 
Wade Schuette, MBA
Posted by  "Schuette, Wade" <wschuett at jhsph.edu>
posting date  Mon, 29 Jan 2007 10:42:00 -0500