Hi Ken,
I'm puzzled by your description of system dynamics. It bears very little
relation to my experience in using system dynamics modelling and being part
of the international systems community. Perhaps there is something I'm
missing, but your explanation doesn't ring true to my experience and
understanding. For example:
One of the key reasons for using systems dynamics modelling is that its
predictive power *doesn't* depend on having significant amounts of data.
Instead it requires a) the sufficient and often simple models of causal
relation between factors; and b) data about single real world 'snapshots'
with which to test the correlation of the system dynamics model with the
real world.
Second, the measure of predictability typically used in the system dynamics
field is *usefulness* - the dynamics models are created because observing
their outputs is more useful than guessing the future in situations that are
too complicated for humans to manage 'in mind' or 'in discussion'.
Third, one of the useful attributes of system dynamics modelling is that it
*doesn't* require clear boundaries, nor does it require exactly modelling
the behaviour or relationships of each and every factor involved. In that
sense, you could see it as very like Physics in which the properties of
situations are idealised to only contain the dominant influences.
Fourth, a great advantage of systems dynamics modelling is that it is
relatively cheap and fast to undertake, particularly compared with trend
data analysis or attempts to classically mathematically model situations:
the conditions where you really do need to collect each and every bit of
data about the situation.
Fifth, in my experience there are only two flavours of system dynamics:
'system dynamics' and 'non-linear systems modelling' (the heavy duty
mathematical approach from which systems dynamics emerged that originated in
non-linear control theory). There is however, a wide variety of software for
system dynamics modelling.
Sixth, using system dynamics for design requires very little resources. For
most projects it certainly *doesn't* require large expert teams. For
example, the last time I had contact with the CSIRO project to redesign the
planning policies for tourism and economic development of the Ningaloo Reef
(which uses system dynamic modelling as one of the main predictive tools) it
had the system dynamic modelling undertaken at Curtin University being
managed by a research fellow and a PhD student.
Seven, the need for iterations is primarily only found in agent-based
modelling, which might be considered a sort of bottom-up way software for
system dynamics modelling
Eighth, if you feel system dynamics is so poor at predictability, I'd be
interested to know your explanation as to why IBM has been and is investing
so heavily in system dynamics modelling for prediction of design outcomes
for planners, policy makers and national governments - see, for example:
http://www-03.ibm.com/press/us/en/pressrelease/35204.wss
http://engagingcities.com/article/system-dynamics-planning-smarter-cities
http://eyeonibm.com/2011/08/09/building-smart-cities-through-simulation/
http://www.ibm.com/smarterplanet/us/en/smarter_cities/overview/
IBM have been backing the use of system dynamics for making predictions for
40 years or so!
(http://news.google.com/newspapers?nid=1982&dat=19740107&id=5SVHAAAAIBAJ&sji
d=czMNAAAAIBAJ&pg=2748,883053 )
Systems dynamics modelling is already in widespread use for prediction to
support decision making across several areas of interest to design
professionals. This has been so since the 70s. A quick google search
indicates bodies of literature describing projects using systems dynamics
in a predictive capacity in: Health service policy design, rework management
in the construction industries, management of national education systems,
investment strategies (national level downwards); power systems managment
(electricity, gas and oil); public health promotion systems; shipping
design and investment; entrepreneurial support systems; supply chain
behaviour; industry investment incentives; effects of environmental
policymaking; behaviour of industrial/product design industries; design of
short life cycle products; sustainability design; collaborative design
processes; teaching and learning; R&D systems; foresight; and strategic
planning, to name a small subset.
I remain puzzled at your explanation. From my experience, system dynamics
is a useful tool for addressing complicated and complicated design
situations.
I look forward to your clarification.
Best wishes ,
Terry
---
Dr Terence Love
Director,
Love Services Pty Ltd
PO Box 226, Quinns Rocks
Western Australia 6030
Tel: +61 (0)4 3497 5848
Fax:+61 (0)8 9305 7629
[log in to unmask]
--
-----Original Message-----
From: PhD-Design - This list is for discussion of PhD studies and related
research in Design [mailto:[log in to unmask]] On Behalf Of Ken
Friedman
Sent: Tuesday, 20 August 2013 9:13 PM
To: [log in to unmask]
Subject: Re: Epistemologically Valid Theory and Dynamic Modeling
Hi, Terry,
This is just a quick note to say that I will respond to your note on systems
dynamics in a day or so. I've been down with a cold, and between trying to
stay on top of my obligations while reviewing some literature, it is taking
me time to write a cogent reply.
I do, indeed, know about systems dynamics. When I respond, however, I will
explain why systems dynamics is not appropriate for "understanding or
predicting the dynamics of the outcomes of" the specific issue of "national
change to design education in India." One cannot use systems dynamics in
every instance and systems dynamics is not appropriate for any intervention.
There are several flavors of systems dynamics. While some versions of
systems dynamic are suited to heuristic modeling for understanding, they may
not be suited to "predicting the ways things will change over time in a
highly interacting political environment."
Predictive modeling is extremely difficult. As I will explain in my next
post, systems dynamics has achieved nothing like the level of predictive
capacity of physics, chemistry, or molecular biology. The predictive power
of physics differs to that of systems dynamics by several orders of
magnitude. There are good reasons for this, and I will explain them. The
mathematical operations are not at issue. The issue involves connections
between data and symbols, between symbols and what they represent in the
world.
The joke about economists - "economists use this kind of modeling on limited
aspects of economic systems, but there are many jokes about throwing out the
data that don't fit the theory" - wasn't a joke about systems theory or
systems dynamics. It was a joke about the notion that we have workable
systems for dynamic modeling of complex adaptive structures in the real
world of organizations embedded in the context of national political life
that enable valid predictions.
Systems dynamics does permit valid predictions, in cases with clear
boundaries and limits. I will describe these in my post.
I have myself done some work in systems dynamics. This was forty years ago
when I studied organizations and agents of change for my PhD. Most people in
my cohort of PhD students were headed toward careers in psychology,
anthropology, or urban planning. My project team had the outliers - I was
one. The others included the deputy mayor of one of America's largest cities
and the captain of an aircraft carrier earning a PhD in human behavior on
the way to an admiral's flag.
One clear aspect about the different flavors of systems dynamics is that
predictive systems dynamics requires significant amounts of data. Accurate
and responsible predictions usually require a large, expert project team
with sufficient time to gather data and model it through many iterations
well before the results are due. Relatively short-term projects do not
permit this unless the financial stakes are so great that they justify
investing in large teams, massive data, and heavy computing power. Frequent
changes to governments, stakeholders, policies, and decision-makers affect
most contexts such as those that Ranjan describes. These can benefit from
visualisation for understanding and heuristic flavors of system dynamics,
but they generally don't permit or justify the costs or work required for
predictive modeling.
The reason I did not think of systems dynamics in the context of your post
is simple. The capacity to engage and deploy systems dynamics for predictive
modeling rests on far more than educational background. The linked
constraints of time, money, and quality - the iron triangle - come into
play.
You specified prediction as well as understanding. To me, that ruled out
systems dynamics.
I asked, "What kinds of symbols permit us to undertake dynamic predictive
modeling of complex adaptive structures in the real world of organizations
embedded in the context of national political life? . If you have not done
this kind of work, can you suggest some responsible publications that
demonstrate such a system or show it in operation?" I was hoping for working
examples. None of the papers in the links you posted meet these criteria.
Neither do your papers. These papers do not involve predictive dynamic
symbolic modeling of the kind you described to Ranjan or to me. Rather, they
are papers stating your views about systems dynamics. They bear roughly the
relation to quantitative, predictive systems dynamics that papers on
philosophy of science bear to physics, chemistry, or molecular biology.
Systems dynamics has many values and uses. Some are heuristic. Some involve
understanding. The most difficult form of systems dynamics involves
predictive modeling. Most of the people who do this kind of work publish
their findings in peer-reviewed journals. I've observed that the articles in
systems dynamics journals are modest and limited in their claims, and the
systems dynamics literature is very carefully bounded.
There are also systems dynamics people who work in consulting, government,
and other fields - I'd guess that what they have in common is an
appreciation of the different uses of systems dynamics and the different
kinds of systems dynamics that are available for those uses.
There is also a certain measure of humility that seems to typify anyone who
really works with these systems in a professional way. By the time you take
on responsibility for a capital ship of more than 100,000 tons plus 80 or 90
fighter jets and advanced missile systems to protect them, you grow cautious
about the power of prediction.
We had a standing joke in our team, attributed to the physicist Niels Bohr
(and to others):
"Prediction is very difficult. Especially about the future."
Yours,
Ken
Ken Friedman, PhD, DSc (hc), FDRS | University Distinguished Professor |
Swinburne University of Technology | Melbourne, Australia |
[log in to unmask]<mailto:[log in to unmask]> | Mobile +61 404 830
462 | Home Page
http://www.swinburne.edu.au/design/people/Professor-Ken-Friedman-ID22.html<h
ttp://www.swinburne.edu.au/design> Academia Page
http://swinburne.academia.edu/KenFriedman About Me Page
http://about.me/ken_friedman
Guest Professor | College of Design and Innovation | Tongji University |
Shanghai, China
-----------------------------------------------------------------
PhD-Design mailing list <[log in to unmask]>
Discussion of PhD studies and related research in Design
Subscribe or Unsubscribe at https://www.jiscmail.ac.uk/phd-design
-----------------------------------------------------------------
|
