Ventana software support forum

Posted: **Fri Apr 07, 1995 11:54 am**

-------------------------- [Original Message] -------------------------
Those interested in the history of system dynamics should also read Brian P.
Bloomfields: Modelling the world: the social constructions of systems
analysts (Basil Blackweel, 1986). In a recent email Jay Forrester wonders
why world dynamics had such a large impact (it attracted a lot of attention
from the media), given for instance the fact that the model was constructed
in two weekends and the book is technical for the lay person. Bloomfields
book suggests that this was largely due to the type of controversial
conclusions and proposed policies rather than the model per se!

Jac Vennix
J.Vennix@maw.kun.nl

Posted: **Fri Apr 14, 1995 6:11 pm**

I struggle with when model-building is the most appropriate approach to a
problem situation. I am wondering if any SD practitioners have a set of
criteria for when SD modeling is the best choice among all the possible
means of intervening in a problem situation.

In other words, when should a problem-solving effort use the SD approach
versus other valid approaches like Soft Systems Methodology ala Checkland
and Scholes, Human Performance Analysis ala Gilbert, Harless and Mager,
Statistical Process Control and root cause analysis, ala Deming and most
quality approaches, etc., etc., etc. almost ad infinitum. Given the
dizzying array of possible problem solving interventions available to
choose from, my questions are:
1. What are your CRITERIA for when SD models are the BEST
intervention for a specific problem situation?

2. How specifically do you determine whether these criteria are
being met in a given situation? i.e., what METHODS do you use
when initially responding to client requests that allow you to
test for your criteria, whatever they may be?

I look forward to any thoughts on this subject.

Joe Kilbride -- jk@mcs.com
------------------------------------------------------------>
Training & consulting in Quality/Process/Systems Improvement
PO Box 64 in Downers Grove, IL 60515
Ph:708-515-9882 FAX:708-515-9883
<-----------------------------------------------------------
The real voyage of discovery consists not in seeking new
lands, but in seeing with new eyes. -- Marcel Proust

Posted: **Mon Apr 17, 1995 4:23 pm**

Joe Kilbride wants to know about proper criteria and methods for determining
whether SD is the best tool for the job. As a full-time SD consultant, I am
often faced with clients who are already excited about SD based on some
previous exposure, but are not sure to which of their own issues it should be
applied. So together we look around for issues that have DYNAMIC COMPLEXITY.
Such issues involve strategic/policy decisions that have both intended effects
and unintended effects, and where the effects take months or years to develop.
They may be (1) problems that have defied prior attempts at solution, or (2)
controversies surrounding the long-term pros and cons of alternative courses
of action. We do NOT address issues in which all of the impacts occur quickly
(as in many operational settings better addressed by TQM, etc.), or in which
the impacts are already well-defined and easily computed.

We start by conceptualizing the issue, and press the scope of analysis outward
to include all those actors (customers, suppliers, partners, and competitors,
whether outside or inside the company) who may react to our decision in some
way that affects us significantly. This "soft" process opens ones eyes to
possibilities not previously discussed, and may occasionally be enough to
break a deadlock or resolve an issue. But all the arrows and loops typically
do not lead to a firm conclusion about the issue at hand, and leave the client
wondering, "Well, which loop really is stronger? And under what conditions?"
So then we move on to computer modeling.

We proceed with an SD model even when many of the desired data are missing or
sketchy, because we know we are dealing with a closed-loop system in which
insights and answers may be found even in the face of numerical uncertainty.
Also, people know more about their system than they think they do, certainly
more than their memos and documents would ever suggest, and it often requires
the computer model to bring this experiential knowledge to the fore. Unlike
other modeling techniques that lock one into a particular set of variables, SD
is especially good at helping people to get more precise and logically
detailed about medium-to-long-term problems that seem fuzzy at the start.

Jack Homer -- 70312.2217@compuserve.com

Posted: **Tue Apr 18, 1995 2:50 am**

I use S-D modeling when the purpose of the inquiry is to gain understanding
by studying system behavior over time, and there appears to be a complex
"web" of interconnecting feedback loops and/or significant delays in the
response of the system to stimuli...suggesting the presence of accumulators
or storages.

Wayne Wakeland,
Adjunct Assoc. Prof. of Systems Science
Portland State University

Posted: **Tue Apr 18, 1995 10:57 am**

Re Joe Kilbrides query regarding how to decide on which modelling
approach to select for a particular intervention. A substantial
literature exists on this issue. One particular approach to
selecting a complementary group of methodologies is
presented in Flood and Jacksons book (Creative Problem Solving:
Total System Intervention. John Wiley. 1991). I seem to recall they
have recently updated this approach in a paper in the Journal of the
Operational Research Society. Also a key reference is the Jackson
and Keys paper of 1984 (Towards a system of system methodologies.
Journal of the OR Society, Vol 35, p.473-486).

--------------------------------------------------------------------
T.F.Burgess Phone +44 (0)113 233 2614/5
School of Business and Economic Studies Fax +44 (0)113 233 2640
University of Leeds T.F.Burgess@leeds.ac.uk
Leeds LS2 9JT England
--------------------------------------------------------------------

Posted: **Wed Apr 19, 1995 9:58 am**

May I urge all to be extremely cautious about
using the Flood and Jackson work. There are two
reasons.

1) The system of system methodologies (SoSM)
that they employ as their main aid to choosing
systems methods has come under a barrage of
criticism within the UK/European OR literature.
Sample references:
* Tsoukas, H. (1993)Systems Practice 6(1):53-70
* Responses to the above and Tsoukas response
to the responses: Systems Practice 6(3)
* Checkland, P. (1992)JORS, 43:1023-1030
* Viewpoints sections of JORS, 1993, 44(6)

The criticism has been at the deepest
philosophical level (is the use of Habermas
Critical Theory ideas in columns of the SoSM
correct and valid?), at the theoretical level
(how do I choose the appropriate Morgan
metaphor to steer me through the SoSM), the
practical level (is it really possible to
understand the range of different methodologies
so well that an informed choice between them is
truly possible?)and the level of tone (some
observors object to the arrogance of tone of
the book and the way that it parades around in
a rhetoric of emanicipation)

Flood does attempt to redress some of these
issues in the paper mentioned by T. Burgess
{its JORS 1995, 46(2):174-191, by the way} but
there remain, in the views of many within the
OR and Systems communities here, quite profound
difficulties.

2) The treatment of System Dynamics in the F&J
book is laughably ill-informed. From the
description given there, SD a way of using
engineering or physics-type system analysis to
study the visible, tangible elements making up
the logistical parts of an organisation. Any
one recognise this? Yes, its the old "but SD
is just control theory isnt it?" argument.
Nothing here about models as transitional
objects, nothing about experiential learning,
representation of values and goals within
models, or the aim being to obtain qualitative
insights . . . . I could continue. Some of the
same comments apply to the description of
Beers work and although space constraint will
have been a factor, in the case of SD I - and a
few other system dynamicists who have come
across F&J, feel that the authors simply do not
understand SD. For some more discussioon on
this issue you might be interested in, in
addition to the above more general references,
the following, which engage with F&J from the
point of view specifically of SD:

* Lane & Jackson. 1995. Only Connect! Systems
Research. To appear.
* Lane. 1994. System Dynamics PracticeJORS
45(3):361-363
* Lane. 1994. With A Little Help From Our
Friends SDRev 10(2-3)
* Lane. 1994. Social Theory and SD Practice.
Stirling Conference Proceedings or updated
working paper available from the author.

In particular, the SD Review paper puts the
F&J book in the historical context of the flow
of ideas in OR.

***********************************************

I would wish to add that I believe that the
issue of choice of method is one of the key
research areas in OR/System Science in the
1990s - and probably beyond. I do not believe
that SD can solve all problems; but then it
does not claim to. So this is a good discussion
to have. However, for any participants wishing
to look at the Flood and Jackson work I would
wish them to be aware of the raging argument
going on in the UK about the validity of their
approach and also of the inadequate treatment
given there to SD.

All comments, alternative views, experiences
and rotten fruit projectiles gratefully
received.

Dr. David C. Lane
Operational Research
London School of Economics

Ventana software support forum

When to Model

When to Model

When to Model

When to Model

When to Model

When to Model

When to Model