REPLY SD Impact on National Government Policies (SD6186)

[System Dynamics Mailing List]

Posted by  j-d <jaideep at optimlator.com>

Greetings Everyone:

I have been away from the list for a while but have read this topic with
much interest. About 5-6 years ago we used to have similar debates and I
have seen the same kinds of debates in Operations Research and Urban
Planning communities. It is also interesting to see how the writings and
thinking of familiar writers on this list has matured.

First a small background - I had academic system dynamics training (Ph. D.
level) and my thesis involved work with large and complex nonlinear
optimization models. I had a one-year system dynamics consultancy stint at
Sandia Labs, and worked for one year PostDoc fellowship building and trying
to sell system dynamics models. Realizing that the market trends, current
academic paradigm and hype was simply not working for me, I decided to
slowly change to more exciting ventures (for me, that is) in terms of
designing web/Windows applications backed by SQL server databases, and I
have been very successful at that, and can see gratifying results when
people use and compliment about my applications. I went over to the "dark"
side of linear, flat thinking, so to speak. Hopefully I have a little bit of
insider-outsider experience now which informs my perspective - please allow
me to share some of it here, especially as I have studied SD, tried it as a
consultant, and have given up (mostly) on it.

1. Some people say that SD works for large organizations, but it works so
well that it is kept a highly guarded secret and is hidden. I simply do not
buy this argument for a minute - with the explosion of information and
movement of people within industries, anything that works will not be a big
secret for long. We all basically know what makes Google, Yahoo, eBay and
Amazon such great successes - these are essentially technology companies -
the problem is that people simply don't have the strategy, the resources and
the fortitude to carry out a vision, even if they know what can work. If
system dynamics worked as some say it does, then the SD vendors/consultants
will themselves push its sales very aggressively, just like Microsoft,
Oracle and IBM do for its products, and we will see open-source,
service-based competitive offerings everywhere for SD software and tools.

2. There are a wealth of freeware, shareware, beerware tools for almost any
other aspect of technology - not so many for system dynamics - we will see
many such tools if people found them useful. Simple market dynamics. If
anyone knows of good inexpensive SD tools, I am interested in knowing which
ones.

3. Some people say good SD is hard, but then programming and designing
databases is hard - there are amazing tools to make these tasks easier (I am
familiar with vb.net and sql server so that is where I am coming from). I
haven't seen the most recent versions of SD software, but what I had seen
was either too simple (Stella/iThink), or with many more exciting features
but not so easy to use (Vensim, e. g.).

4. The returns from process improvements and usage of even simple
applications using programming and database tools is amazing (in my view
actually small apps and changes are better than big apps and projects, as
they allow for agility and dumping of bad ideas very quickly). SD simply
does not have quick returns. Great insights based on simulations but then
what. Earlier on the list, I used to be a big advocate of good clean data,
because I was realizing that non-SD users simply didn't trust my story after
a while - it almost always seemed like a good but toyish exercise to the
customers. I had the feeling that data had a lower value to simulation-based
insights for people on this list and even for some SD software vendors. I
remember once showing my work in an SD job interview where I was asked why a
particular relationship wasn't included even though intuitively it made
sense to have it - I said I didn't include it because previous regression
studies showed no need to include it - the person insisted that it should
have been there - I didn't get the job and I didn't care much for it either
after these kinds of questions. Point of story: even SDers fail to see their
own biases and throw out econometric results just because it is "evil,
linear thinking".

An example from aikido as a martial art that I have practiced for 15 years
now: you can do all the simulations and kata in the dojo (hence gaining
great insights, just like in current SD way of doing things), but a dose of
street reality (real world business) can very quickly wake you up (or knock
you out, which seems to be happening to SD). For people who follow martial
arts, think of what happened when Gracies came to the UFC scene, demolishing
the myths of many simulation/insight-based arts. Extending Richard
Stevenson's analogy of cats, think of independent cats doing aikido and then
rushing out to streets on their own to greet thuggish wild cats - not a
pretty outcome.

The biggest benefit from SD and systems thinking to me so far has been to be
careful of unintended effects in designing large/complex databases/apps, and
to think in terms of chunks/interfaces and so on (of course, these insights
are also  available in OO programming). Also, computers have become much
more powerful (in ten years, when I finished my Ph. D. work on SLOW
macintoshes), so I believe a lot of useful work can be done even on home
computers (I am talking about multi-player dynamic game optimization stuff).
But without good data, linking with other fields, easy-to-use software, and
quick demonstrable returns, SD has little hope.

Best regards - still in love with SD but separated and living with someone
else

Jaideep

Jaideep Mukherjee, Ph. D.
Application Services Manager
Star of Hope Mission
Posted by  j-d <jaideep at optimlator.com>
posting date  Fri, 12 Jan 2007 12:30:11 -0600