The return of the System Dynamic Zeitgeist?

Extraordinary how for years you hear nothing about a meme, and then it come in threes in a matter of hours....on Saturday I went to an old colleague's birthday bash, and it resolved itself into one of the larger collections of business System Dynamics / game theorists outside of a conference on the stuff. Last night I read this note, arguing that game theory predicts the bailout won't work, and today on Twitter I see Kathy Sierra also talking about System Dynamics. (Heres is a simple System Dynamic model dealing with "Bricks to Clicks industry outcomes - you can run it yourself on iThink over here)



Interesting its coming back, the last time all this stuff hit anything like mainstream public consciousness was with the book The 5th Discipline, in the early 1990's. (The last time there was serious property financial trauma, interestingly enough)

However, having written many types of simulation models, business games etc and consulted around this space for the best part of 30 years, I can testify to a number of issues the approach has as a methodology - best encapsulated by one of my early lecturers in Operations Research:

People would rather live with a problem they can't solve than a solution they don't understand

The problem with simulation modelling for client work seems to be twofold - firstly, persuasion:

- the high level models give the best learnings and insights, but are attacked by critics as not being comprehensive enough.

- more detailed models are extremely laborious to construct, and very hard to test without an ongoing live environment to run off - and are then criticised by nay sayers for being too complex etc.

Secondly, timing - even the best models find it very difficult to predict "when" - our models of the UK housing bust showed it would be 2005/6, not 2007/8, for example (there is a whole lesson there about what I now term the Momentum of Vested Interests - ie its amazing how you can see broad collusion to push something well past its "natural" zenith).

The other downside is it is costly to build anytin but very high level models. When it works, it can lead to amazing insights, but it needs clients who are open to the approach - and is thus sensitive to people who would oppose the predicted outcomes, which is why it is not more commonly used.

However, I do note a number of interesting trends - the internet is effectively a closed loop system system, and that allows it to be used as one huge System Dynamic playground. The growth in data collection by Web businesses of all stripes, the increasing use of heavy duty analytics in understanding that datastream is thus no accident. Also, note the small but growing numbers of predictive system plays that are popping up.

In addition, emerging applications such as Social networks, and social media overall, are probably very fertile fields to apply SD thinking and modelling

And the large numbers of quamt jocks forcibly leaving the financial sector gives a one-off infusion of the people with the sort of skills to do this work.

Update - I saw this article in the WSJ, about using video and games to make business more interesting, and it did occur to me that there is probably a whole new approach to simulation gaming possible. Years ago I worked on DVD based business video games, but modern technology is a lot better and a lot cheaper to produce content with.