Hi,
I am relatively new to system dynamics. Just a little bit of background. I am based out of India and primarily worked for around 10 years in media planning and market research.
Around 3 years, back, the motherhood question on advertising started troubling me - does it work? how to measure ROI?..etc
And I started doing market mix modeling in my company - where we were essentially using the methods of econometrics to explain the effects of various elements of the marketing mix on sales.
I am nowhere close to being an expert on econometrics but I must confess that I did not feel in sync with some of the answers it threw up in our modeling exercises. Additionally, the need for high quality time series data was always a limitation - data collection methods in India are still largely manual, and hence both the number and quality of data points are limited. And, atleast amongst the econometric modelers I worked with, there was a tendency to 'terrorize' the client with terms like canonical correlation, heteroskedasticity etc!!
More fundamentally, I guess, I was not fully comfortable with the world view of this kind of modeling - from whatever extent I knew about it.
By sheer luck, through a friend, I found out about System Dynamics, and am really liking it! Have been reading Sterman's book for the last 9 months. And also enrolled for the online graduate certificate program at WPI. ...and hopefully work towards the online masters program too.
I would really appreciate if someone could give me tips on how to explain SD to a non-SD audience. In the last few months, I have found myself generally discussing SD with friends, colleagues etc. and getting lost in the details - in one discussion with my boss on the benefits of SD, I really lost his attention while explaining the criticality of 'DT'.
Essentially, what would be the best way to communicate the world view of SD to a non-SD audience? I feel that SD as a philosophy is possibly a little easier to explain to an audience than say, econometrics - where, the need to know the terminology and the finer details of the skill are very high. But, how does one do it efficiently?
Secondly, any recommendations on additional books beside John Sterman's would also help me in my gaining knowledge of this fascinating field.
Many thanks in advance. And apologies for this rather long note.
Malli
Explaining System Dynamics to others
explaining system dynamics to others
Hi Malli
The best thing to my opinion is to ignore theory and show concrete examples, starting with
simplistic examples and gradually increasing the material of the model, verifying at each step that the person has fully understood what you try to teach.
To verify that somebody has understood, he must be able to explain the thing back to you, using different parameters.
Example:
Start with a simple influence diagram, with no stocks, purely static, that illustrate the notion of causality.
Number of customers, widgets bought every month by customer, price of a widget, average margin by widget, the result being the total margin by month.
At this stage one must introduce the notion of unit that is fundamental in SD.
Widgets / customers, Dollars / widgets etc…
Step two: adding dynamic.
You introduce a stock, without feed backs:
for instance, the accumulation of the margin during a certain number of time steps.
One can at this stage, vary the time step, one day, one week or one month.
After one has well understood the notion of accumulation into a stock, the level three would be to show that the stock can diminish too. On can add that every time step, there are some fixed costs diminishing the margin, adding the notion of cash.
There are still no feed backs.
Level four: the first feed back:
The cash being placed in a bank, earns some interests that increase the cash in return, generating a first feed back.
Level five: adding a second stock with a second feed back.
On can consider that as the cash is increasing, a part of it can be spent in publicity that may increase the number of customers, that was fixed until now. To represent this one must change the representation of the customers into a level, that will increase with the amount of cash, and the increase of customers will increase the amount of cash, that will in return increase the amount of customers.
You can explain that both feed backs are positive and add word of mouth influence as a third positive feed back.
You can then introduce a balancing feed back that is more subtle to understand, for instance by introducing a total market and considering that the efficiency of the publicity may decrease when the number of potential customers is decreasing too.
This cannot be done in one day, but after three or four days, and one hour every day, that would do.
Regards.
JJ
The best thing to my opinion is to ignore theory and show concrete examples, starting with
simplistic examples and gradually increasing the material of the model, verifying at each step that the person has fully understood what you try to teach.
To verify that somebody has understood, he must be able to explain the thing back to you, using different parameters.
Example:
Start with a simple influence diagram, with no stocks, purely static, that illustrate the notion of causality.
Number of customers, widgets bought every month by customer, price of a widget, average margin by widget, the result being the total margin by month.
At this stage one must introduce the notion of unit that is fundamental in SD.
Widgets / customers, Dollars / widgets etc…
Step two: adding dynamic.
You introduce a stock, without feed backs:
for instance, the accumulation of the margin during a certain number of time steps.
One can at this stage, vary the time step, one day, one week or one month.
After one has well understood the notion of accumulation into a stock, the level three would be to show that the stock can diminish too. On can add that every time step, there are some fixed costs diminishing the margin, adding the notion of cash.
There are still no feed backs.
Level four: the first feed back:
The cash being placed in a bank, earns some interests that increase the cash in return, generating a first feed back.
Level five: adding a second stock with a second feed back.
On can consider that as the cash is increasing, a part of it can be spent in publicity that may increase the number of customers, that was fixed until now. To represent this one must change the representation of the customers into a level, that will increase with the amount of cash, and the increase of customers will increase the amount of cash, that will in return increase the amount of customers.
You can explain that both feed backs are positive and add word of mouth influence as a third positive feed back.
You can then introduce a balancing feed back that is more subtle to understand, for instance by introducing a total market and considering that the efficiency of the publicity may decrease when the number of potential customers is decreasing too.
This cannot be done in one day, but after three or four days, and one hour every day, that would do.
Regards.
JJ
I would purchase some slides from the "Learning Fables" series from Pegasus Communications.
http://www.pegasuscom.com/fables.html
http://www.pegasuscom.com/fables.html
Explaining SD to non-systemdynamicists
Hello Malli,
your initial question is a real challenging task that I am coping with as well (as quite a number of systemdynamicists actually do as I learned yesterday at the workshop of the German Chapter of the System Dynamics Society in Mannheim).
You have to find stories that the others can emotionally connect to their own experience without even naming it SD, ST or alike. Just do the plain happenings ask people what they think the explanation -from their point of view- could be.
From there on -as Jean-Jacques already has outlined- you could go deeper, mentioning an article that is well covering the question in place, do the static analysis, and so on.
And be prepared to wait for a long time until people get the understanding as they are mostly not thinking in this way for decades and therefor it is like learning a new language (this takes time of practice).
Best regards
Ralf
your initial question is a real challenging task that I am coping with as well (as quite a number of systemdynamicists actually do as I learned yesterday at the workshop of the German Chapter of the System Dynamics Society in Mannheim).
You have to find stories that the others can emotionally connect to their own experience without even naming it SD, ST or alike. Just do the plain happenings ask people what they think the explanation -from their point of view- could be.
From there on -as Jean-Jacques already has outlined- you could go deeper, mentioning an article that is well covering the question in place, do the static analysis, and so on.
And be prepared to wait for a long time until people get the understanding as they are mostly not thinking in this way for decades and therefor it is like learning a new language (this takes time of practice).
Best regards
Ralf
Ralf Lippold
chicken and egg?
Isn't the chicken-and-egg problem a basic SD example? You can't answer the question who was first, the chicken or the egg. You have to look at the chicken-and-egg system.
[Edited on 10-30-2008 by johan]
[Edited on 10-30-2008 by johan]