Why is system dynamics used by so few when it is so good?

System dynamics solves difficult problems with ease, but it cannot deal with simple ones.

- Not the bathtub now again...! My children are looking at each other around the dinner table, shaking their heads. Daddy has just presented his all-purpose solution to any of their problems. - Look at it as if it was a bathtub, I explain to them. - A good relationship is like a high level of water in the tub. It builds up over time as the result of opening the kind-words-and-deeds faucet, accumulating trust, confidence, and friendship between people.

bath tub.pngThe bath-tub metaphor has become an integral part of my thinking. The concept of reservoir (tub), accumulation (faucet), and draining is the simplest, most general and powerful way I can think of when it comes to explaining situations (states), how they appear, and how they can be changed into more desirable situations.

System dynamics is a field that tries to take the full potential out of the bathtub metaphor, applying it to areas such as markets, population dynamics, national and global economies, human relationships, climate changes, productivity, inventory management, product life cycles, and so on.

Using only two basic building blocks (stocks and flows) system dynamics modellers can capture the essence of change. I think this is why system dynamics is one of the most beautiful disciplines around. With its tiny vocabulary (stock and flow) it can express key elements of any change that we experience – most of the time not even being aware of it.

System dynamics uses easy and intuitive graphical diagrams to illustrate the composition of models. This makes system dynamic models much easier to understand, communicate, and maintain than corresponding spreadsheet models.

I created my first system dynamics tool in the late 1980's. It was a DOS-based simulation tool called SimTek, and we used it to develop training simulators for people operating valuable and potentially hazardous industrial processes.

System dynamics is better known for its ability to model soft systems, and the company I ran at the time, also started to move towards business modelling. In 1993 Powersim Constructor was finished, and in 2000 Powersim Studio was released as the last product I was responsible for during my time with Powersim.

My son has always shown an interest in computers, and when he turned 10 years old, I thought it was about time that he started to learn how to model. The initial problem I challenged him to solve was to model his own savings account. He was already fluent in bathtub thinking, and could easily imagine his bathtub account with an inflow of deposits and outflow of withdrawals. An extra inflow was needed for adding interest paid into his account by the bank for letting them borrow his money. Below is a picture of the account model (using a PowerPoint-inspired version of the stock-and-flow diagram. In Dynaplan Smia we call this a box-and-arrow diagram).

account.jpg

The exercise was a disappointment to both of us. On the conceptual level, my son had no problems to assemble the model together. But he wanted the model to become operational, in the sense that he would enter information about every deposit and withdrawal he made and every change in interest rate that the bank might do. It turns out that you must be quite an expert in order to create a system dynamics model that represents accurate, transactional information. In addition, the graphical user interface of Powersim Studio, and any other system dynamics tool I know about, does not provide convenient means for the user to enter and display data at the level of detail that we needed for my son's first model.

System dynamics is slightly older than the spreadsheet. A system dynamics model is more visual, has better quality control, and it can capture the reality more closely than a spreadsheet. However, the level of acceptance among users is quite opposite. Spreadsheets are used by almost everyone, system dynamics by very few – at least relatively speaking.

I believe that part of the reason is related to the technology that has sprung out of the system dynamics field. Using system dynamics tools, it is almost as hard to model a teenager's savings account as it is to model the world economy. (Forrester's world model has only five stocks). By reformulating the statement, we see how powerful system dynamics can be: It is almost as easy to model the world economy, as it is to model a teenager's savings account. (An aggregated view of the world model is displayed below).

world model.png

This observation is somewhat in line with Forrester's claim that important questions are not harder to solve than the unimportant ones. I agree with this. But from a market perspective, how many people are involved in solving global issues at work every day? (Such problems are "solved" by the most of us over a beer after work).

The spreadsheet is good for solving trivial problems. But since everyone knows how to use it, it is tempting to (ab)use spreadsheets to solve important, dynamic problems that are way beyond the spreadsheet's league.

For system dynamics, the matter is the opposite. It is a field that is perfect for dealing with holistic approaches to long-term, dynamic problems, but it is unsuited in dealing with conceptually simple questions that millions of office workers face at work every day.

Out of one hundred problems that require modelling, I guess that only one is framed as a non-linear, dynamic problem. The rest are static or linear in nature. Using system dynamics, it is harder to solve simple static problems than it is to solve complex dynamic problems. I think this in itself is enough to explain why it has never become the first choice for the masses. In addition, there are other hurdles that prevent system dynamics from entering the mainstream.

tools problems.png

You may already be aware that spreadsheets and system dynamics have their strong and weak areas in different places. With the development of Dynaplan Smia we try to provide a technology that maintains the strong sides of both platforms, while avoiding the weak spots.

PS
Is system dynamics really a "good" tool? The answer depends upon the application. When problems are addressed in a wide context, using a holistic, aggregated approach, it typically turns out that most problems are dynamic, involving feedback, delays and non-linearities. Then system dynamics is your best tool. On the other hand, if you are focusing on details, short time horizons, and isolated parts detached from the big picture, there are probably better tools than system dynamics to meet your needs.

Like what you read?

Give me more You must log in before you can subscribe to email notifications on new articles.

Comments

Awesome article... you've helped me answer a question that's been bugging me: It’s very interesting that SD seems to have fallen off the radar screens of the large consultancies… After a surge in interest n the 1990’s following the popularization of the “Beer Game”, the People Express (and other) Management Flight Simulators, and The Fifth Discipline… recent activity has become more “academic” and less “practitioner” based (corporate and consulting). A media / web search reveals: • Number of mentions in the Financial Times, 2003-2008: 3 for “system dynamics”,6 for “systems thinking”, 64 for balanced scorecard (as an example of a popular business method) • Mentions in the Economist, 1997-2008: 2 for system dynamics, 4 for systems thinking • Mentions in New York Times, 1981-2008: 4 for system dynamics, 8 for systems thinking • Mentions in Harvard Business Review since 1979: 5 for system dynamics, 10 for systems thinking, 50+ for balanced scorecard • Mentions of system dynamics in a 100-page bi-annual survey by Bain and Co. of tools used by management: zero. (In an earlier ranking around 1992, “systems thinking” ranked 23rd, used by less than 5% of respondents). Trying to solve all problems with SD approach could be part of the answer.

Very informative article. I would like to add a couple of comments:

1. "1 out of 100 problems is framed dynamic" - here emphasis should indeed be on framed. Most problems are presented or modeled in a way that suggests they are static - the concept of time is left out or greatly simplified. But actually most problems in business are dynamic that is time is important to the understanding of a problem. This starts with the fact that a problem has a history. It is good System Dynamics practice to start with charting time series for crucial variables or parameters of a problem starting with some time back and plotting possible or feared projections also. Besides for good old OR-problems requiring linear or quadratic programming a great deal of problems should better be looked at as being dynamic. (By the way next to linearity it is especially delays that screw up spreadsheets and the like - but are natural to SD)

2. System Dynamics tends to offer a high level view of a problem that is, it is a strategic method. It helps to figure out how to set up things so that they can smoothly evolve in everyday operations. And again, a lot of typical consulting work could very well profit from System Dynamics being involved.

3. System Dynamics needs people being educated to think that way. Unfortunately SD is not very widely spread at schools and there is much to gain from it. It offers a very great explanatory power if one has succeeded to simplify complicated models for presentation. Again building highly aggregate models like Forrester's World Model is an art and needs a lot of practice (like writing a short letter). I would believe that System Dynamics has even more to offer for politics and political journalism than in Business. Public Policy problems are actually much better suited for SD:  they typically involve many different stakeholders and subsystems, long term effects are especially important and there is a policy setter here that wants to ensure that the system can run without influence once he has set the rules. Much more work should go in this direction.

Thank you for your comments, Guido! Your points reinforce what I try to communicate in my original posting.

  • The spreadsheet is a software without methodology, and still it is extremely successful.
  • System Dynamics is software with methodology, but it struggles in vain to gain market share.

Could it be that the reason why dynamic thinking (methodology) and dynamic tools are underutilized is to be found within system dynamics itself?

  • The narrow focus of the system dynamics community contributes to defining it as a niche activity exercised by a few experts.
  • The limited feature set of the traditional system dynamics software narrows its usability in government, business, education, and homes.

The focus of Dynaplan is to address strategic as well as operational problems, and to provide a technology that beats the spreadsheet not only on dynamic problems, but also in areas where the spreadsheet is at its best.

Best regards,

Magne

I can't help it but this time I would rather have some objections:

  1. Most system dynamics software can be used as pure sketchboard for mapping processes, networks, decision trees or causal loop diagrams - so they are very, very flexible. Indeed some of them are available for free and worth just for these kind of mappings. It is not the narrow bounds of a methodology that hampers SDs success.
  2. Looking through the papers presented at the Conferences of the System Dynamics Society will provide ample evidence for the breadth of applications as opposed to a narrow focus. You are right though that it is mostly experts doing it but that again supports what I said about the education system:  Most people know how to add numbers (use a spreadsheet) but very few know to draw a stock and flow diagram.

  3. The tools out on the market can do stochastic sensitivity runs (Monte Carlo), dynamical optimization, calibration (econometrics), games and can be quite easily coupled with spreadsheets and external programs allowing for a vast range of applications (up to agent based modeling). So it is definitely not the limited feature set that blocks the application of SD.

I will take a look at Dynaplan but to me it is more strategic applications that matter. As Barry Richmond put it: SD is for taking the 10.000 feet view of things. If you are down to counting individual widgets you might have missed the point.

Thanks for your input, Guido. It is much appreciated.

Assuming that you are right — that there is nothing wrong with the way system dynamics is practiced and that the tools are just perfect — what is in your view the reason why the field has not become stronger after 40 years of development?

Regarding your objections, I actually agree that system dynamics technology has a lot of functionality. However, some critical features are practically missing. Features that can explain why the technology competes poorly with alternative modelling tools.

You are also right that one should avoid focusing on a too low level. But it is equally bad to sit on a cloud too high up in the sky. The challenge is to find the right balance between too much detail and oversimplification.

Best regards,
Magne

Let us know what you think

You have to log in before you can comment on this article.

Please enter your account information to log in.

Tags