Thanks for a very informative and timely post. I am sure you simplified it too much to ensure the post is understood by less technical readers. But even if you don’t skip any of the details, do you think this is applied in any of the living banks, no matter what size? You gave a couple examples of study/research, but can it be applied in a raw business environment where C-level folks are just responding to their own ego trips and where revenue generation of today is seen more important than sustenance of the system long term?
Hi Anand, I think system dynamics can work for any organization in any industry, but not if the short-term is prioritized over the long-term at the company. Executive buy-in and having a champion are vital for success: you'll interview and gather data from people in different roles across the organization, validate model boundaries and key flows and model outputs with them iteratively, and need to get their buy-in on changes implemented as a result.
Every model skips some details. The keys are getting the boundaries right and including enough detail to identify all the important flows and leverage points, but not so much detail that the model becomes unwieldy and too difficult to understand. (For example, the procurement department flows are probably unimportant for a bank-run model.)
If you want to read some detailed case studies about *how* system dynamics teams got buy-in, identified some counterintuitive leverage points, dealt with pushback, and succeeded in transforming companies for the better, I highly recommend John Sterman's book Business Dynamics (https://www.amazon.com/Business-Dynamics-John-D-Sterman/dp/0071068120). It's almost 1000 pages and doesn't skip any of the details (equations, first-order and second-order delays, etc.) but is surprisingly engaging; you can start modeling using this book too!
There's also a System Dynamics Summer School that's online across various time zones in July (membership fee and course fee, scholarships available): https://systemdynamics.org/summer-school/
Also, to add: in some of the cases, the Sterman book focuses on how the initial champion of system dynamics got buy-in by letting stakeholders participate in and see results of the modeling process through use of tools like MIT's The Beer Game, which encapsulates the concepts in an engaging, immediate-feedback way. (Unfortunately, it's pay-to-play: https://mitsloan.mit.edu/teaching-resources-library/mit-sloan-beer-game-online)
Anand -- your comment about C-level has a lot of merit. The very thing that can lead to C-Suite CAN be in conflict with systems approach, at least IMHO. I have old colleagues who remained at a company which was system-oriented and employee-owned. They went public and reoriented structures and processes more into vertical divisions. Not sure this is real cler but I agree with your points somewhat.
And yes, I can definitely write a system dynamics article focusing on climate change. I'll probably walk through examples of how system archetypes map to various climate change scenarios and (maybe a separate essay) highlight leverage points identified by the En-Roads climate simulator (https://www.climateinteractive.org/en-roads/).
Thanks for letting me know what you'd like to see more of! Much appreciated.
Maybe something like the David Chapman argument that AI is mainly a PR stunt. What would getting enough bad PR on AI that it slows or stops look like? Is there a leverage point there? For climate change, what I'm interested in is what are the leverage points to use to actually get useful action on it.
Re: bad PR: it largely didn't stop the Bing rollout, though Microsoft did shorten user sessions, so it slowed it down a little and raised awareness. As another proxy, maybe watch what happens next with Tesla self-driving: this (https://www.washingtonpost.com/technology/2023/03/19/elon-musk-tesla-driving/) is some pretty bad PR stemming from accident reports and complaints in a true safety context; will it have an impact on slowing or stopping pursuit of self-driving (arguably it may have), on re-thinking the approach, on improving controls?
Interesting ideas, I'll see if I can map them out and how they look.
Awesome synthesis of a complex topic! Thank you for making it so easy to understand.
I have been thinking about a similar approach to build a better understanding of causal events and their relationships that lead to patient harm in a surgical procedure involving a medical device. There are a lot of moving parts, from start to finish, in a typical develop-build-launch-manufacture-ship-use cycle of a typical medical device. As a result, if you look at adverse events data, generally a direct link to a device issues is not found when injuries are reported. Yet, patients are getting hurt. So how do we improve our system? Difficult question, but I have a feeling a systems approach can help.
Thanks, Naveen! For a deep dive into how system dynamics modeling works--and to start building your own models that capture key flows in the development and launch process and run simulations to try to discover if changes to certain flows increase or decrease incident rates--I'd highly recommend John Sterman's Business Dynamics (https://www.amazon.com/Business-Dynamics-John-D-Sterman/dp/0071068120).
Also, see my response to Anand's comment - there's a System Dynamics Summer School in July run by the System Dynamics Society: https://systemdynamics.org/summer-school/
Very well written! Getting people and business buy-in for system thinking is hard and making it accessible helps a lot. While it is implicit in some things, once it involves people there seems a lot of resistance in my experience.
Yeah, see my response to Anand's comment; getting buy-in for changes that can take a while to flow through the system pretty much requires an internal champion who can defend the project until its benefits become clear.
That might be a senior manager who brought in system dynamics consultants, it might be a C-level executive, it might be the head of a modeling department that starts using system dynamics on a pilot project as proof-of-concept, but someone needs to press the gas pedal until momentum starts doing part of the work.
I worked for a long period for a scientific company with a large subset of employees with advanced degrees. It was employee-owned so less of the Q2Q pressure. System thinking pervaded the place. I thought the whole world thought this way and realize nearly no one has gone this route. It helped that significant portions of our work was designing systems and human controls. Definitely not for everyone and it is only part of a curriculum for a small number of people and mostly electives.
It would be amazing to talk more with you about this, since it sounds like you saw an ideal use of systems thinking (and system dynamics?) in action. Will send you an email.
Thanks for a very informative and timely post. I am sure you simplified it too much to ensure the post is understood by less technical readers. But even if you don’t skip any of the details, do you think this is applied in any of the living banks, no matter what size? You gave a couple examples of study/research, but can it be applied in a raw business environment where C-level folks are just responding to their own ego trips and where revenue generation of today is seen more important than sustenance of the system long term?
Hi Anand, I think system dynamics can work for any organization in any industry, but not if the short-term is prioritized over the long-term at the company. Executive buy-in and having a champion are vital for success: you'll interview and gather data from people in different roles across the organization, validate model boundaries and key flows and model outputs with them iteratively, and need to get their buy-in on changes implemented as a result.
Every model skips some details. The keys are getting the boundaries right and including enough detail to identify all the important flows and leverage points, but not so much detail that the model becomes unwieldy and too difficult to understand. (For example, the procurement department flows are probably unimportant for a bank-run model.)
If you want to read some detailed case studies about *how* system dynamics teams got buy-in, identified some counterintuitive leverage points, dealt with pushback, and succeeded in transforming companies for the better, I highly recommend John Sterman's book Business Dynamics (https://www.amazon.com/Business-Dynamics-John-D-Sterman/dp/0071068120). It's almost 1000 pages and doesn't skip any of the details (equations, first-order and second-order delays, etc.) but is surprisingly engaging; you can start modeling using this book too!
There's also a System Dynamics Summer School that's online across various time zones in July (membership fee and course fee, scholarships available): https://systemdynamics.org/summer-school/
Wonderful, positive, step-by-step -- so great
Also, to add: in some of the cases, the Sterman book focuses on how the initial champion of system dynamics got buy-in by letting stakeholders participate in and see results of the modeling process through use of tools like MIT's The Beer Game, which encapsulates the concepts in an engaging, immediate-feedback way. (Unfortunately, it's pay-to-play: https://mitsloan.mit.edu/teaching-resources-library/mit-sloan-beer-game-online)
Anand -- your comment about C-level has a lot of merit. The very thing that can lead to C-Suite CAN be in conflict with systems approach, at least IMHO. I have old colleagues who remained at a company which was system-oriented and employee-owned. They went public and reoriented structures and processes more into vertical divisions. Not sure this is real cler but I agree with your points somewhat.
I'm interested in modelling the system dynamics of climate change mitigation and AI X-risk. Think you'll get out articles on those at some point?
I have a topic in my to-write list on "Meadows’ Leverage Points Applied to AI Regulation" referring to this article: https://donellameadows.org/archives/leverage-points-places-to-intervene-in-a-system/ - I can move that up the list and start writing it. Are there other specific angles on AI risk you'd like to see me cover too?
And yes, I can definitely write a system dynamics article focusing on climate change. I'll probably walk through examples of how system archetypes map to various climate change scenarios and (maybe a separate essay) highlight leverage points identified by the En-Roads climate simulator (https://www.climateinteractive.org/en-roads/).
Thanks for letting me know what you'd like to see more of! Much appreciated.
Maybe something like the David Chapman argument that AI is mainly a PR stunt. What would getting enough bad PR on AI that it slows or stops look like? Is there a leverage point there? For climate change, what I'm interested in is what are the leverage points to use to actually get useful action on it.
Re: bad PR: it largely didn't stop the Bing rollout, though Microsoft did shorten user sessions, so it slowed it down a little and raised awareness. As another proxy, maybe watch what happens next with Tesla self-driving: this (https://www.washingtonpost.com/technology/2023/03/19/elon-musk-tesla-driving/) is some pretty bad PR stemming from accident reports and complaints in a true safety context; will it have an impact on slowing or stopping pursuit of self-driving (arguably it may have), on re-thinking the approach, on improving controls?
Interesting ideas, I'll see if I can map them out and how they look.
Awesome synthesis of a complex topic! Thank you for making it so easy to understand.
I have been thinking about a similar approach to build a better understanding of causal events and their relationships that lead to patient harm in a surgical procedure involving a medical device. There are a lot of moving parts, from start to finish, in a typical develop-build-launch-manufacture-ship-use cycle of a typical medical device. As a result, if you look at adverse events data, generally a direct link to a device issues is not found when injuries are reported. Yet, patients are getting hurt. So how do we improve our system? Difficult question, but I have a feeling a systems approach can help.
Thanks, Naveen! For a deep dive into how system dynamics modeling works--and to start building your own models that capture key flows in the development and launch process and run simulations to try to discover if changes to certain flows increase or decrease incident rates--I'd highly recommend John Sterman's Business Dynamics (https://www.amazon.com/Business-Dynamics-John-D-Sterman/dp/0071068120).
Also, see my response to Anand's comment - there's a System Dynamics Summer School in July run by the System Dynamics Society: https://systemdynamics.org/summer-school/
Very well written! Getting people and business buy-in for system thinking is hard and making it accessible helps a lot. While it is implicit in some things, once it involves people there seems a lot of resistance in my experience.
Yeah, see my response to Anand's comment; getting buy-in for changes that can take a while to flow through the system pretty much requires an internal champion who can defend the project until its benefits become clear.
That might be a senior manager who brought in system dynamics consultants, it might be a C-level executive, it might be the head of a modeling department that starts using system dynamics on a pilot project as proof-of-concept, but someone needs to press the gas pedal until momentum starts doing part of the work.
I worked for a long period for a scientific company with a large subset of employees with advanced degrees. It was employee-owned so less of the Q2Q pressure. System thinking pervaded the place. I thought the whole world thought this way and realize nearly no one has gone this route. It helped that significant portions of our work was designing systems and human controls. Definitely not for everyone and it is only part of a curriculum for a small number of people and mostly electives.
It would be amazing to talk more with you about this, since it sounds like you saw an ideal use of systems thinking (and system dynamics?) in action. Will send you an email.