In Systems Thinking, A Slinky Does Its Thing.

January 4, 2011 1:10 PM

Why do we continue to make so little progress in addressing our many, very pressing social concerns? My answer is that the way we think, communicate, and learn is outdated.... it is the premise of this Chapter (and Systems Thinking) that it is possible to evolve our thinking, communicating and learning capacities. (pdf)

The above chapter is from a software manual that morphed into a book, An Introduction to Systems Thinking with STELLA by Barry Richmond. More articles here.

Download the free iSee Player and check out proposed models of global problems in the "Stories of the Month" section of the software site.

posted by zennie (30 comments total)

Well, we've made significant progress on AIDS...
posted by Chuffy at 1:18 PM on January 4, 2011 [1 favorite]

All systems fail (PDF Warning)
posted by Chuffy at 1:21 PM on January 4, 2011 [3 favorites]

I'm having trouble reading the PDF in the main link... anyone else?
posted by codacorolla at 1:29 PM on January 4, 2011 [1 favorite]

"The significant problems we face cannot be solved at the same level of thinking we were at when we created them." -- Albert Einstein
posted by blue_beetle at 1:30 PM on January 4, 2011 [6 favorites]

Didn't the last time we tried evolving our thinking, communicating and learning capacities end up involving large amounts of drugs and tie-dyed shirts? I don't mind the drugs, but do we really want to risk tie-dye again?
posted by happyroach at 1:30 PM on January 4, 2011 [5 favorites]

I'm finding his odd emphases, exclamation points, and pacing to be extraordinarily distracting! And, it's making me immediately skeptical of his thesis!
posted by thinkpiece at 1:38 PM on January 4, 2011 [8 favorites]

I'll admit that I just skimmed the first few pages of the linked paper, but his premise is a bit damaged by his apparent lack of knowledge about the pervasive use of systems theories and system thinking in the social sciences, particularly in social work at all levels (micro, mezzo, and macro). What if we've already been trying the solution he thinks he's identified? WHAT IF THE CALL IS COMING FROM INSIDE THE HOUSE?
posted by OmieWise at 1:47 PM on January 4, 2011 [5 favorites]

This is interesting. I actually find it really intriguing, though I don't really agree with much of it.

My first thought was that the writing style and framing reminds me distinctly of a Scientology manual. I don't mean that in a bad way; I own several, because I find them to be lots of fun to go through trying to sort out exactly what they're trying to say about thought.

I've read ten pages of the main link now, and I'll go over the rest too, but I have two immediate objections. First, it's sort of quaint that this assumes that learning is possible. That might be true, but we need to say a lot more before we can make a bold claim like that. Second, I'm skeptical of any systematic explanation of thinking that doesn't even seem to mention sex.
posted by koeselitz at 1:49 PM on January 4, 2011 [1 favorite]

from main article, p.11: “Each employee interacts with a raft of people outside the organization who, in turn, interact with others, and so on. So, in the social domain, being able to think horizontally is essential!”

Ah. Maybe I was wrong...
posted by koeselitz at 1:53 PM on January 4, 2011 [3 favorites]

STELLA was a great piece of software, and I saw (back in the day) how powerful it could be when you started working with, say, secondary science teachers on the crazy new field of computational science. I know it has been covered here before, but it is always worth mentioning Stafford Beer when talking about attempting to Design Freedom. Tangents: Heinz von Forster I liked for being highly technical about the small stuff and then connecting it to the big picture; Herbert Brun was a composer and more, connecting cybernetics to electronic music and with a plan on how opera could save humanity.

For those readers that need sex, I suggest Maturana and Varela's The Tree of Knowledge on the biological basis of human understanding. Thanks for posting this, STELLA was pretty visionary WRT to believing how significant this change in perspective could be.
posted by cgk at 2:14 PM on January 4, 2011 [2 favorites]

Didn't the last time we tried evolving our thinking, communicating and learning capacities end up involving large amounts of drugs and tie-dyed shirts? I don't mind the drugs, but do we really want to risk tie-dye again?

Seriously, I think the drugs had a lot to do with our not evolving since then (but believing we had), and tie-dye turned out to be a gateway to disco polyester.
posted by oneswellfoop at 2:38 PM on January 4, 2011 [1 favorite]

I thought the main problem was that we're not ants.
posted by ZenMasterThis at 2:42 PM on January 4, 2011 [1 favorite]

Why do we continue to make so little progress in addressing our many, very pressing social concerns?

REPUBLICANS. Next!
posted by JHarris at 3:10 PM on January 4, 2011 [13 favorites]

Chuffy: Good link, anyone interested in where Dave Woods got his ideas should check out Jens Rasmussen's risk management framework. It's an attempt to describe complex system failures as a dynamic sociotechnical system.
posted by anthill at 3:12 PM on January 4, 2011 [1 favorite]

I think the drugs had a lot to do with our not evolving since then

Not enough drugs, lack of drugs is what shut it all down I swear we were so close.
posted by Meatbomb at 4:08 PM on January 4, 2011

Most of what I'm getting out of following the links in this posting is that there's a rift in the "systems" community between System Dynamics and Systems Thinking. Many of the articles here have the theme that "those darn System Dynamics folks think that Systems Thinking is just a step in their process, but we think that System Dynamics is just a step in ours." Huh.

I see benefit to using modeling to think about complex systems, but I don't see the reason for this brouhaha. For my part, I've enjoyed playing with SimGua. It's got an awkward name, but their academic (and personal use) license is very affordable, and it's a nice enough tool.

I'm still looking for more insight into how to actually do system modeling, especially the verification/iteration part.
posted by dylanjames at 4:28 PM on January 4, 2011 [3 favorites]

The writing is opinionated, pseudoscientific and mathematically unrigorous. It is littered with assertions that are heavy-handedly simplistic, or supported in the literature. Figure 1-15 (page 31) is a ridiculous caricature; it is offensive to the fields of psychology, physics, and biology—nice job in achieving that simultaneously. More than that, it highlights the fundamental weakness of this approach: supposing someone constructs a really bad model like the one in Figure 1-15, there's still no way for the tool to help validate it. Garbage in, garbage out.

I can see the appeal in using concise, formal models to describe phenomena and even simulate them to get quantitative feedback. But if the author can't recognize or address the limitations and philosophical issues that are bound to be present in any such endeavor, it somewhat discredits their work.
posted by polymodus at 5:07 PM on January 4, 2011 [1 favorite]

This gentleman has a lot of answers.

But this gentleman has even more answers! To questions you didn't even know you had!
posted by Sidhedevil at 7:07 PM on January 4, 2011 [1 favorite]

: The writing is opinionated, pseudoscientific and mathematically unrigorous. It is littered with assertions that are heavy-handedly simplistic, or supported in the literature. Figure 1-15 (page 31) is a ridiculous caricature; it is offensive to the fields of psychology, physics, and biology—nice job in achieving that simultaneously. More than that, it highlights the fundamental weakness of this approach: supposing someone constructs a really bad model like the one in Figure 1-15, there's still no way for the tool to help validate it. Garbage in, garbage out.

As I read it, the document is about thinking, not about being right. The author seems to agree about it being a caricature. From page 4:

... all models (mental and otherwise) are simplifications. They necessarily omit many aspects of the realities they represent. This leads to a very important statement that will be repeated several times throughout this Guide. The statement is a paraphrase of something W. Edwards Deming (the father of the “Quality movement”) once uttered: “All models are wrong, some models are useful.” It’s important to dredge this hallowed truth back up into consciousness from time to time to prevent yourself from becoming “too attached” to one of your mental models. Nevertheless, despite the fact that all models are wrong, you have no choice but to use them—no choice that is, if you are going to think. If you wish to employ non-rational means (like gut feel and intuition) in order to arrive at a conclusion or a decision, no mental model is needed.

Models are meant, ultimately, to replicate specific observed phenomena. Sometimes you care about the intermediate steps (academic, mechanistic), and sometimes you only care about inputs and outputs (applied, empirical). The thinking, the defining of a possible model, would be the first step in a long, iterative process of refinement based on comparing the model to the real world.
posted by zennie at 8:03 PM on January 4, 2011 [1 favorite]

Models are meant, ultimately, to replicate specific observed phenomena. Sometimes you care about the intermediate steps (academic, mechanistic), and sometimes you only care about inputs and outputs (applied, empirical). The thinking, the defining of a possible model, would be the first step in a long, iterative process of refinement based on comparing the model to the real world.

Fig 1-15 is so bad, it's not even wrong. Anybody with a cursory acquaintance with the liberal arts would observe that the 3 "time constants" are bullshit. Therefore there exists no refinement, using his framework, that leads to an improved model. That the author fails to recognize this, and instead willfully shoehorns in bad/ambitious examples into even the first Chapter of his book (more like marketing document), I find hugely problematic.
posted by polymodus at 8:41 PM on January 4, 2011

Thanks for your thoughtful reply, zennie.

zennie: “As I read it, the document is about thinking, not about being right.”

Ah, but even if thinking isn't about being right, isn't learning about being right? That is: isn't learning a method to get to knowing? Furthermore, if that's true, don't you have to nail down what knowing is before you can say exactly what learning is, and how you go about it? These seem like much more difficult things to do than the text implies; there's a sort of pragmatist simplicity about it all, as if learning and knowing fit into simple models, as if they're things we already understand fully.

To which, as you say, the response is: of course not. All thinking is done via simplified models, as is pointed out in the bit you quoted:

from text: “... all models (mental and otherwise) are simplifications. They necessarily omit many aspects of the realities they represent. This leads to a very important statement that will be repeated several times throughout this Guide. The statement is a paraphrase of something W. Edwards Deming (the father of the “Quality movement”) once uttered: “All models are wrong, some models are useful.” It’s important to dredge this hallowed truth back up into consciousness from time to time to prevent yourself from becoming “too attached” to one of your mental models. Nevertheless, despite the fact that all models are wrong, you have no choice but to use them—no choice that is, if you are going to think. If you wish to employ non-rational means (like gut feel and intuition) in order to arrive at a conclusion or a decision, no mental model is needed.”

This is a satisfyingly reflexive claim. After all, a model is not strictly a mental thing at all; it's a physical representation of something. So even the concept of a "model" is one of these metaphorical "models." This reinforces the sense that everything we think is an inadequate, simplified model. But I have two problems with this picture of thinking.

First of all, if it's true, how can we possibly know that this is how thinking works? If it's really true, then the things left out of our models might be the most essential things, the small details that show just how wrong the rest of our picture really is. We might in fact have the whole thing wrong. In fact, if this is really how thinking works, we can almost give up on the possibility of knowing, and thereby give up on learning; because anything we think we "know" is only a vague approximation, and it may turn out that our model is excluding the very footnote that defines the whole system entirely. It's not even any good to say that we can approximate learning – because again, how could you know that? It could be that every mental model we build takes us further and further from the truth about the world. It seems as though that's often what happens when we work with less than all the facts, right?

Second of all, isn't it possible that this picture of knowing is false? Sure, much of thinking is done by way of inadequate models – perhaps even most of thinking is done that way – but can we say that all of thinking is thus inadequate? What if there is a way to get to thinking about things as they truly are? It seems as though this would be the sort of thing that we can't really know until we get there; after all, we haven't thought all possible thoughts, so there may be a way to know. Until we actually know, it's impossible to eliminate true knowing as a possibility.

Finally, you (and the article) mention "the real world." I don't think there's any such simple thing as "the real world," and I think it's useful to take apart what exactly we mean when we use that phrase.
posted by koeselitz at 8:50 PM on January 4, 2011

(Also, a propos of nothing, I sincerely doubt that W Edwards Deming really meant to signify all models of thought when he said that. It's much more likely that he was talking about statistical models used in marketing products. That was his field, after all, and there's a significant difference between marketing and cognitive philosophy.)
posted by koeselitz at 8:53 PM on January 4, 2011

Fig 1-15 is so bad, it's not even wrong. Anybody with a cursory acquaintance with the liberal arts would observe that the 3 "time constants" are bullshit.

The figure isn't totally obvious but his model of a capacitor, at least, is valid. It's pretty well established that capacitor discharge is a first order process governed by time constant (RC). One could say that friendship decay is a time-dependent first order process as well. Maybe that is wrong but as an approximation of the simplest case it makes some sense. I think his point is that these processes may have something in common, even though, on the surface they don't. And, further, it may be possible to build a visual language that can describe memory loss and capacitor discharge without resorting to complex math that is out of reach to most people. And this language can help people understand diverse problems which they could not otherwise.

In interests of full disclosure I've used STELLA before so that probably helps.
posted by euphorb at 10:10 PM on January 4, 2011

Just so people are on the same page here, when we talk about discharging capacitors, you can mentally substitute in a different analogy (which is part of the point! Certain simple equations pop up in a lot of places). Most people don't have experience with discharging capacitors, but they likely have experience with objects changing temperature.

When an object is moved into an environment with a different ambient temperature, the object quickly starts changing its own temperature to try to reach the ambient temperature. The greater the difference, the faster the change. The change is not instantaneous; we capture exactly how fast the change is with a "time constant".

Regarding Fig 1-15 and who it insults: There are many different kinds of memory (implicit, semantic, episodic, sensory, ...). I can't make any statement about most of them. But I should note that I worked in a perception lab for a couple years and one of the projects I worked on involved sensory memory (not "I remember how the crayons smelled when I was in kindergarten", rather "My brain can automatically 'replay' a sensory experience that's occurred in the last couple seconds, so you don't need to repeat yourself; I'll just reinterpret the replay").

To avoid getting bogged down in a lot of detail that would not be of sufficient interest to most readers, I'll try to give the short and sweet version. My PI had found a visual illusion regarding stimuli that changed as they moved, and data regarding how people misperceived what was displayed to them, but he did not have an explanation for that data. He asked me to come up with one.

I said "Let's make a wild guess that sensory memory responds to new sensory input in the same way that a capacitor responds to a new voltage". I'm not a neuroscientist or anyone with any qualifications to understand sensory memory on a biological level; I'm a CS guy that understands capacitors better than neurons. I only mention this anecdote because I fit that model to the data and it predicted people's responses essentially perfectly (less than 1% relative error; error was indistinguishable from the quantization noise introduced by our measurement methods). The model only had one parameter- the time constant that specified how quickly sensory memory could update in the face of new sensory input- so I'm confident that the excellent fit was not a matter of overfitting.

My PI was disappointed in the work because it was not sufficiently theoretically motivated. He thought that we couldn't publish a wild guess unconnected to prior research, and he was probably right. But given the perfect fit of the model to the data, as far as I'm concerned, sensory memory does update in the same way (with a change proportional to the extant difference) that a capacitor (dis)charges.
posted by Jpfed at 2:52 AM on January 5, 2011 [1 favorite]

it's sort of quaint that this assumes that learning is possible
what

I don't think there's any such simple thing as "the real world,"
wait, what?
posted by tylermoody at 6:02 AM on January 5, 2011

: Finally, you (and the article) mention "the real world." I don't think there's any such simple thing as "the real world," and I think it's useful to take apart what exactly we mean when we use that phrase.

Interesting. All I meant by 'real world' was the observations of phenomena meant to be replicated. In Jpfed's excellent example, that would be the sensory memory data. I'm kind of surprised that the PI was disappointed, because it seems to me that Jpfed's empirical model would be an excellent starting place for what the PI was no doubt after, an explanatory mechanistic model (a good theory). Neurons are not entirely unlike capacitors; they store up chemical energy and then discharge like a camera flash. If the memory data was limited to a very brief interval after stimulation, then why wouldn't this be a good explanation? Well, lots of reasons, but that's where you go get more "real world" information from published literature or from additional experimentation, and you test anything that could prove the working mechanistic model wrong.
posted by zennie at 6:38 AM on January 5, 2011

This seems like an intro to systems theory for a self-help-guru seeking audience. Another book for people who want to know why Bill Gates makes more money than they do and who think they can remedy that with know-how and gumption.
posted by outlandishmarxist at 8:19 AM on January 5, 2011

I am very, very skeptical that social ills difficult to solve because we aren't using the right conceptual flowchart. Rather, I suspect they are difficult to solve because 1) social problems are very complex 2) potential solutions are not amenable to the sort of empirical tests available to physics, 3) we disagree about what social facts constitute social ills, and even when we do not, we disagree about the relative severity of various ills.

For example, is the decline of opposite sex marriage rates a social problem, or a social achievement? Even if it is a problem, should we invest much or any money into a remedy, or would that money be better spent on improving the free lunch program for impoverished public school students? And if we do invest money into a remedy, what do we have to invest in to make the marriage rate go up?

Even if we adopt the author's magic conceptual bullet, we won't get answers to any of those questions. (The answer, by the way, is forget marriage and invest in free lunch). We won't even get explanations for those social facts. What we will get is a chart that gives the appearance of formal rigor when we really have judgment calls.

And even if the flowchart method does give us a good solution, how are we to implement it? Will a flow chart help me convince Congress to enact my plan of having the schools with the highest and lowest per-capita student spending trade funding levels? (Now let those smug suburbanites tell me that funding doesn't matter!) Or enact a federal guarantee of adequate funding for every public school student? Not likely. People interested in these problems are better off with a Sociology 101 textbook and leave STELLA for making illustrations.
posted by Marty Marx at 11:40 AM on January 5, 2011

*social ills ARE difficult to solve. Of course.
posted by Marty Marx at 11:53 AM on January 5, 2011

It seems to me we could solve quite a few social problems just by looking around and copying what other countries do. But that presupposes that your problems and my problems are the same, which is probably not the case. For example, if my children go to a private school, then a reduction in my taxes will increase funding for their education. The system works!
posted by ryanrs at 2:22 PM on January 5, 2011