OK, this post will involve a bit of what I like to call "the conceit of the engineer", which is that many of us tend to think of our skill set as a hammer and when we look around, we see a lot of nails, most of which need a good pounding. So, I tend to see anything that has a bunch of pieces, or can be modeled with a differential equation, as an application of systems theory. So, today I'm going to simply write a bit about two basic examples in straightforward terms: Climate Science, and Health Care Delivery. Finally, I'm going to describe a tiny bit about how I see systems theory applying to one of science's current Big Debates: Open Access.
Climate Science: I'm no expert. This will be essentially at the third grade level. But this is an area where the applicability of systems theory should be apparent upon a moment's reflection to anyone. The water cycle, for example, is a subsystem of climate science. And a deeply troubling one. As the earth warms, enormous quantities of fresh water currently trapped in ice are at risk to melt. At the same time, warming air will hold more water vapor. Having huge amounts of water in these more volatile forms (ice is pretty sedentary) allows for it to move and change form more readily, meaning additional rainfall, heavy snows when the temperature does drop below 0 degrees Celsius, and rising coastlines.
Of course, there are also a lot of hollywood-style doomsday scenarios to which you probably shouldn't give a lot of credence. It is incredibly difficult to make specific predictions regarding local effects of climate change in any area smaller than say, Western Canada. Localized effects of these very complex systems are (a) highly sensitive to initial conditions, (b) not fully understood even in the realm of theoretical abstraction. As a result, even if we had an incredibly faithful climate model of the whole world, that modeled down to the level of the briefest zephyr, it's going to get the specifics wrong most of the time.
So why bother? Because it will probably get the global behavior correct in ways that are critically important and vital to planning for the consequences of climate change. It doesn't matter if you successfully predict the weather in Astoria, Oregon on Sept 3rd, 2012 (probably kind of grey and rainy). It matters a lot if you want to know things like "within 10%, what will the mean high tide on the western seaboard be over the course of the next 20 years". Climate science is doing an excellent job of generating useful predictions of those sorts. So local weather effects, like hot or cold waves, like tornadoes, hurricanes etc., are not going to be predictable in any useful way by climate models (though, they will likely be good at predicting that they will increase, or change basic locations, like moving north). Systems theory is good at predicting global (system-wide) effects, rather than local (small "features") effects.
So in summary, pay attention to the global predictions. Small-seeming effects over large areas and long time periods. These are the things that climate science is really good at predicting. Weather? That's always going to be random. But the media gets it wrong a lot. Go to the source, if you can, and you'll see that the actual papers explain their confidence intervals, limitations, assumptions, etc. Those things are rarely reported.
Health care delivery has similar strengths and weaknesses. The models are, as yet, not nearly as sophisticated as those in climate science. This is due to a few simple things: the discipline is younger, and so hasn't had as much time to develop, and health care delivery deals with units which are not as easily divisible as those in climate (i.e., it's more natural to deal with non-integer quantities of water then with people).
However, many key concepts are similar. People and material flow from place to place over time. Stocks and flows and relationships between objects and subsystems all contribute to the global behavior of the system. I work at the middle level of systems analysis when it comes to healthcare delivery. Large scale healthcare systems modeling looks like climate science or economic modeling. Determining how huge numbers of people and processes move with very little specificity. Small scale systems work in healthcare delivery is at the level of the individual: feedback control systems for artificial limbs, mannequin simulators, anaesthesia pumps.
Mid-level systems work in healthcare delivery is what I do: optimizing clinical systems. Determining how clinical policy and capacity is likely in influence the health and outcomes of the patients who visit that clinic. When the policy or infrastructure of a hospital, or a clinical system, changes, this influences how the subsystem of providers interacts with the subsystem of consumers. Ensuing increased or decreased efficiency will result in increased or decreased capacity. Those who do receive care may have better or worse outcomes. It will be non-obvious how these things interact. To do this, we need to model how patients receive care, how they arrive, what care and processes they will need, etc.. Then, but tuning the system, changing policies, capacities, resources, we can simulate how best to serve the population that depends on the system for care.
Finally, I'm going to wade into dangerous waters. Open Access. There's a big movement in the scientific community to require that all research be published in open access journals, meaning that authors pay to publish their research, presumably using grant or institution money, and then the papers are free to libraries and individuals. It's an interesting model. Currently, most science is published by journals which (usually) publish without charge to the author, and then charge for access by libraries an individuals. Journal subscriptions are very expensive, and often bundled in ways which are not beneficial to the consumer. It's a serious expense for libraries and researchers.
I'm not here to take a position on the debate as to which one is better. I would merely like to point out that the systems which produce and publish scientific knowledge are vastly complex and interrelated, and form a complex system. It involves universities, governments, libraries, private research facilities, for-profit corporations and non-profit foundations. Many are intertwined in ways that are difficult to understand (and I don't claim understanding).
Many foundations publish their own journals, under the closed access model, and take the proceeds from that to further their foundation's agendas, and reinvest in granting or other research activity. Some closed access journals are run from universities, and support professor's lines. The flow of science and knowledge and money and education is enormously complicated. Changes to this system will have unpredictable consequences.
People who argue on both sides of this debate seem to me to take very reductionistic views of how changes will influence the industry. If systems theory tells us anything about these very complicated interactions with many different agendas and agents, it's that we cannot predict the outcomes simply by thinking about them. I suspect that a large model (presumably using systems dynamics) which captured the flow of money, time, resources, and publications in the various types of publishing models would be very enlightening. We could see, what are the global effects on publishing if we augment the open access sector, or if we institute policies requiring greater levels of open access? What happens when non-profit models (and open access does not necessarily mean non-profit) are given favor over for-profit models?
Making predictions about how large systems will change based on the adoption of new policy is extraordinarily difficult. It requires a lot of math, and a lot of computing power. People who tell you how their policy will improve or change an entrenched system, without doing some kind of reasonably sophisticated systems analysis, will almost certainly be wrong. Remember 50 years ago (I don't, but I've read about it) when people were worried about a coming ice age? They weren't stupid. They just hadn't done the systems theory.