Systems Thinking and the emergence of new Life

T.S. Eliot, in the Four Quartets , said

We shall not cease from exploration
And the end of our exploring
Will be to arrive where we started
And know the place for the first time.

I'd like to suggest, as others have, that the process of exploration is more a spiral or helix than a circle, and we, individually and collectively, may go through multiple cycles of coming back to where we were, except that now it's "different" than the last time we were "here."

There is a motion, very much akin to an actual mechanical screw, in which we apply a rotational force or motion, but the result is a much harder to achieve motion in a wholly different direction.

As I sometimes do, I was rereading Practical Challenges of Systems Thinking and Modeling in Public Health this morning, seeing if the article had changed in meaning since I last read it [ Trochim WM, Calebra DA, Milstein B, Gallagher RS, Leischow SJ, American Journal of Public Health, March 2006, V96, No. 3]. Most of those people are at Cornell, in fact, in my old stomping ground, MVR Hall. Bobby Milstein is at the CDC, and Scott Leischow at NCI.

As one of my favorite philosophers, Charles Schulz's dog Snoopy, noted one day:

Did you ever notice
that if you think about a problem at 2 AM
and then again at noon the next day
you get two different answers?

Anyway, I got a different answer I wanted to share.

Still reverberating from my post on the Ten Lessons from Physics yesterday, I realized there is another candidate for important lessons in reducing complexity to a manageable level from physics that may be relevant to "Systems Thinking."

I already described yesterday how immensely complex interactions at one level, such as gas molecules, can get reduced to a few very simple but aggregate-scale concepts on the next level, such as "pressure, volume, and temperature", that are meaningless words on the individual molecule's level. In between those scales there was a forbidden zone, computationally intractable from either end.

Near the first easy end, only a few things only occasionally interacting, we have some room to actually compute the motions of several planets or billiard balls - much to the dismay of students of introductory physics.

What I can bring home from my foray into physics, however, is the news that there is also an easy zone at the other end - the end of immense numbers and immense interaction.

There is a continuum actually, and a sort of "continuous paradigm shift" as you go from the first easy end (few things, little interaction) to the other easy end (huge number of things and interactions) and that is in "what goes away."

At the lower easy end, the interactions go away, and we think in terms of "objects", mostly, that have properties, such as position and velocity.

At the upper easy end, it is the objects that go away, and we think in terms of persistent interactions that have self-energy and properties. This "regime" is the one that most of the visible universe inhabits, except for some relatively cold rocks (planets). It is the world of plasma, of high-energy physics, of stars, of the huge interactions at the centers of galaxies.

This is akin to the near-field, mid-field, and far-field of a radiating antenna - the math is relatively easy at either very near the antenna, or very far from the antenna - it's only at intermediate distances away that things become totally strange and slippery. I discussed that issue before.

In any case, we have here a two-dimensional entity, not a one-dimensional one, so it is now capable of being non-transitive and forming what Hofsteader called "a strange loop", akin to M.C. Escher's waterfalls or staircases. As the number of actors and intensity of interactions increases, one dimension, objects, diminishes in importance, the other dimension, interactions, increase in importance, until we reach the limit point, mathematically, and physically, where the initial objects simply drop out of the equation entirely and we are left only with the persistent interactions - which, voila, close the loop, the snake bites its tail, and these persistent interactions become the "objects" for the next level of existence and we start our next turn
around the greater helix of Life.

It's a smooth transition, easily modeled, not a "leap". It only looks like a leap if we go across the screw threads, not along them.

So, the meaning of "an entity" is really only locally defined, in an asymptotic way.. well below its level, the concept or entity is too large to be perceived, as "temperature" is to a molecule or "admission to Harvard" is to one of our body's cells. Then, as we think about sliding up to larger and larger scales, this concept has very strong meaning and dominates the local equations in the local time frame at the local scale, and then, as we keep on going, it goes away again forever, as does our problem starting our car on January 12th in terms of international relations.

OK, nice metaphor, but how is it useful? Do we have a name for these "persistent interactions"?

Aha! In physics, one type of persistent interaction is a "soliton". I'll skip the details, but these are waves that don't die out as they travel, but just keep on going. Fascinating.

What I do want to suggest for these persistent interactions is, at one scale, the idea of self-aware regulatory feedback loops. This concept is applicable at any scale, size, or time-frame and seems to be invariant and a core building block of life. Cells do it. The endocrine system does it. People do it. Corporations do it. Nations do it. Ecosystems do it. Policy makes and lawyers actively work on "regulations" that are, one realizes, actually supposed to be part of an on-going dance and regulatory feedback loop. Hospitals and doctors and the Centers for Medicare and Medicaid do this dance. The fact that there is a dance is the constant, and it's a very specific kind of dance that Control System Engineers know very well, and have tools to deal with.

I've gone on at great length in my recent MPH Capstone presentation at Johns Hopkins about these loops, and the fact that a closed, regulatory feedback loop is a qualitatively different animal than a "network" of interactions or a "web" of interactions or some Systems Dynamics Causal Loop Diagram. The loop has several properties that are very distinct from what is sloppily called "feedback":

• The loop persists over time.
• The loop is self-aware and can tell "itself" from "other" and do repairs, that is it has some kind of rudimentary immune system and damage repair system.
• To accomplish stability in a changing world, and overcome the problems Godel pointed out, the loop probably has formed alliances with other loops, either horizontally (peer loops) or vertically (management loops) where "you tell me if I'm losing it without realizing it and I'll do the same for you."
• The goal of the regulatory process has to include stability and survival, yes, but on top of that there is some goal, or direction, or "intent", and that goal may, in fact, be set by other loops (management) or peers (norms) or the surrounding context (tissue), or distant context (the pituitary gland).

Such loops will consequently differentially survive and persist, and end up dominating the landscape. They will tend to survive and persist best, if they are compatible with the larger scale activity that is going on, of course - if they are "aligned" with larger and larger contexts - but in Life this is a bidirectional game and the larger context goals may still be in flux and controllable by the lower level actors - until we get "phase lock" in that vertical dimension, the image finds a home and latches into place, and the whole entity is now stable on multiple levels at once. A solution has been found. Life, at a whole new level, has emerged.

Such loops are also exquisitely powerful in a computational sense. It's a long story, but closed feedback loops are IIR and open feedback loops are FIR, and IIR is infinitely more powerful than FIR. I'll get back to that tomorrow. In any case, no Control System Engineer would even imaging using an "open-loop" system if a "closed-loop regulatory feedback control system" were available. You get much better performance, stability, cost, robustness, response-time, tunability, etc. with closed-loop systems.

So, where's that get us? It's time to head off to my day job. Let's lash this down.

Of all the possible, N-factorial loops on causal loop diagrams, or in physical reality of the whole hierarchy of self-organizing Life, which includes people and corporations and nations, the winners will be the relatively persistent closed feedback loops. These will form, in many real senses, meta-life, or actors in their own rights with their own agendas and local worlds and local reference frames in which they are important and most of the rest of life is reduced to "an environment" in which they swim.

This "meta-life" will emerge at ever higher and higher organizational levels.

If you want to tweak, or tune, or improve the behavior and outcomes of any of these local entities (cells, the pancreas, the endocrine system, a person, a team, a corporation, the tobacco industy, the auto industry, a whole culture or nation or broken nation or proto-nation), then this model suggests focusing on finding a way to nudge it towards this stable and viable condition. "Nudge" means find existing proto-regulatory feedback loops that are almost closed, and close them. This pushes the whole "holon" (Ken Wilber's word) towards a stable attractor, a hold-fast, a sort of resonant notch or Lagrange point, where it can exist consistently with Life above and Life below it efficiently and effectively.

This is consistent with the focus on "microsystems" of the Institute of Medicine's Crossing the Quality Chasm, but goes beyond that and suggests a type of completion or stabilization of an actual meta-life process is the key variable to a sustainable intervention. It suggests what to look for, what to measure, and how to tweak it, and a fundamental principle of physics that supports that type of intervention as being principle-based.

It's also consistent with Senge's Presence and Stephen Covey's "8th habit" and focuses on the use of local feedback processes to bring about a change in a person or small team or an entire corporation that needs "restructuring" to return to an innovative, agile, productive state.

And it's consistent with my previous posts that Public Health, by its own logic, has to embrace improving the performance of corporations as well as populations of people, because both types of meta-life interact strongly, and, as many people in southeast Michigan are aware, if the corporate health is not well, the employee lives will not be well either. These do not need to be in conflict, and should be totally consistent objectives from both sides.

Neat. We should try simulating this and see how it behaves. We need to know how to measure such things, how to parametrize them, how to reverse engineer the underlying simple model from observable features, and how to nudge the various mathematical poles in the control-world towards the correct quadrant and tell (with p<0.01) whether our intervention is working or not - or, more precisely, we need to navigate and steer the intervention, in a minute-by-minute short-scale piloting fashion, around the rocks and to completion and need this compass to recover our direction after each swerve to avoid some obstacle or seize some favorable wind.