What I term "adapting systems" is a way of looking for organizational patterns which suggest principles of arrangement in the universe. Systems are sets of related objects, and the relations that bind them. Adaptive systems reconfigure themselves to the changes around them, which reconfigurations in turn trigger cascading ripples of adaptation in the surrounding systems. This is the "shifting [fitness] landscape" I alluded to in my title, above.

 

As General Motors, once a behemoth of American industry, shed jobs by the thousands over the last decade, it illustrated well the adage that those who can't adapt become nonviable. The June 2009 bankruptcy was inevitable: eventually, if GM could not produce enough value (cars, trucks, vans), then its obligations (pensions, wages, rents, fees) would become greater than its capacity to meet them. Size doesn't matter. It doesn't matter how big you are, how many battleships and aircraft carriers you have, how much gold bullion you have stacked in the vault: if you don't adapt you will become void. The Tyrannosaurus Rex may once have been the biggest and toughest creature on this planet, but the planet changed (think: Chicxulub) and the tree shrew adapted better to the changed biosphere. Ergo, Homo Sapiens. None of us should be so arrogant to think that we don't need to change.

 

How do collective entities respond to the changing environment, maintaining viability? Only systems that adapt successfully will remain. Systems here means humans (as biotic systems) as well as man-made, artificial systems (robots, corporations), and also abiotic physical systems (weather, climate, tides, solar system), and non-physical systems such as ideologies (idea systems).

 

I am interested in the process of change as it manifests itself through different systems. One, which simple sets of rules can engender change over the long run, and yet maintain systemic continuity? And two, what are the risks, and what are the payoffs, in this process of adaptive change?

 

Regarding the first question, the system's rules must be simple enough for everyone in the system to remember. There can't be "15 simple rules", there should only be a few. Too many rules and the homeostatic balance becomes cumbersome and leads to stasis and eventual insolvency. My sense is that there should be one guiding principle, one meta-rule, "one ring to rule them all": simultaneously manifest in two aspects: one for internal coherence and another one for external integration into the larger system (remember we are dealing with hierarchical [nested] systems here).

 

Each internal/external aspect would in turn have a set of subsidiary, reinforcing rules, which in turn guide other sets of rules. Then you can address complexity and change without upsetting the raison d'etre.

 

Regarding the second question, an obvious risk of nonadaption is that one can quickly become nonviable. So change may well be seen as imperative, given the changing environment (think, the Red Queen Hypothesis here). Those who sense the blowing winds of change around them, and respond appropriately, might retain viability a bit longer. So, which kinds of response heuristics are better, and which are worse? How might one do cost/benefit analysis in evaluating different "rules" (response patterns)?

 

Conversely, stability and continuity in the rule search (heuristic) remain crucial. The historical pattern of previous events (environments and associated response patterns) have gotten us to this point, and therefore have some meaning. Trying to do some blank-slate "It the Year Zero" nonsense is like jumping off a cliff; don't jump unless you really, really have to. Instead, recognize the value of continuity, and try to change in manageable increments.

 

But insufficient change will also lead to systemic failure, as will standing pat. If an agent just changes the bare minimum, and then says, "I'm different now", and folds its hands and rests, it will probably become nonviable shortly. Maybe tomorrow instead of today, but finished nonetheless. Rather, the agent should endeavor to transform as much as is practical, and necessary, without losing contintuity with the process that brought them there. The goal might be to become an outlier, near the tip of the bell curve, but not falling off the cliff. The ones on the tip of the bell curve will probably get the greatest reward/benefit.

 

The trick seems to be to glean enough from the environmental cues: "You're at the fat part of the niche here; don't move. Let the center come to you." And, conversely, one perhaps may get cues from the surrounding meta-system: "Lean times ahead, buddy. Better move, and move fast -- go over that way."

 

For instance, look at the so-called "horsetail", the little herb equistum arvense. It is a rather primitive plant which has remained largely the same over millions of years. We might say that it found a nice niche and stayed put. Yet it was able to change over time: it once grew to 60 feet tall, and now it tops off at 2 feet. The dinosaurs came and went, and little equistum is still viable. So I like the idea of finding a "fat" niche, exploiting its potential, and maintaining some kind of equilibrium there. It seems as if luck, or chance, i.e. a favorable roll of the dice is indeed helpful. But exploiting the outcome, once the dice are at rest, is where our decision-making process becomes salient.

 

I call this "Agent heuristics in multivariate settings".

 Last modified 1/28/12