Equilibrium Part 2 - Heuristics for Intervention

What are some principles for generating upward momentum (or avoiding downward spirals)?

Note: This is Part 2 of a three-part series. You can read Part 1 by following the embed below.

Equilibrium Part 1 - Virtuous and Vicious Cycles in Ecology

At the outset I said that I loved ecological analogies to other complex systems. One such complex system is cities. Thinking of a city as an ecosystem has helped me understand what cities are and how they function under the surface. Not only can natural complex systems help us understand cities, but natural and cultural systems are bidirectional analogies. We can use cities to help us understand ecosystem dynamics and use ecosystems to teach us what makes cities function and dysfunction.

The difficulty with understanding these systems is not so much that it is impossible to figure out what is going on. The real difficulty is to model what would happen if we changed things, to understand which interventions work and which hurt. I want to propose here a set of heuristics that could be used to test potential interventions. They are heuristics because most of us do not have the luxury of time or resources needed to experiment. Indeed, with complex systems, modeling may be practically impossible and inadvisable, although I do not want to foreclose that possibility entirely. But we could be led more often rightly than astray by a minimal set of guiding principles that stand in advance of experimental results, rather than relying on the results of simulated models that are incomplete by definition. We could maximize upsides and minimize downsides by carefully choosing our interventions. And we could avoid unintended consequences by keeping it simple and staying humble.

1. Observe Carefully

When I read about a common staple item like lima beans, I marvel at (and am grateful for) the poor experimenters who have gone before me, suffering gastrointestinal distress, illness, and even death to find that these lima beans are toxic unless cooked sufficiently. I sometimes wonder how much of the discoveries of plants with medicinal (or toxic) qualities originated from the essentially random process of eating everything to see what would happen and how much originated from careful observation. We move too quickly these days, as a rule, to observe with enough detail to provoke wonder and it’s hard-nosed twin, the experimental hypothesis. But attending to the complex systems and the web of interdependencies we inhabit teaches much and suggests room for intervention.

Passing a street corner at a high rate of speed in a car affords far fewer opportunities for observation, understanding, and intervention than walking, running, or riding a bike. The next time you have an opportunity, walk a route you have previously driven and notice how much more of that place you can observe. When repeated many times, one comes to notice the same people, the same places, the growth of flowers and passing of seasons, the mail carrier and the children getting off the bus from school, when one's eyes and mind are not focused so much on piloting a large, two-ton machine at a high rate of speed. 

Observations can also always be generative, especially when one has the capacity to act immediately upon them. It's much easier to pick up a piece of windblown trash when walking down a street, making some small gesture of help, than to even consider doing the same from the comfort of your own vehicle. Close and repeated observation also helps to spike the majority of wrongheaded interventions from the immediate outset. If you are deeply aware of the pace and tenor of a neighborhood by presence of proximity, your intuition about the likely effect of a change to that neighborhood will be better-informed.

I want to caution here that, because complex systems are at many scales, the parochial intuition about a local intervention may run counter to the intuition one would develop when observing it and adjacent localities, or dynamics at higher scales or even lower scales, say the dynamic within a family living in that neighborhood. Close observation should indeed lead you to notice the interdependencies between the node in the network you occupy and its neighbors.

2. Disturb Tactically

In part one I used the word perturb to describe an intervention to an ecosystem that could send it in a positive, upward spiral. We tend to use perturb in negative contexts, but destruction is often necessary. As a child I had surgery to correct a hollow pocket of blood and air that had developed in my jawbone. The prescribed intervention? Drain the pocket, scrape (literally, damage) the inner wall of the hollow, and close it back up. The destructive act of scraping the bone inside the pocket induced it to regrow and fill in the pocket with new bone. Sometimes, and usually in economic contexts, this perturbation is described as 'creative destruction' by the more eager disturbers among us. But intervention can often be a salutary shock, an awakening against the enervating lull of a well-known cycle. It can be generative only as it is also really destructive of something that came before – something that has, likely as not, been successful in its own right and on its own time.

But at risk of overusing this repeated analogy of a spiral (upward or downward), it is important not to disturb so violently that the momentum and centrifugal force of the spiral is arrested. A system will need that momentum to get it around the next swing. I originally considered writing that one should disturb minimally, but I don't think that is quite right. To require interventions be minimal may immobilize or dilute the work needed. It could be better to say that one should not intervene any more than is required.

But required for what? That is why I prefer to say that one should disturb tactically. The term ‘tactics’ strikes the right balance in my mind, and I hope in yours, because tactics can often be quite disturbing, but they do not have the same universal ambitions or broad requirements of a strategy. They are situational, small enough in scale and ambition that they leave open possibilities, and also do not preclude additional, simultaneous tactics that run other experiments or attempt incursions at other fronts. On the other hand, you can have only one strategy.

3. Keep Your Options Open

Indeed, some kinds of decisions are final: they foreclose other possibilities. Other kinds of decisions do the opposite. They form the building block for many different kinds of future actions. A characteristic of industrialized agriculture is that the decision to pick a particular cash crop deliberately eliminates all other possible land uses. To the industrial farmer, polycultural forays are anathema. Planting multiple kinds of crops, using the inevitably variable topography of the land to plant drier and wetter species, planting with multiple timescales and for multiple purposes, planting species for which no immediately productive use can be made, planting in spatial patterns inimical to mechanical harvesting, are all headaches. In the words of James C. Scott, the fingerprints of whose book Seeing Like The State are all over this essay, polycultural planting makes the field less legible to outside command and control interfaces. These interventions get in the way. In the way of what? Yield. But the row- and mono-cropping farmer has made a decision that forecloses for them all other possibilities. If short-term, season-by-season yield maximization is truly the outcome for which a farmer is optimizing, then polyculture is a nightmare. It makes mechanical harvesting impossible. It mis-aligns the seasonality of picking labor from the need for hands on the land. It also just doesn't look neat and tidy: it doesn't look like big-A Agriculture. It violates our aesthetic expectations, which are far too often determinative for us.

So too do polycultural cities make command and control difficult. The provision of centralized services and the exertion of centralized control are much more difficult in a city with variable density, an irregular road network, and a thick bed of informal cultural and economic practices and relationships. If the central planner's decision is to optimize for the efficient allocation of resources, then it is better to bulldoze the slums (read: anything with the external appearance of untidy or unplanned construction), build clean repetitious grids of orderly buildings zoned in rational, single-use districts, and so provide a flattened field in which to distribute water and power and police and schools and convey wastewater and solid waste. Incidentally, this mode of planning makes the city legible and explicit; it lays it bare for the central gaze. So too does monocultural farming. It's easier to spot a weed when everything that doesn't look like your cash crop is unwanted.

There are cautionary tales aplenty in the story of industrial agriculture and central city planning. The untold human devastation wrought in the heady days of urban renewal, suburbanization, and mass road construction has so broken the psyche of the city planners of today that it has in many places essentially suffocated even genuine and 'tactical' interventions to the planning of the built environment. Likewise, the quiet but desperate war against ever-strengthening pests and ever-more-devastating blights in industrial agriculture is a cautionary tale that big decisions that foreclose other possibilities leave one open to a single point of catastrophic failure.

That is the crux of the problem with decisions that have yield, or vehicle conveyance maximization, as their only criterion for success: that it leaves systems vulnerable at a single point of failure. Single-mode transportation fails when one accident shuts down a freeway. A monocultural farm (or industry, as Florida's orange growers have learned) is absurdly vulnerable to pests. By so attenuating our concerns and our goals, we have undercut the very informal and distributed secondary systems that could sustain or at least mitigate the damages to an orange grove through a blight or a freeway through an accident. Therefore, decisions to intervene in these complex systems should in the aggregate add to, not eliminate, the optionality already there.

There is a profound humility about this approach that I think is worth elevating in its own right. We cannot know all ends when at a beginning. The arrogant start to a project would be one which requires for success a single outcome. The humble beginning recognizes that there are enough unknowns in the future that while a single end can be hoped for or hypothesized about, it cannot be relied upon. Optionality builds up resilience for a system; it makes it possible to fail a hundred times and still ultimately succeed.

There are very limited realms within which single-criterion interventions can be successful. To use one final, and tangential, example, venture capital firms are famously willing to take catastrophic loss and failure again and again while searching for the one wildly successful, high-growth company that will wipe out their losses and provide a return to their investors. This is, of course, a successful model within the narrow slice of the economy it operates in. But imagine if those same economic dynamics of feast or famine were applied to the economy at large. It is precisely the distributed, risk-averse, impossibly complex economic system that venture capitalists operate within that minimize the downside of their experimentation and the churn it creates. Not many people are risk-tolerant enough to bank with, buy groceries from, invest in, or take jobs from startups that could be gone the next month. Nor should they be. It is the exception that proves the rule.

4. Take Care of the Smallest Element

A familiar frustration for designers and planners of all stripes is the misuse (in their eyes) of their new creation. What has happened is that as they have planned or designed a city, a product, or a system, they have implicitly or explicitly modeled the ways that a very complex system (a person, or a community) will use and respond to an intervention. And they see it as a failure when that model is violated. This is of course another vote for optionality -- plan and design for misuse! -- but it is also a point in favor of paying careful attention and taking care of the smallest divisible unit in the complex system. In cities, it is the individual. Making interventions to a city that give an individual the autonomy, the tools, the resources, or the opportunities to be successful at the smallest scale trickles upward toward the broader goals for a plan. Another way of saying this is that leaving optionality open for an individual lets them use your work to optimize locally.

This concept requires us trusting, in the case of cities, that individuals tend to understand 'conditions on the ground' -- their own and their immediate network's needs and goals -- better than we do and that they will be relatively rational in their use, or tactical misuse, of your interventions to achieve those goals. Creating the conditions for success, rather than trying to manufacture the success ex nihilo, entails risk, as all trust does, but counterintuitively may involve less ultimate risk of the failure of the intervention as a whole.

Much of polyculture involves building and taking care of the soil. The microorganisms, decaying matter, insects, and worms that build soil are not flashy; they're squishy and dirty and smell bad and we usually cannot see them or readily see the work they do. But attending to them and giving them conditions wherein they can be successful literally lays the groundwork for the higher-order species who rely on them.

Ideally, this investment at the smallest divisible unit yields the accumulation of many small upward motions in our analogical spiral. The theory is not that these are sufficient to the task alone but that these create the conditions for larger and more coordinated upward motions in the future, ones which may arise organically or with minimal intervention. The virtuous cycle, the spinning up, from a lower-order equilibrium to a higher, may require only that the conditions be right. So many of these flourishings in human and natural culture seem magical, or spontaneous. The unaccountable success of a city without the right geographic or historical conditions; the profusion of a micro-environment within a vacant lot, may not be as spontaneous as they appear. They may be the sum total of good conditions matched by the teeming motions of many small elements.

5. Don't Jump Ahead

We tend to look at natural or cultural ecologies and wish for them the highest-order equilibrium possible without delay. The profusion of mass tree-planting competitions or initiatives worldwide is laudable in some sense but reveals pathologies of impatience and misunderstanding in us and the organizers. Forests of trees arise from and thrive in a complex set of conditions that take time to set down. A forest can be thought of as the culmination of many smaller steps taken over many years, not a thing you can get at once by planting hundreds of thousands of seedlings in the ground. Efforts like this try to construct the end point of the desired outcome from a complex system, but do not spend the time cultivating the conditions required. This advice is really about the dynamics of equilibrium. A forest requires an extraordinary cyclic momentum to survive in the long term. Without that momentum, an intensive perturbation like a mass planting will often settle back down to a natural equilibrium, leaving a lot of dead saplings in its wake.

6. Sometimes, Do Nothing

If the conditions are already right and the trend is already upward, just leave it be. I've heard this tendency described in relation to hand-wringing about vacant lots. "Let's just plant some trees there, fix it up!" If a place was until recently a forest, it very well may return to it over time as long as it isn't disturbed. A vacant lot that is regularly mown to keep up its aesthetic appearances never will; the unruly, 'weedy' appearance is really a profusion of competing pioneer species attempting to restart the upward equilibrium of the successional forest it used to be. We lack the patience for 250-year or even 50-year timescales, urgent as we are to deliver aesthetic results for our vacant lots and for their neighbors. But if we can bear a vacant lot that looks like an actual meadow for a time, we may in turn find the lot surprising us once the conditions have been made right again.

On the other hand, not every lower-order equilibrium can be jump-started into a higher-order equilibrium. A desert may very well 'want' to remain a desert -- perhaps it only has the environmental conditions suitable to an ecosystem less dense with biota but no less productive within its own scope. In this context, doing nothing would yield nothing different, but would not waste effort on a system already operating at what is truly its local maximum.

The nice thing about writing heuristics is that I can fall back on the definition of the word to forestall any criticism of them. A heuristic is imperfect and lossy; it elides details or edge cases with a wave of the hand. This is as far as most of us can go when it comes to studying, understanding, and proposing interventions for complex systems. As often as not, heuristics like this can help us point out where existing interventions have gone wrong, or help us quickly judge the likely results of a proposed intervention. But they do not contain in themselves good ideas for intervention. Ingenuity, humility, and careful attention are only some of the characteristics of a successful cultivator of natural or cultural systems. It is easier to be a critic than to cultivate.