← Back to Archive

Why feedback appears everywhere once you have seen it

The same loop, recognised outside engineering.

Feedback appears everywhere once it has been properly seen because the pattern behind it, an output looping back around to influence its own cause, is not a piece of engineering machinery at all, it is a shape that any system made of interacting parts can fall into, whether those parts are thermostats and heaters or habits and moods, and once a reader has genuinely understood the shape in one setting it becomes almost impossible not to notice the same shape recurring in settings that have nothing to do with engineering.

Meadows' Thinking in Systems is nominally a book about feedback loops, stocks and flows, and delays, but the effect of actually reading it is stranger than that description suggests, since the same simple diagram it teaches, an arrow looping back from an effect to one of its own causes, keeps reappearing afterward in places that have nothing to do with engineering at all, a habit, a conversation, an argument that keeps escalating for reasons neither person quite intended.

Introduction and overview

A feedback loop, at its most general, is simply any arrangement where an output of a process becomes, sooner or later, an input back into that same process, closing a circle rather than running in a straight line from cause to effect and stopping there. Engineering happens to be full of clean, deliberately designed examples of this shape, a thermostat's measured temperature looping back to control the very heater producing that temperature, but the shape itself was never actually invented by engineers, it was simply noticed, formalised and put to deliberate use by them. The same closed loop shows up just as reliably in systems nobody designed on purpose, a population of animals whose numbers affect how much food remains, which in turn affects how many animals survive to reproduce, or a habit that becomes easier to keep the more consistently it is kept, each outcome quietly becoming one of the inputs shaping what happens next.

The messy-desk comparison

A desk that has been allowed to get properly cluttered demonstrates a feedback loop with no engineer, sensor or diagram anywhere in sight. The more cluttered the desk becomes, the more daunting it feels to start tidying it, since a small mess invites a quick five-minute fix while a large one demands a proper, dedicated effort that is easy to keep postponing, and every day the tidying is postponed adds a little more clutter, making the next day's version of the same task feel even more daunting than the last. The clutter is both the outcome of the avoidance and, looping back around, a cause of the next round of avoidance, precisely the same closed circle a thermostat runs, just built from procrastination and mess rather than from temperature and a heating element. Recognising the loop suggests the fix directly, breaking the circle by making the very next tidying pass small and easy enough that its own outcome, a slightly tidier desk, stops feeding the avoidance rather than reinforcing it.

Why some loops grow and others settle

Not every feedback loop behaves the way the desk or the thermostat do, and the difference matters. A loop where the output works to correct or reduce the very thing driving it, a thermostat's heater switching off as the room warms, a tidier desk making tidying feel easier rather than harder, tends to settle toward some steady state rather than spiralling away from it. A loop where the output instead reinforces and strengthens the thing driving it, the messy desk feeding its own avoidance, an argument where each sharp reply provokes an even sharper one in return, tends to grow instead, sometimes slowly and sometimes explosively, until something outside the loop finally interrupts it. Both kinds are feedback in the fullest sense of the word, and telling them apart is usually the single most useful thing understanding the shape actually buys a reader trying to change how one of these systems behaves.

The one number worth remembering

A habit reinforced by even a small feedback loop, each repetition making the next slightly easier or slightly more rewarding, can become dramatically more entrenched after only a few dozen repetitions than the same habit attempted without any such loop supporting it, since the loop is not simply repeating the same small effort each time, it is compounding a slightly reduced effort onto the last slightly reduced effort, a pattern of accumulation that a single glance at any one repetition in isolation would never reveal.

Where the same loop hides in a supply chain

A well-documented example of a reinforcing loop growing rather than settling is the bullwhip effect, in which a small, ordinary fluctuation in customer demand at the retail end of a supply chain gets amplified at every step further back, since each supplier along the chain reacts to what the step in front of it has just ordered rather than to the real underlying demand, and each of those reactions becomes, in turn, an exaggerated signal for the supplier one step further back still to react to. A shopper buying a few more tins of something than usual one week can end up, several links back along the chain, causing a factory to swing its production by a proportion vastly larger than anything the original shoppers actually did, purely because each link in the chain was feeding its own reaction back into the next link's decision rather than ever seeing the original, much smaller signal directly.

What follows from this

Once feedback is recognised as a general shape rather than a piece of engineering vocabulary, a genuinely useful habit of mind follows automatically, looking for the loop behind a pattern that otherwise seems to have simply appeared or worsened on its own, a system's parts affecting one another in a circle rather than a straight line running quietly out of sight. The rest of this set follows that same habit of mind directly into problems that look, on the surface, nothing like the thermostats and control loops covered earlier, and finds the identical shape waiting underneath every one of them.

More on Systems