feedback loop

If you see a behavior that persists over time, there is likely a mechanism creating that consistent behavior. (Donella Meadows, Thinking In Systems)

Feedback is when the effects of a system return as inputs to the system itself. Donella Meadows defines it in terms of stocks and flows:

A feedback loop is a closed chain of causal connections from a stock, through a set of decisions or rules or physical laws or actions that are dependent on the level of the stock, and back again through a flow to change the stock.

systems thinking describes two kinds of feedback loops:

• Reinforcing (or “positive”) feedback amplifies the output of a system, resulting in growth or decline. For example, your bank balance and interest accrued: as one increases, the other increases as well.

• Balancing (or “negative”) feedback dampens the output of a system, stabilizing the system around an equilibrium point (a state known as homeostasis).

Balancing feedback loops are much more common than reinforcing ones because they are more sustainable, but they are also harder to see: because they produce stability, their effects are easy to take for granted. Still, everything in the world that isn’t spiraling out of control is the way it is thanks to balancing feedback loops.

In systems with multiple feedback loops, one loop is often dominant, meaning it’s behavior governs the behavior of the system. When a reinforcing feedback loop is opposed by a balancing feedback loop, the system remains in equilibrium so long as the balancing feedback loop is dominant. If it is weakened or the reinforcing feedback loop strengthened, the system can switch—surprisingly in some cases—into unchecked growth.

Reinforcing feedback loops can be driven both ways

Reinforcing feedback loops are powerful when used to your advantage, but carry the inherent risk that their exponential growth can turn into exponential decline. Consider a growing company: they hire more employees which lets them better serve customers, which in turn earns them more profits with which to hire more employees: a classic reinforcing feedback loop. But if conditions turn against them, they will lose income, not be able to employ to many people, not be able to serve customers as well, losing them income: the same reinforcing feedback loop. (As an aside, this is why I prefer the term “reinforcing” feedback rather than “positive”: because it can lead to desirable outcomes just as easily as undesirable ones.) A strategy to counteract this is to hedge your bets and add balancing feedback to your system (such as setting aside savings and diversification of investments) which come at the cost of powering further growth, but likewise provide a buffer if that growth turns to decline.

A helpful example of this is what Elizabeth Sawin in this TEDTalk calls “multisolving”. When you have multiple problems compounding in a reinforcing feedback loop, that can be scary but it is also an opportunity to solve multiple problems at the same time.

Living Systems Have Unimaginably More Feedback Loops Than Artificial Ones

From a talk by John Gall (of Systemantics fame):

The amount of feedback that is built into living organisms differs by many orders of magnitude from the amount that we build into manmade systems. Living organisms such as human beings have feedback from almost every cell of the body all of the time.Every event that occurs either to or within a living creature has a feedback component—that is, every part of the rest of the organism knows of it in some way, usually within a fraction of a second.

There is not one living creature that lacks that integrated, total unification through feedback. Compared to a frog or a salamander, some of our biggest manmade systems seem like simplewind-up toys. On a supertanker, an alarm bell rings if the engines aren’t working, but doesn’t indicate where the malfunction is or what kind it is. It could be a loose wire—or it could be that an eighty-foot wave has just crushed in the bow of the ship, flooding the engine room.Either way you get that same little bell tinkling. Then there is that little idiot light on the dashboard of your car that blinks on and says, “Check engine.” A living creature organized like that wouldn’t last very long.

“Make a Friend of Feedback”

Gall, in the same talk:

Many systems don’t want feedback because typically their system is not designed to respond effectively to it. They may even regard the bearer of feedback as an enemy of the system. But ignoring feedback merely means that the system will eventually experience a massive unpleasant surprise rather than a small unpleasant surprise. So my advice is: Make a friend of feedback. Plan for it, welcome it, and be sure to make the necessary indicated changes in your system promptly, before the feedback gets really threatening.