Systems

In spite of what you majored in, or what the textbooks say, or what you think you're an expert at, follow a system wherever it leads. It will
be sure to lead across traditional disciplinary lines.

-Donella H. Meadows

Feedback Loops

Listen and incorporate.

Feedback loops are everywhere in systems, making them a useful mental model.

Think of feedback as the information communicated in response to an action. Whether we realize it or not, we give and receive different forms of feedback every day. Sometimes feedback is more formal, as is the case with performance reviews. Other times, it is less so. Our body language is a form of feedback for people interacting with us. The tone you use with your kids is feedback for them.

A feedback loop is a process in which the output of a system also acts as an input to the system, helping to refine and improve the system over time. It's like a conversation in which each reply helps to shape the next question and answer, making the discussion better and more focused.

Once you start looking for feedback loops, you see them all over the place, giving you insight into why people and systems react the way they do. For example, much of human behavior is driven by incentives. We want to take actions that lead to us getting something good or avoiding something bad on a range of timescales. The incentives we create for ourselves and other people are a form of feedback, leading to loops that reinforce or discourage certain behaviors. If you get visibly upset whenever someone at work offers you constructive criticism, you'll incentivize your colleagues to only tell you when you're doing something well-thereby missing out on chances to improve.

The challenge in using this mental model is that the ubiquity of feedback loops can become overwhelming. How do you know which ones to pay attention to? Or which ones to adjust to improve your outcomes?

We are constantly offering feedback to others about our feelings, preferences, and values. Others, in turn, communicate feedback on the same things, but we don't necessarily receive it or interpret it correctly. A critical requirement is learning how to filter feedback. Not all of it is useful. The quicker you learn to identify feedback that helps you progress, the faster you will move toward what you want to achieve.

Learning to communicate feedback in a way that makes it easy for others to receive is a valuable skill to develop. Feedback is crucial to relationships of all kinds. The skill is in giving feedback in a kind and clear way, in hearing it without getting defensive, and in partnering with people who can receive it on a regular basis. If you feel you can't communicate important thoughts and feelings, you won't be happy in that relationship or situation long-term. Much of relationship therapy consists of a therapist helping a couple tell each other things they have been afraid to say-when that feedback should have been free-flowing for years.

There is a larger implication here about working with the world. The world offers us feedback, but do we listen and incorporate, or do we just keep wanting it to work differently than it does?

The technical definition of a feedback loop comes from systems theory.

A feedback loop is a process in which the outputs (information) of a system affect that system's behaviors. Depending on the complexity of a system, there may be a single source of feedback or multiple, possibly interconnected sources. It helps to first consider feedback in a simple system as we do in the following, but keep in mind, we are part of many large systems that contain many interconnected feedback loops.

Feedback loops are a critical model because they are a part of your life whether you are aware of them or not. Understanding how they work helps you be more flexible with the variety of feedback you receive and incorporate, and then you can offer better feedback to others.

There are two basic types of feedback loops: balancing and reinforcing, which are also called negative and positive. Balancing feedback loops tend toward equilibrium. Your thermostat and heating system run on a balancing feedback loop. Information about the temperature of the house is communicated to the thermostat, which then adjusts the output of the furnace to maintain your desired temperature.

Reinforcing feedback loops amplify a particular process. They don't counter change, like your thermostat does. Instead, they keep the change going, as with the popularity of trends in fashion (in which styles become ubiquitous within months, only to disappear soon after) or the loops usually involved in poverty (in which different but related circumstances can compound the problem). Breaking out of reinforcing loops often requires outside intervention or a new change in conditions. Or, as with fashion, they simply burn themselves out after a while.

Within complex systems, feedback is rarely immediate. It can take a long time for changes in flows to have a measurable impact on how the system works. This delay complicates establishing cause and effect. In our lives, problems arise when the feedback for our actions is delayed or indirect, as is often the case.

A challenge to improving our decision making is getting accurate feedback on decisions. On one hand, consequences may take a long time to become apparent or may be hard to directly attribute to a particular decision. On the other, we may trap ourselves in maladaptive behaviors when something we do receives positive short-term feedback but has negative long-term consequences. Thus, it's important to remember immediate feedback isn't the only feedback. When you eat junk food, there is an instant hit of pleasure as your body responds to fat and sugar. After a little while, though, you receive other feedback from your body that indicates your choice of junk food has negative consequences. And over longer periods, conditions such as type 2 diabetes and high blood pressure provide more feedback from your body about the effects of your eating habits.

The faster you get useful feedback, the more quickly you can iterate to improve. Yet feedback can cause problems if it's too fast and too strong, as the system can surge. It's like when you press the gas or brake too heavily when first learning how to drive.

Feedback loops are a useful mental model because all systems have them, and we operate in a world of systems and subsystems.

Adam Smith and the Feedback Loop
of Reactions

You probably know Adam Smith as one of the most influential economists of all time, notable for his notion of the "invisible hand" of the market. But Smith's first book, The Theory of Moral Sentiments, is a work of philosophy. In it, he describes a different sort of invisible force that guides us: how the approval and disapproval of others, real or imagined, influences our behavior.

We are, by nature, selfish. We value ourselves above all other humans. Smith illustrates his point by suggesting that the news that your little finger must be amputated would likely be more stressing than the news of the deaths of a huge number of strangers overseas. Yet despite our inherent selfishness, the majority of people the majority of the time are cooperative and kind to one another. Smith believed our interactions with others are responsible for our well-established reciprocity. He saw others' responses to our behavior as feedback guiding how we act in the future. To do something selfish usually warrants a disapproving reaction. To do something selfless usually merits an approving one.

The feedback loop of others' reactions to our actions is the basis of civilization. Russ Roberts writes in How Adam Smith Can Change Your Life, "Smith's vision of what sustains civilization is the stream of approval and disapproval we all provide when we respond to the conduct of those around us. That stream of approval and disapproval creates feedback loops to encourage good behavior and discourage bad behavior."

This type of feedback doesn't just cover formal punishments and prohibitions according to the law where we live. It also covers the ways people respond to behavior that is just considered the norm. If a friend says hello to you on the street and you fail to acknowledge them, you haven't broken any laws, but they're likely to respond in a negative fashion. So you adhere to the norm. Smith writes: "When Nature formed man for society, she endowed him with a basic desire to please his brethren and a basic aversion to offending them. She taught him to feel pleasure in their favorable regard and pain in their unfavorable regard. She made their approval most flattering and most agreeable to him for its own sake, and their disapproval most humiliating and most offensive."

Smith asked readers to imagine a person who grows to adulthood without any interactions with other people. He believed such a person would have no awareness of their character and no notion of the right or wrong way to act. Our desire to be loved and accepted prompts moral behavior relative to the standards of our society. We in turn respond with approval of the same behavior on the part of others.

Everyday Loops

We can address many of the challenges we face every day by adjusting feedback loops. Figuring out how to change behavior (ours and others'), dealing with inaccurate information, and building trust are ongoing challenges. How to get customers to buy your product and not the competitors'; how to sort through information to find what's relevant to your decision; how to cooperate effectively with others: these are common situations many of us face.

All these dynamics play out on the larger social scale as well. How do societies incentivize the behavior they want and disincentivize the behavior they don't want? How do they get people to trust one another enough to keep society functioning?

Any system with an unchecked reinforcing feedback loop is ultimately unsustainable and destructive. Balancing feedback loops are more common in systems because they are sustainable-they maintain a system long-term, whereas reinforcing loops can crash and burn. In many societies, a legal system has historically served to stop reinforcing feedback loops from crumbling the social infrastructure and to promote balancing feedback loops that support desired dynamics. How they do it suggests options for addressing similar issues with feedback loops in our own lives.

Let's explore four aspects of social systems through the lens of feedback loops:

1. Creating the right future incentives

2. Influencing behavior at the margins

3. Dealing with information cascades

4. Building trust

Creating the Right Future Incentives

We want to minimize, as much as possible, making a choice today that creates a negative reinforcing feedback loop down the road. So, we consider the future incentives a decision will create.

A classic example of today's solution inadvertently creating a reinforcing feedback loop of future problems is paying off kidnappers. The immediate problem is someone being kidnapped and a ransom being demanded. If you have the resources to meet the kidnappers' demands, you might want to pay the money right away. You save a life and solve the problem.

However, your response communicates to the kidnappers that you will meet their demands. You thus create an incentive for them to kidnap again, as well as signal to other would-be kidnappers that there is money to be made. By paying a ransom, you create a powerful reinforcing feedback loop that causes more problems in the future.

In many legal systems, each decision by a court becomes a bit of information that moves via a feedback loop into the stock of legal options to influence both how the system responds to future cases and how judges form future decisions. Each decision becomes legal precedent. In The Legal Analyst, Ward Farnsworth explains that in making decisions courts will consider "what incentives people will have after the case is over." Courts need to be careful. If they compensate for a wrong now, they could create a climate that increases the chances of that wrong happening again.

One set of issues courts often face is questions of liability. If something bad happens to me, then does someone else need to pay to compensate? Sometimes the answer is yes, but not indiscriminately so. If we go back to the kidnapping example; let's say an American citizen is kidnapped overseas and the kidnappers demand the government pay for release. Is the government liable to compensate the victim's family for the loss of life if they choose not to pay the ransom? Most courts will answer no. If I am held liable, it incentivizes me to pay in the future, and we are back in the same reinforcing feedback loop. When considering certain instances of liability, Farnsworth explains, "Instead of looking back and deciding who should bear the suffering, [a court] can look ahead and decide what ruling will make the suffering less likely to occur later."

In some situations, choosing an immediate benefit creates reinforcing feedback loops that remove the possibility of future benefits. For cases like these, laws are designed to support balancing feedback loops, such as protecting attorney-client privilege, or copyright and patent laws. Although one could argue for the immediate benefit of, say, forcing a defense attorney to testify about what their client disclosed, the feedback loop created would make clients disinclined to disclose things to their attorney. As Farnsworth summarizes concerning copyright protection, "Once books and music exist . . . there's a great case for free distribution of them. But then they are less likely to exist at all next time."

People look around and often see what they view as unfairness-but they don't realize that unfairness sometimes has a greater purpose. Unfairness in specific cases creates fairness on the whole, as in the aforementioned cases. Often things happen that look like an injustice to the individual, and they may be so. But those things may create greater justice for the collective. Think of someone being "excessively" punished; that may seem unfair to them, but if it is successful in deterring others from committing the same infractions, it's not always such a bad idea.

Influencing Behavior at the Margins

Not everyone is likely to change their behavior in response to pressures, such as social or economic changes, at the same intensity and rate. Some people need more convincing; others need more time.

A good customer retention strategy doesn't lump all customers into one group. It might, for example, treat customers of ten years better than customers of six weeks. These are examples of margins. "Margins" can mean a lot of things in economics; in this case we mean "by increments," or "small changes"-shades of gray instead of all or nothing. Farnsworth explains thinking at the margin as "looking at problems not in a total, all-or-nothing way . . . but in incremental terms: seeing behavior as a bunch of choices about when to do a little less along one dimension and a little more along another."