The quality of feedback loops

A primer on the cybernetic fundamentals

Feedback is a key (perhaps the key) notion of a branch of mathematics called cybernetics. Norbert Wiener invented cybernetics and introduced it to the world in 1948. He coined the term from the Greek word meaning “steersman.” Wiener defined “cybernetics” as the science of communication and control in the animal and the machine. The field has been broadly developed and applied to such diverse areas as engineering (control systems), communication, information theory, general systems theory, cryptography, epistemology, and management, among others. It offers insight and utility in understanding and improving the performance of all systems. It is often very misunderstood. Isn’t it just the comments you get back from people?

Here are the conditions that define feedback.

First, someone who can use the information as feedback must have a goal. Without a goal, there’s no way to figure out whether this person’s subsequent behavior treats the information as feedback or not.

Second, the information received by this person must allow him/her to relate the effect of his/her actions to the goal. Otherwise, the information is useless as feedback.

Third, the person receiving the information must choose to construe this information as feedback—to interpret it as a report of how well the action succeeds in approaching the goal. If the person doesn’t want to construe the information as feedback, it remains merely information. Finally, this interpretation must influence the person’s subsequent actions in attempting to reach the goal. Unless the construed information helps the person move toward achieving the goal, the information is not being used as feedback at all.

Herein lies the common misunderstanding of “positive” and “negative” feedback. Negative feedback isn’t necessary bad; it just means information that is construed in such a way that it results in a change in direction in the pursuit of some goal. Inversely, positive feedback isn’t necessarily good; it just means information that is construed in such a way that it results in a continuance of the same course. In nature, positive feedback loops are often responsible for a state called “runaway,” unchecked behavior that can cause a disaster. In general, negative feedback leads to stabilization or homeostasis, while positive feedback can (but needn’t always) lead to runaway.

In sum, the four conditions are these: (1) a goal, (2) a way of relating the effect of actions to the goal, (3) choosing to construe the information as feedback, and (4) influential on behavior.

Quality Criteria for Feedback Loops

1. Right recipients: gets to the people who need it.
2. Willing recipients: recipients have to want to construe the information as feedback. They won’t if they feel offended by the information. Therefore, the information must be inoffensive.
3. Actionable: formatted in a way that is easy to understand how it relates to the goal’s criteria for success; if it is diverging from the goal, the feedback should have sufficient detail to allow swift identification of the problem to allow solutions to be easily identified.
4. Timely: the value of the information diminishes with elapsed time, because the recipients have moved farther down what may be the wrong road. The sooner they receive the information, the sooner they can correct their course. Untimely feedback loops are called “sloppy.” Fixing them is called “tightening” the feedback loop.
5. Fail-safe: a common problem is to permit feedback loops to rely solely on someone’s memory. When they forget (as often happens), the loop is broken. It is better to set up some precautions, reminders, or automation to ensure that feedback occurs.