Applying Feedback Techniques
Dan North talked about models and techniques for giving and receiving feedback and how to apply them effectively at the QCon London 2016 conference.
How feedback works
The term feedback is derived from systems theory. In dynamic systems some of the output reenters the system as input via feedback loops. Adaptive systems rely on feedback.
You can look at people as a system, and considers a group of people as a system of systems said North. For instance by trying to find levers which are feedback points where a small change creates a large impact.
In a system you can have delays. Delay in the feedback lead to uncertainty. This is one of the reasons why you strive to have small batch sizes in lean, which reduces delay and leads to less uncertainty and stress said North.
There are several kinds of feedback loops. One is reinforcing loops, where feedback either amplifies and accelerates the behavior or where it suppresses and diminishes the behavior.
North gave the example of the monkeys in a cage where there was a ladder with bananas on top of it. When a monkey tried to go to the bananas all monkeys were sprayed with water. This way the monkeys learned that they shouldn’t try to go to the bananas. When one of the monkeys was replaced by a new monkey and that monkey tried to go for the bananas, it was stopped by the other monkeys because they didn’t want to get wet. In time all monkeys were replaced, so none of them had ever experienced getting wet when they tried to go for the bananas. Still, none of them would do that. This is known as learned helplessness said North, where people won’t try something because they are afraid that it might go wrong, even though they never did it.
Another kind of feedback loop is stabilizing feedback. This is where feedback supports a system to stabilize and move towards a stable goal. And there is oscillating feedback where a system flips between states. Oscillation can happen when the feedback time is relatively short and the system is reacting to quick. The solution would then be to change the system so that it will pause before responding to feedback.
Small and frequent feedback is better than large and infrequent feedback said North. Delay in feedback affects a system’s ability to adapt. He gave the example of an annual performance review, where the feedback is large and the time is too long. Which such a system the vast amount of feedback is wasted.
Giving and Receiving Feedback
After explaining how feedback works, Dan North talked about giving and receiving feedback. We ask for feedback to improve or modify our behavior, to get help when we are stuck, or to get recognition when we think we are doing well sais North.
There are several reasons why people offer feedback. One reason is to improve the system of work, for instance when working in a team. Another reason is to model a culture of encouraging feedback, aiming to create a situation where feedback becomes acceptable.
Offering feedback can be misused for instance when trying to control people, or to demonstrate superior knowledge. An example of the latter is when people use feedback to convince that somebody is wrong and that they have the right answer.
The feedback cycle starts when feedback is offered or sought. The next step is that the feedback is heard and recognized. The third and last step in the feedback cycle is that action is taken. North mentioned that when exploring feedback you should be aware of all three steps and check to make sure that they are all working.
North gave an example of how to effectively use feedback when testing user interfaces with actual users. In stead of asking them what they think of the system you should sit with users and see what they do. When people work with the system and sigh, there’s usually a problem said North. At that point in time, ask them what the problem is.
North explained how people can effectively deliver feedback. Feedback should be about actual behavior and it should be specific. Feedback is most often taken at a personal level, this is why people receiving feedback interpret it different from the person delivering feedback. E.g. "Your work is sloppy" becomes "You are a sloppy worker".
The most effective model for structuring feedback that he knows is Situation, Behavior, Impact (SBI) said North. With this model you are giving feedback about observed behavior, together with the impact that the behavior has on you. The observed behavior should be factual, not judgmental, and you should describe how you feel said North.
North explained how he used SBI in a situation where two person were pointing to each other and talking over each other in stead of with each other. He ask the persons to sit together and come up with examples of the behavior, and explain what that did to them. He stated to them that if they can’t think of a situation with actual behavior then it didn’t happen.
North mentioned the ladder of inference by Chris Argyris. This concept explains how information that people take in will go through filters. The observable data and experiences will be filtered, next people will add meaning, make assumptions, draw conclusions and adopt beliefs before they take action. This is often ongoing at an unconscious level said North.
There are three models for structuring feedback, which you can practice to develop your feedback skills. The first one is porpoise feedback. With this model you offer specific positive feedback. You will only reward things that are going well and ignore all the bad stuff. The underlying assumption is that everything which goes wrong will self correct. This model can be used in low trust situations.
The second model is sandwich feedback. It has three parts: Specific positive regard, growing edge, and a general positive summary. You can use this model to provide something critical in a safe way. People can only give feedback this way if they genuinely care and notice things said North. The feedback should not be given as criticism, it should be "towards" motivation rather than "away from" motivation. This model can be used in medium trust situations.
The third model is Atkins feedback, which consist of only offering a growing edge. It’s the sandwich feedback without the bread, which works in high trust environments. The assumption is that people value each other’s opinion and trust each other.
For receiving feedback, the main thing is to listen and say "thank you". You can process the feedback later.
Feedback affect the system said North. You have to be honest about your own motives, why do are giving feedback. And practice giving and receiving feedback to develop your skills.