Variability is both good and bad

Creativity is inherently variable. We value ideas from left field, out-of-the-box thinking, and idiosyncratic thinking. A degree of randomness is desirable in many processes including evolution.

But variability cuts both ways. Kahneman defines noise as “unwanted variability” in judgment and it is a direct result of our causal thinking. Whereas bias is a predictable error which sends a judgment in a particular direction, noise is unpredictable. And while bias can give us an explanation because it’s causal (say, a biased decision because of a sexist decider), noise is invisible and only detectable statistically.

In Noise, Kahneman, Sibony, and Sunstein tell the story of how they became interested in noise as a flaw in human judgment. They studied the decisions of underwriters in a large insurance company. Underwriters were shown summaries of risks and asked to make a decision on the appropriate premium to assign. The researchers measured the difference between the premium values that two underwriters assigned to the same risk. The executives expected the difference to be about ten percent. But the typical difference was fifty five percent. “Astonishing.”

After conducting research in fields as diverse as medicine, justice, and recruiting, they concluded that noise was as significant a problem as bias in human decision-making. In fact, noise is worse because it is invisible.  The authors explain how noise is unnoticed because we have the illusion of agreement: we think there is only one way to see the world. “We view the occasional disagreements with colleagues as lapses of judgment on their part.”

Causal reasoning means humans can avoid unnecessary effort while remaining vigilant to abnormal events. But it also means that humans naturally look for causes to everything even if they do not exist or if they are idiosyncratic to personal experience. This introduces variation in judgment and interpersonal conflict because each individual’s experience is different.

Noise in human judgment is a direct consequence of our natural causal thinking. Our intuitions develop from our understanding of cause and effect and it feels good to use them. When we use our intuition and make personal judgments in complex or ambiguous situations, we receive an emotional signal termed “judgment completion.” This reward signal delivers satisfaction and is difficult to resist.

Great Human Strength: Variability in our thinking introduces randomness which can be desirable and adaptive.

Great Human Weakness: Variability can lead to disagreement between people because if there is more than one to see something, people will see things differently. When we need decisions to be reliable and repeatable, variability in decision making can be dangerous.

Machine Opportunity: Designs which enhance variability in our thinking, duality in perspective, the ability to deal with tradeoffs and dilemmas or when we want to be creative. Designs which give us repeatable and reliable predictions when we need them (and tell us when that might be).

Machine Threat: Designs which amplify undesirable variability, reduce our ability to truth-seek or when we need consistency and singular perspectives.