Nobel Prize-winning Israeli-American psychologist, Daniel Kahneman, explains at the Wharton People Analytics conference why it is almost impossible to hire the perfect candidate for every position
Daniel Kahneman, who won the Nobel Prize in 2002 for his pioneering work in behavioral economics, had revolutionized the hiring decisions of the Israeli Army as their 22-year old psychological evaluator.
At a time when most hiring decisions came down to intuition, he provided the interviewers a structured system of measuring young men on six dimensions. It had proved to be so effective that the army continued to use his exact method for decades to come.
The same expert who has long shown organizations the way to hire better, recently admitted that there was probably no such thing as perfect hiring for all positions. In an interview with Dan Pink at the Wharton People Analytics Conference on April 7th, he explained that no matter how advanced our systems get, we will probably never reach a point where a perfect candidate could be hired for every position. This is because some part of people’s job performance is unpredictable.
He gave the example of Google’s hiring process, which employs structured interviews and multiple interviewers to reduce the chances of one person’s faulty judgment. He didn’t think that any psychologist could help Google predict any better than they already were.
“The problem”, he said, “is not that we're poor at predicting; the problem may be that [performance is] unpredictable."
"There are chaotic interactions between the characteristics of the individual and the events that happen to them on the job that change them and that change the way they are viewed and treated by other people." Further testing of the individual certainly could not predict these outcomes in advance. It is simple but important point for people and organizations to keep in mind – they should make use of the most sophisticated hiring tools and analytics while being cognizant of the fact that human behavior may never be 100% knowable.
This sounds intuitively right, since as individuals, we are constantly in the process of self-discovery, unable to predict the course of future events and how they will affect us. Analytics can model our general behaviors better than we know, but it is hard to account for all swerves. In some ways, this reminds me of a mathematical limit where we say that a value can never be zero. It can get close — really, really close — but never become it.