Random Beats Merit In Promotions?

In 1969, the Canadian psychologist Laurence J. Peter posited the “Peter Principle”: people in a workplace are promoted until they reach their “level of incompetence.” This happens, Peter argued, because we wrongly assume that people who are good at their jobs will also be good at jobs that are one rung up on the corporate ladder — so we promote them. But often the new job is so different from the previous job that the employee can’t handle it. Now performing incompetently, the employee stays in place, dragging the efficiency of the firm downward. Eventually the entire economy becomes like the paper company Dunder Mifflin in “The Office” — clogged with incompetence.

Is there any way to avoid this trap? Yes, by promoting people at random. That’s what a trio of Italian scientists discovered this year. They created a computer model of a 160-person corporation and programmed it with Peter Principle-like logic: the best performers were promoted, but they had only a random likelihood of being good at their new jobs. Sure enough, the firm was soon cluttered with incompetents, and its efficiency plunged. But then the researchers tried something different: they reprogrammed the firm so that it promoted people entirely randomly, and the overall efficiency of the firm improved.

They also tried alternately promoting the absolute best and absolute worst performers. That, too, worked out better than promoting on merit.

via Clive Thompson www.nytimes.com

Reading about this research back in July (see New Research Discovers Way To Avoid Organizational Incompentence: Promote Randomly) first gave me that Matrix sensation, like when Morpheus explains to Neo that he has always felt that the world was somehow off, strange, and irreal. The idea is so off kilter — that random promotions would be better than evaluating people closely to try to determine who would perform best — that it makes you wonder about management theories in general. What other ‘soundly established, beyond doubt’ premises are actually wrong?