The Peter Principle

Scott Adams, who writes the syndicated cartoon strip "Dilbert" wrote an article ("Manager's Journal: The Dilbert Principle") for the Wall Street Journal in 1995 in which he discussed a concept that he called the "Dilbert Principle", which states that organizations tend to systematically promote incompetent employees to management in order to get them out of the day-to-day operations of the organization.  Adams further explained, "I wrote The Dilbert Principle around the concept that in many cases the least competent, least smart people are promoted, simply because they're the ones you don't want doing actual work.  You want them ordering the doughnuts and yelling at people for not doing their assignments - you know, the easy work.  Your heart surgeons and your computer programmers - your smart people - aren't in management.  That principle was literally happening everywhere."

A similar principle was articulated by Laurence J. Peter and Raymond Hull in their book, The Peter Principle.  The so-called "Peter Principle" is slightly different.  Whereas the "Dilbert Principle" assumes that incompetent individuals are promoted into management positions, the "Peter Principle" assumes that people are promoted because they are competent.  The important difference is that according to the "Peter Principle", individuals are promoted until they eventually reach a position in the organizational hierarchy that requires knowledge, skills, and talents that they do not possess.  At that point, they become stuck in their current position, having reached a plateau.  Peter and Hull explained their principle by classically stating, "In a hierarchy, every employee tends to rise to his or her level of incompetence."  Peter also offered a corollary, "In time, every post tends to be occupied by an employee who is incompetent to carry out its duties."

While both the "Dilbert Principle" and the "Peter Principle" are meant to be satirical, there is evidence to support, at least in part, the "Peter Principle".  Importantly, the concept now known as the "Peter Principle" has been known for over 200 years.  Apparently, the Enlightenment playwright and philosopher Gotthold Ephraim Lessing wrote about a sergeant who avoids an opportunity to be promoted in his 1763 play Minna von Barnhelm, stating "I am a good sergeant; I might easily make a bad captain, and certainly an even worse general.  One knows from experience."  Similarly, the Prussian general and military theorist Carl von Clausewitz wrote that "there is nothing more common than to hear of men losing their energy on being raised to a higher position, to which they do not feel themselves equal" (this sounds a lot like the so-called imposter syndrome to me).  Finally, the Spanish philosopher José Ortega y Gasset articulated an early version of the  "Peter Principle" in 1910 when he stated, "All public employees should be demoted to their immediately lower level, as they have been promoted until turning incompetent."

It is with this background in mind that I read with great interest the 2010 winner of the Ig Nobel Prize in Management by Alessandro Pluchino, Andrea Rapisarda, and Cesare Garofalo, "The Peter Principle Revisited: A Computational Study" (this group also won the 2018 Ig Nobel Prize, which I wrote about in a recent post, "Lucky Breaks").  These investigators used an agent-based simulation model to show that in a hierarchical organizational structure in which the mechanism of promotion rewards the best members, then not only is the "Peter Principle" realistically possible, it is unavoidable and results in a significant reduction of organizational efficiency.

The investigators modeled a pyramidal organization made up of a total of 160 positions distributed over six hierarchical levels (there were 81 agents in the bottom layer, denoted level 6, 41 agents in level 5, 21 in level 4, 11 in level 3, 5 in level 2, and 1 in level 1).  Each individual agent was characterized only by age (ranging from 18-60 years, which increased by 1 for each cycle of the simulation) and degree of competence (ranging from 1-10) - these were randomly determined based on two normal distributions at the start of the simulation.  At each cycle of the simulation, all the agents with a competence under a fixed competence threshold (to model termination of employment) or over a fixed age threshold (to model retirement) left the organization, leaving their positions empty.  These empty positions were then replaced by promoting an agent from one level down and so on.  Empty positions at the bottom level were filled randomly with new agents (to model hiring at the base level).

Two possible mechanisms of transmission of competence were modeled (in other words, when one agent from the level immediately below was promoted to fill an empty position in the level immediately above) - the common sense hypothesis and the Peter hypothesis.  With the common sense hypothesis, the promoted agent inherited his/her previous level of competence with a small random variation (+/- 10%), while in the Peter hypothesis competence was randomly determined independently of the old degree of competence.  

Finally, three different strategies were used to promote agents from one level to the next - Best, Random, or Worst.  With the Best strategy, the highest rated agent (in terms of competence) was promoted, while in the Worst strategy, the lowest rated agent was promoted.  With the Random strategy, the agent who was promoted was randomly determined.  Overall efficiency of the organization was calculated based upon the summing the total competences of each level, which was then weighted by multiplying the sum of competences by the hierarchical level (1, 2, 3, 4, 5, or 6).

The simulation was run for 1,000 cycles and was further averaged over 50 different iterations of the initial baseline conditions.  The results show that if the best person is always promoted (as would be typical for most organizations I think), then the best organizational efficiency depends upon the common sense hypothesis being valid and true.  Moreover, the worst results occur if the Peter hypothesis is valid and true.  All of that makes sense, but surprisingly the best results were obtained when the worst performer was promoted (Worst strategy) and the Peter hypothesis was valid and true!  Promoting an agent at random was intermediately effective.  Just look at the following Figure and Table from the manuscript, which compares the gains and losses (from initial baseline conditions) for all of the combinations of promotional strategies and transmission of competence:

The most consistent (and probably safest) strategy, regardless of whether the common sense hypothesis or Peter hypothesis was true was to either promote by random completely (essentially drawing a name from a hat) or selecting the best and worst performers and then choosing at random.  In other words, the best way to improve organizational efficiency (as a proxy for organizational success) is to promote people randomly or to shortlist the best and the worst performer in a given group and randomly select between the two.  Finally, when they analyzed the progression of agents as they ascended the organizational hierarchy, they found that if the best performers are always promoted and the Peter hypothesis is valid and true, then all the agents will eventually end their career at the level where their competence is minimal (or conversely, where their incompetence is greatest), thus confirming the "Peter Principle".

So, what does all of this mean?  Let's take a look at a couple of the assumptions made by these investigators.  Is it reasonable to assume that individuals will maintain their previous degree of competence whenever they are promoted to a new and higher position in the organizational hierarchy?  Probably not (in other words, the common sense hypothesis Pluchino and colleagues describe doesn't always hold true).  However, I do think it is reasonable to assume that individuals will maintain at least some degree of their previous level of competence, so just assigning the level of competence at random (which is what Pluchino and colleagues did with their Peter hypothesis) doesn't seem correct either.  Regardless, the results of this study are certainly compelling and suggest that promoting the "best qualified" candidate may not always be the best strategy for the organization!