Organisation design is not always a piece of cake
Introducing the Nobel Prize winning theory of mechanism design
“You don't have to be a mathematician to have a feel for numbers” — John Nash
“If performance evaluations were a drug, they would not receive F.D.A. approval, they have so many side effects, and so often they fail.” — Bob Sutton
Why are we so bad at designing effective organisational mechanisms?
In so many aspects of organisational life, we operate a long, long way from the optimum. Take, as just one example, performance reviews. Performance reviews are almost universally conducted and the outcomes are almost universally terrible. A mere 14% of employees strongly agree that their performance reviews inspire them to improve1. Even worse, about 30% of traditional performance reviews are so bad that they actually drive performance down!2
The extant problem with organisational design is two-fold. Firstly, our collective intelligence is diminished by our faith in managerialism. The erroneous belief that an elite group of people (managers) exclusively possess the knowledge and skill necessary to run organisations places far too much weight on their unavoidably biased opinions. The situation is compounded by giving the same people asymmetric power for decision-making.
Secondly, designing procedures to generate desirable outcomes in groups of humans, or any adaptive agents for that matter3, is difficult. An individual’s goals are usually only partially aligned with the organisation’s and frequently at odds with others’.
The first problem, reversing managerialism by dismantling bureaucracy and managerial hierarchies, is the central theme of this publication so I won’t go into it here.
Today, I want to offer some thinking around the second problem.
Mechanism design
As mentioned, designing procedures to generate desirable outcomes in groups of people whose motivations are not necessarily known or aligned is difficult. In 2007, the Nobel Prize in Economic Science was awarded to Leonid Hurwicz, Eric S. Maskin, and Roger B. Myerson for laying the foundations of mechanism design theory, a toolkit for addressing this exact type of problem.
Mechanism design is arguably the best framework we have for designing organisational procedures, yet it is rarely included in MBA programs.
Mechanism design is sometimes called ‘reverse game theory’. In game theory, we analyse a particular game to predict what players will do. In mechanism design, we start with the desired outcome and design a ‘mechanism’ that will achieve it.
A mechanism is an institution, procedure, or game for determining outcomes.4
Unfortunately, the language of game theory and mechanism design is rather technical. However, as an introduction, a lot can be gained without maths.
I’ll use a simple example to give you a taste. Say I have a piece of cake that I want to divide fairly between my two sons, Jack and Tom. No matter how carefully I try to divide the cake, I can guarantee there will be an argument, “his piece is bigger”, or “his piece has more icing.”
However, there is a simple mechanism to produce the optimal outcome whereby both are happy: let one son, say Jack, cut the cake, but let the other choose which slice they want. Jack is strongly incentivised to cut the cake fairly and Tom is happy because he gets to choose his own piece. Problem solved.
Of course, organisational design mechanisms will be a little more complicated. Let’s return to the performance review challenge.
eBay’s peer review system
eBay China devised a nice mechanism to address a common problem when transitioning to Agile: “Scrum is all about the ‘team’. People self-organize and share the team’s performance. But how about the individual’s performance within the team?”. 5
They devised a system based on the insight that team members who work closely with each other on a daily basis are the ones who know most about the performance factors that make the team successful; thus, they can give the most meaningful feedback and evaluation.6
Before looking at their system, let’s apply a mechanism designer’s lens across the challenge.
What is the desired outcome?
The first step of mechanism design is defining the desired outcome. Many performance review mechanisms are highly confused. They try to compress improvement guidance, bonus calculations, promotion or firing justification, accountability, peer feedback, and legal compliance into a single mechanism. It’s way too unwieldy.
For simplicity, let’s focus on actual performance improvement, leaving any bonus and other considerations aside for now.
We might define the goal of the mechanism as providing the optimum level of feedback for catalysing individual performance improvement and growth.
Based on performance research7, we might decide that the mechanism should be:
contextual: providing performance feedback on factors most important to the collective performance of the team
accurate and legitimate: as free as possible from bias and poor judgement
timely: delivered frequently enough for learning to take root
succinct: rapidly digestible and not a burden to perform
future-focused: focused on future actions rather than past performance
How does eBay’s system measure up? Well, here’s how it works.
Once a month all members of the team rate each other against agreed criteria using a simple multiple-choice form, with one free-text field. So that meets the timely and succinct objectives.
The developers themselves came up with their own performance criteria (communicativeness, quality of work, collaboration, continuous improvement, role sharing, energising influence, overall satisfaction, general comments). These were not imposed by management. This means the feedback is contextual and legitimate. Also, because the resulting score is aggregated, bias is reduced and accuracy is increased through the ‘wisdom of the crowd’ effect.
The results are not published until all inputs are made to prevent anchoring (following other’s lead)
The aggregated feedback is made public, providing a strong incentive for self-improvement
The data can be visualised in various ways, allowing problems to be pinpointed and addressed quickly
The online system does not provide future-focus, per se. The individual might reach out to others or a coach to obtain guidance on how to improve, or a coach observing the dashboard might offer assistance to individuals struggling in some area.
Is it implementable?
A core focus of mechanism design is the notion of ‘implementability’, which tries to determine if the predicted outcome will equal the desired outcome.
To do this we need to consider the player’s incentives, what payoffs they get for each alternative strategy that they might adopt.
For the eBay system to work well, each team member should provide candid feedback to each other member. One strategy a player can adopt is to provide candid feedback, as desired. But another strategy might be to provide false feedback. They may wish to rush through the survey to get to an appointment, or they might want to punish a person they are having a disagreement with, for example.
Because the survey is anonymous there is no direct downside for not giving honest feedback. The recipient of the feedback can’t come back and query the scorer. On the other hand, providing candid feedback would give a player the positive payoff of having a voice. One would anticipate that most players would choose to provide candid feedback rather than withhold it.
However, it may be the case that some team members are disgruntled or disengaged and obtain some sort of ‘dark’ payoff by providing dishonest feedback. If this is the case, the mechanism designer might add another feedback question, such as: “How would you rate this person’s willingness to provide candid feedback?”. This would have the effect of strengthening the mechanism by making the quality of one’s feedback also subject to review.
I have barely scratched the surface of mechanism design in this article. Please note that the eBay example is not supposed to be any sort of best-practice case study. To do a proper mechanism design would involve mathematical modeling that is beyond the scope of this article.
The purpose of the article is simply to point a budding organisation designer towards some powerful theory that is rarely spoken about in management literature.
It is a toolset with enormous potential for designing better organisations.
Learn More
For a good (non-mathematical) introduction see Eric Maskin’s Introduction to Mechanism Design at the Warwick Economics Summit
For those who are mathematically inclined, start with Mechanism Design: The Implementation of Society's Goals. If some of the concepts such as Nash Equilibrium need deeper clarification, there are plenty of game theory tutorials on the web.
Journal of Mechanism and Institution Design
For more about performance feedback research, see:
Jackie Gnepp et al., ‘The Future of Feedback: Motivating Performance Improvement through Future-Focused Feedback’, PLoS ONE 15, no. 6 (19 June 2020), doi:10.1371/journal.pone.0234444.
Sutton, R., Wigert, B., 'More Harm Than Good: The Truth About Performance Reviews', Gallop Workplace, May 6, 2019
Cappelli, P., Tavis, A., 'The Performance Management Revolution', Harvard Business Review, September 2015
Image Credits
Cake: Couleur via Pixabay
eBay screenshots: Daniel Gu
Sutton, R., Wigert, B., 'More Harm Than Good: The Truth About Performance Reviews', Gallop Workplace, May 6, 2019
Kluger, A. N., & DeNisi, A. (1996). The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin, 119 (2), 254–284. https://doi.org/10.1037/0033-2909.119.2.254
Remember Microsoft’s racist chatbot?
Maskin, Eric S. "Mechanism Design: How to Implement Social Goals." The American Economic Review 98, no. 3 (2008): 567-76. http://www.jstor.org/stable/29730086.
ibid.
See Learn More section for some performance research resources