About an hour ago, it was annouced that Paul Krugman was just awarded the Nobel Prize in Economics. Jumping onto wikipedia to confirm, I soon succumbed to the problem with Wikipedia and ended up reading up again about mechanism design.
I remember first finding out about mechanism design almost exactly one year ago, when the 2007 Nobel Economics Prize was announced. Mechanism design is a simple idea: If you make some radically simplifying assumptions about people’s desires and methods, you can predict how they will go about behaving under certain scenarios. If you then know *enough* about how people will behave, you can structure the *rules* of the system to cause certain outcomes. For example, whether an auction is blind or non-blind can have an impact on how much people bid. What Hurwicz, Maskin & Myerson had done was to build powerful analytical tools to formalise this intuition and present it in a highly rigid, mathematical manner.
Which was great, it was fabulous and it was directly in line with what I was working on. I too was looking for how to understand the intersection between the rules of the system and people’s behaviour. But mechanism design was a formal, analytical and bottom up approach that only captured a tiny part of the hugely complex, social context of the real world. Which made me really wonder: How come I had never even heard of people trying to study this from a more holistic manner?
Let me give a simple example: The implicit rules I was brought up with for group discussion was that if you had a point to make, you raised your hand. The moderator of the group would then pick who had their hand up for longest. This was a *horrible* way to conduct group discussions because by the time they reached you, the group had moved on and the point you were trying to make was half an hour in the past. It didn’t hit home to me just how bad it was until I attended a philosophy discussion group which added just a tiny bit more process into the system.
To make a point directly related to the current point, you raised a single finger. To make a completely new point, you raised an entire hand. All the fingers got to go before any of the hands but as a finger, you were not allowed to speak more than 30 seconds and it was very poor form to meander too far off the original topic. As a result, digressions were quickly but efficiently resolved and the level of the conversation in the room was markedly more efficient.
The rules were slightly different and, as a result, the conversation was *better*. But when I was asked to *explain* why it’s better, the best I could do was to piece together ad hoc bits of common sense that I knew at the time. What I was seeking was some sort of framework that I couldn’t fit my explaination into.What was really surprising to me was how hard it was to find such a thing. Wherever I looked, nobody seemed to be studying this kind of thing. So part of this blog post is an appeal to people to point me to others who are doing the same kind of work because it’s still hard for me to believe that something so big and so obvious could have received so little study. My work has been so interdisciplinary I can only surface skim over a lot of different fields and it may well be I’ve completely missed an entire area of research.
So this is the message that I’ve been trying to craft for the last 2 years: Rules matter when you’re designing social systems and right now, people don’t have anything but intuition and rules of thumb to guide them. Mechanism design is awesome but it’s impossible to use in any practical sense. Without some sort of way of thinking about how rules should be structured, each system is at the whims of the individual talent of the designer. What’s more, without such a framework in place, it’s difficult for people to even recognise the aspects of mechanism design and clever and innovative designs are not propogated. This, to me is a situation which needs to change and it needs to change quickly.