Thursday, October 7, 2010

Asymmetric Accounting & Adaptive Networks

In this recent post, Gregory Rader explores the implications of asymmetric accounting. For those who need a quick catch up, asymmetric accounting is when two parties in a transaction account for the transaction differently. This process is a little different than we are accustomed to, as the concept of "price" generally ensures there is a symmetry in accounting between buyer and seller. Here are a few additional thoughts for your consideration.

Let's consider an adaptive neural network. The basic rule in any adaptive network is: try a bunch of different configurations, and strengthen those connections that lead to a preferred result. While our brains are one example of an adaptive network, this pattern also manifests in the evolution of ecosystems and even our social systems. In fact, our economy as a whole could be thought of as an adaptive network that reinforces configurations (innovations) that add value, and eliminates those that don't.

So what does this have to do with asymmetric accounting? If we were to consciously design an algorithm for the economy's adaptive network to function by, I would argue that insisting on symmetric accounting is a big mistake. Let's assume that our goal is to encourage lots of innovation / experimentation and provide reinforcement for the production of goods / services that add value to people's lives. By this logic we would want to measure the value-add of any good or service provided. However, you will notice that the price is agreed upon before any value is added to the economy. Consider for a moment how much distortion this causes. As a buyer, I might be suckered in by advertising that promises some great value-add for my life, only to discover that the product in question causes me nothing but grief.

Also consider that the buyer has every incentive to understate the amount of value they expect to receive, as this expression takes the form of them parting with their hard-earned money.

In order to overcome these two major barriers to a well-functioning adaptive network, I propose the following. First, we eliminate the concept of expected value as expressed through "price" entirely, and shift to a model where we are measuring the real value of the good or service received. There are, of course, many ways we could accomplish this, ranging from hard data derived from sensors to subjective reviews and ratings systems. The second would be remove any incentive for either buyer or seller to express anything other than their authentic experience around the transaction. And we all know that experiences can be asymmetrical.

Taken together, the picture this paints of the economy is quite different. In the interests of not rambling on for too much longer, I would assert that this algorithm would likely enable the kind of economy I described in this blog post from August.