Sunday, December 26, 2010

Knowledge and Planning

by Puck T. Smith on Sunday, December 26, 2010 at 11:41am

There are resources that exist right now that are not recognized as resources therefore they are not taken into account when calculating the availability and allocation of resources.

In the 19th century the main sources of artificial light was whale oil(1). Had some central planning supercomputer been available at that time then its calculations on the problem artificial light would have been confined to maximizing the production of whale oil. The resources required to discover other sources of light would not have been made available because they would have been misallocated to other uses based on incomplete information.

Centralized planning would only work if the knowledge of initial inputs were complete. That is impossible. It is for this reason that weather forecasting, for all the computing power being thrown at it, is impossible beyond a few days. In order to extend forecasts further than that would require knowledge of the temperature, barometric pressure and humidity of every single point on Earth. Again, that is impossible.

The same dynamical system principles apply to all forms of predictive calculation: only if you have complete knowledge of initial inputs can you extrapolate future behavior. Such complete knowledge does not exist. It becomes even more problematic when some of the inputs are not even recognized as inputs in the first place, e.g., if you don't know about kerosene it could not be factored into the artificial light problem.

This process was first quantified by Edward Norton Lorenz(2), though others, such as Friedrich Hayek(3), had already approached it prior to Lorenz. Lorenz demonstrated the impossibility of accurate future calculation using a relatively simple system comprising only three variable. Considering that there are far more than three variables involved in the dynamical system we call society--some of which variables have not been identified let alone quantified--to think it is possible to accurately predict the future and calculate resource allocation based on those predictions is the height of hubris.

Such calculation can only be made when the allocation decisions are distributed to those with the most complete knowledge of the inputs, those who are most affected by the consequences of the decision making process. Even then the calculations are at best approximations. When errors are made in a distributed system the effects are minimized and local. When those decisions are centralized and enforced by coercion the consequences of error are maximized and global.




No comments: