Constructor Theory - A New Fundamental Theory of Physics

You may have observers that there is not so much physics in UnRisk Insight currently. The simple reason: Michael works hard to transform some new ideas into programs swiftly. For counterparts risk valuation. They are of eminent practical relevance, and were  triggered at a recent workshop with Solventis, here in Linz.

There is no such thing as an abstract program is one of the basic insights of a new fundamental theory of physics the constructor theory, developed by David Deutsch, Chiara Marietto…Oxford Univerity.

I am a lousy physicist, but I dare to write a little about this theory, because I found one example that I (hopefully) understand.

Robot control

You can write an offline, task oriented robot program but its constructor is the robot control. It's the entity that carries out the given task ("..pick a part of the box, put it on the palette...") repeatedly. The robot control is the foundational element - the constructor.

The robot control uses models that are calibrated and constantly re-calibrated to the real working space and situation. It may need sophisticated feature recognition...


In constructor theory, a transformation or change is described as a task. A  constructor is a physical entity which is able to carry out the task repeatedly. A task is only possible if a constructor is capable of carrying it out.

It sounds so simple, but it goes beyond Popper's science theory of falsification, because it touches information, computation and knowledge on a fundamental level. If we, for example, think of the idea of entropy in a thermodynamic system the link to information is strong…(oh, I'm already on icy terrain)

I take the practical view: there is no such thing as an abstract program…

In mathematics, BTW, there is no theorem without a model and a theorem comes to life only based on its operational semantics - the evaluation, the computation...

However one try: If I understand it right knowledge can be instantiated in our physical world and the better the instantiation the better the knowledge. This sounds quite evolutionary?

The evolution of the option theory

In the introductory book Quantitative Methods for Financial Markets (for students!), Andreas wrote: "the principles of risk-neutral valuation transforms the market into a 'fair game'". This rule has been instantiated by the BS formula by 1987. But, with the introduction of far out of the money options the smile was explored.

In the following, increasingly complex option models were/are introduced - among them models that cannot be validated (impossible to calibrate and re-calibrate..).

In the sense of the constructor theory, the task they represent is "impossible".  Too complex models are a fundamental trap.

How to avoid them is not easy, because you need to know in depth, where the computational limits are.    And the borders are moving...