Free riding
(for Zsóka)

In David Lewis' Convention he introduces coordination problems with several examples. One of these is the well known artificiality, convetionality of the traffic regulations. while everyone is driving on the same side of the road it is irrelevant which side they chose. Moreover if only a few participant chooses not to cooperate with the others and take the opposite direction than the rebels won they may be faster than the others. On the other hand if more people prefers to be rebel and go against convention total chaos is resulted.

Cyclists are constant rebels. If they participate in the traffic they can go in any direction practically and they are faster in shorter but reasonable distances (that are driven by cars for the most part in a city). So according to what is said above one would expect that universal cycling results in total chaos...but that's plainly false, that is free riding.

Constructive mathematics and agent-based modeling.

Though I suspect that its not only me who saw similarities between the philosophy of the two, I still find it valuable to write some sentences about this issue here.

For one example Leigh Tesfatsion does explicitly refer to constructivism in the philosophy of Mathematics in his Handbook chapter (CHAPTER 16. AGENT-BASED COMPUTATIONAL ECONOMICS:A CONSTRUCTIVE APPROACH TO ECONOMIC THEORY). He highlights that in constructivist circles one have to have an actual construction of a mathematical object if it is to be said to exist. Furthermore he mentions that proofs can be understood as computer programs in constructivist terms. And this is the line of symmetry between philosophies of mathematics and economics.

The symmetry may seem accidental but it may be not. Computational interpretation of constructive mathematics is most famously developed by Kurt Gödel, that became known as the Dialectica interpretation of constructive mathematics. Before that the notion of mathematical construction was somewhat obscure (that is a philosophical notion). Gödel managed to give a precise mathematical description of constructions with the help of recursive functionals. And recursive functions are computer programs, as defenders of functional programming paradigm like to say.

So I'm about to argue that as in the history of constructive mathematics a human cognitive notion (mathematical construction) was made more and more precise mathematically, the same process can be observed in economics, where human rational behavior is being explained with computer programs (that are precise formal entities with definite properties).

Constructive mathematics is usually disdained by mathematicians working in the classical paradigm (as something unusable) and in the same manner ACE is being overlooked by classical economists. On the other hand development of computer science showed how important and insightful discipline is the constructivist theory of mathematics, and I expect that development of multi-agent systems will display similar success for ACE.

While reading Butterfly Economics, I visited Amazon and ordered another book of Ormerod. The most recent one, with the most interesting title a book on popular economics ever could have: Why most things Fail, evolution, extinction & economics.
After reading the preface and introduction I'm looking forward to reading the whole book.

The first great gain concerning the book is that it refered to Mr. Ormerod's homepage and there I found his paper on Random Matrix Theory and the Failure of Macro-economic Forecasts which contains his arguments against economic predictions in detail. The article describes the problem very differently as I put it in the previous post, and it seems to be more reconciliable with my own views expressed there. Basically as it is expressed in the abstract Ormerod sais that the genuine information content in economic growth data is low, and so forecasting failure arises from inherent properties of the data. Introducing more confusing terms while putting the problem into a radically different perspective. Stating that the data or the subject of forecasting is what causes the problem and not the methods of forecasting. The amiguity, vagueness is due to the ill-defined macro-economic concepts like the GDP itself. So my argument in the previous post is somewhat irrelevant. However it can be restated and for this time we can formulate it as an Ormerodian argument.

Economists in the first place give credentials to data rows, then by established methods create forecasts that at best inherit the orginal (rather low) credentials of the data. Hence economic forecasts result in statments with very little credibility and usually fail.

I think that introducing the concept of information content does not add to the argument, and may even be misleading at least for a philosophically minded reader like myself. The argument can be formulated without reference to the (vague) concept of information, as I tried to show above.

I'm reading Butterfly Economics (by Paul Ormerod) and I'm a bit perplexed about his statements on the practical impossibility of predictions in economics. I understand and since I have very little information I believe him that prediction about for example GDP growth are generally erroneous, however this stance makes the work of financial state departments, national banks or any other economic policy maker institution susceptible. On the other hand any argument stating that the persons are bad economists making such and such predictions that prove to be wrong, these statement are necessarily false, since nobody could make faithful predictions except by chance.

However I believe that there are better and worse economic predictions. There are analysts who may say more informative statements on the rationally expectable economic situation. Ormerod's argument is slippery since he treats predictions as genuine bivalent propositions, but economists using them simply do not treat these as such. Economists usually (as I understand) assign probability to such statements, moreover they are baysian with respect to these. So if there is a prediction that is assigned higher probability by the majority of the expert economists, it is rightfully said credible, but not true in any way.