You may also like...

2 Responses

  1. Thank you for your comments.

    —- Artem wrote > I feel that in your dismissal of game theory, you are targeting a straw man from the 80s.
    ———– Reply > Which is the game theory taught in universities that helps shape the opinions of students and future decision makers. This situation is of grave concern.

    —- Artem wrote > Modern game theory involves models with no rationality assumptions (such as Evolutionary game theory, in which I specialize) and bounded rationality assumptions (Algorithmic Game theory in general).
    ———– Reply > The unbounded rationality assumptions in neoclassical economics are also relaxed at times, in an attempt to provide more realism. However, the underlying core doctrine remains the same, that is, equilibrium and ration choice theory formulation. If the assumptions of game theory continue to be relaxed to attain more realism and the interconnectedness of the agents retained, the model starts to look like an agent based model. This evolutionary path of realisation most probably took Axelrod from game theory to agent based modelling.

    —- Artem wrote > Of course, ABM is often a tool to study inside these theories, however they are also treatable analytically.
    ———– Reply > To obtain analytical solutions in game theory requires gross over simplification of a situation. In contrast the computational solutions to agent based modelling simulation are unconstrained by the requirements of analytic solutions, which allows for more realism. Computing power is becoming cheaper allowing for even more realistic modelling.

    —- Artem wrote > A theory is always stronger when we can actually prove (in the mathematical sense) things about it instead of relying on simulations.
    ———– Reply > A mathematical proof within pure maths is a fine thing to display ones mental agility and prowess. However, within the social field the set of assumptions used to prove a theory are merely another ideology if the theory fails empirical testing or not falsifiable. The other aspect to mathematical proof is that a conclusion can be merely a restatement of the assumptions. Denniss (R 2012, The use and abuse of economic modelling in Australia, The Australia Institue, Canberra) discusses the circular logic of mathematical models. For instance, an assumption is made that modelling government is not required to model the economy and the conclusion made that the government has little effect on the economy.

    Economics seems to collect these zombie theories using mathematical proofs and continues to teach them, see ‘Zombie Economics’ by Professor Quiggin.

    Unless models or theories can be empirically tested, they could well be the celestial mechanics of a non-existent universe, which is fine as an academic pursuit but people actually take these models and theories seriously and use them in policy development.

    —- Artem wrote > This is especially true when our systems are so non-linear and sensitive to specification/initial-conditions and when we have such a poor grasp of how empirically observed and model parameters relate.
    ———– Reply > Testing simulations against reality, provides a scientific approach. The EU acknowledges the failure of traditional economics to predict so adopts agent based modelling

  1. 20th September 2012

    […] The Society Pages “Progressing from game theory to agent based modelling to simulate social emergence” […]

Leave a Reply

Your email address will not be published. Required fields are marked *