artITecture

Game Theory and IT Strategy November 20, 2008

I am currently reading the selfish gene by Richard Dawkins, it is a fascinating book not least because the thesis draws on research which combines (what i believe to be) previously unconnected analysis techniques like Game Theory with genetics with animal behaviors (ethology).

This really got me thinking how techniques from other fields of science could be used in Information Technology (if you consider IT science, perhaps it is an art!) Coincidentally I work in the field of IT strategy and at university my Computer Science dissertation on the card game of Bridge used aspects of Game Theory to formulate the best strategy for any given hand or game, although I can only consider myself, at best, a ham amateur in all of these fields.  (I am curious now why i never quite put IT strategy and Game Theory together before now, and a quick google search suggests that whilst game theory has ben applied to other areas of IT there it has yet to be applied to IT Strategy which is rather surprising)

Wikipedia defines Game Theory as follows ‘Game Theory attempts to mathematically capture behavior in strategic situations, in which an individual’s success in making choices depends on the choices of others.’

actually I would adjust this to be ‘Game Theory attempts to mathematically capture the outcomes of behaviour in strategic situations…’ I maybe saying the same thing as the Wikipedia definition but i want to stress that the analysis of outcomes are key in game theory not part of trailing secondary sentence, albeit that the analysis can only be as good as the model you construct.

My question is could a model be built which would could analyse different IT strategies and allow decisions to be made based on the outcomes (presumably the best outcome) – crucially Game Theory analyses the strategy in the context of other competitors and their strategies, the implication being that an strategy is only as good as the strategies of your competitors – if their strategy is better you are going to lose.

I would suggest that IT strategy and investment planning today probably does consider alternate investments and their likely outcomes but only in the context of the organisation itself and not the competitors. I suspect the vast majority of organisations call this analysis a business case . The measure of a business case is almost always a definite dollar return like headcount reduction which is convieniently provides a very definite and measurable outcome. However this excludes strategies which may bring a competitive advantage based on probabilistic outcomes (which I will call a ‘soft benefit’ which are outcomes like a probable increase to sales).

My experience is that any CFO is very unlikely to fund a project based on a soft benefit return not because CFOs are closed minded, but because the claimed return of soft benefits has wholly inadequate research and justification. The CFO is making a judgment call on the probability of an investment making the claimed return. In my experience soft benefits are frequently (always?) overstated, they are often highly speculative based on gut feel, they rarely (perhaps never) have good analysis and research behind them which backs up their claims. By using game theory the accuracy of the benefits and probability of a particular strategy success can be improved, perhaps to an extent where a CFO might take a chance.

so in conclusion, and a very unsatisfactory conclusion at that, i would like to build, or see a model built which could do this analysis. Perhaps on the train tomorrow i can think more on this and the likely parameters. At the weekend maybe i will search my loft for some of my forgotten economic textbooks.