Eleventh Thesis is famous among Marxist, stating that “the philosophers always tried to understand the world until now, now we should act to change it”. Incidentally, this is the name of an important Marxist book series in Turkey that was published in the 80’s, well regarded as a carrier of a tradition that was struggling for existence at the time. It occurs to me that this thesis is somehow related to how the inverse problems are handled in Bayesian paradigm, which is the idea I explore in this post.
I believe the notion of forward and inverse problems is very important and I am not so sure if this is well understood in the social sciences. The idea is that, forward problems begin with clear axioms and develop from there, as the theory of value started with Adam Smith, passing through Ricardo to be finally developed rigorously by Marx. Rigorous development of a theory helps in understanding what may be the implications of certain set of axioms, that may not be obvious from the start. Yet the rigour should not be confused with truth, as this stage of development is more of an intellectual exercise, lacking connection with reality. This connection (as I also discussed in: http://blog.nus.edu.sg/egealper/2013/06/12/hello-world) is necessary to make the rational decision. Some people say this connection is prediction, but what kind of prediction? It looks to me that passive prediction is what people hope for (observing some facts without any intervention and seeing if they fit the theory); but they tend to get nothing and still be happy about it. Just as the engineering discipline is interested in active prediction, that is predicting a system response after we modify it in some way, the economy-politics should be interested in predicting the social response after we actively attempt to change it in some way. This, I believe, is the likelihood that must be sought. Therefore, the eleventh thesis is not just some revolutionary spirited slogan, but a methodology of scientific progress and filter. Prior theoretic exercises determine the intellectual quality of a thesis, but we may still have tons of them to explain a single phenomenon, all of which seemingly viable. It is the real life’s likelihood that ranks them in the end.