Hmm, I agree, only to a point. The way models are chosen and applied in science and finance are not quite the same, and the fundamental assumptions are also quite different. Science is also about observing black swans and adjusting for them, since they break the underlying hypothesis. On the other hand, a lot of science, especially physics is all about predicting black swans on the basis of observation or unexplainable phenomena. - Deepak Singh

But at the same time a black swan is something that is not considered in a model prior to observation of a black swan - in other words, if you don't know about the black swan, what are you predicting? One the messages I took from the book was that models *will* fail due to black swans - it's up to the modeler to not be dependent on the model. How? By having multiple models of different types. By having data from different source/experiments. Essentially, having a back up strategy - Rajarshi Guha

The thing is, in science it's OK for the model to fail, since it disproves your hypothesis and then you go back to the drawing board. In finance, one you aren't working with all the data, since human behavior isn't an exact science, and the consequences of model failure are a little more tangible. It's not the scientific method. I think the key is how we approach models. - Deepak Singh

In informatics and molecular modeling, we go with too many heuristics, partly due to computational limitations, and partly cause we don't have the right science in place, e.g. protein folding. - Deepak Singh

Coincidentally, a related post .. http://www.avc.com/a_vc... - Deepak Singh

Nice post, and yes you're right, the presence of hidden variables could very well be a root cause of black swan phenomena. But then one might ask, how far out do we need to go to get a fully global perspective. You also note that we use heuristics - I'd argue that as long we have 'models' - reduced representations of reality - we will always be employing heuristics. If we achieved a full description of a system (biological, physical, economical), we wouldn't need the model - Rajarshi Guha

True, a model is a hypothesis. I suppose the point, I am trying to make is that in science, part of our goal is to try and find out how models break, until we can't and they become Law, so we know there are Black Swans in our future, and we're always looking for them. - Deepak Singh

Couldn't resist ... http://mndoci.com/blog... - Deepak Singh

And now I have to run to my gate. Time to board - Deepak Singh

That is one of my favorite books - Black Swan events are intrinsically unpredictable. Self emergent behavior is way too chaotic. All we really have as chemists (at least synthetic) are heuristics. Drawing arrows is better than not drawing arrows but we should always remember it is not reality. - Jean-Claude Bradley