is just one of anal precision: there is a difference between irrationality and non-linearity and they shouldn't be conflated. Not that I know for sure which one the markets are (possibly a mixture of both, or neither), but they are two fundamentally different hypothetical causes of the effects.

I think the problem is more one Herring touches on: To what extent do you let self-interest-maximising models run? Surely at some point an ethical, moral or practical concern may be deemed of greater significance (the right to have a roof over your head, or not starve, for example); I still don't understand how americans can sleep at night knowing what's on their southern doorstep, for example, but maybe that's just me. Of course, those who do best out of self interest-maximising models are against that idea and those who do worst are for it (in many ways it's quite rational [in the economic / game theory sense]).

carefully, although you may find that relatively small investments at very long odds are rational in terms of long-term strategy (the cost of not being involved should x prove successful etc.). Also actions may or may not be rational based on the actions of your market competitors. That pesky feedback again.

I never claimed that investors are entirely rational; although it's important to understand that financial gain is not necessarily the only form of self-interest-maximisation; the preference comes before the quantification thereof, and concerns of rationality are applied to results and information based only on the measure of the a priori preference. When economists use the word 'rationality' in the game theoretic sense they mean something very specific and the rationality (i.e. reason) of the actor is only relevant isasmuch as it produces a preference which can be quantified. The actions with regard to this quantified preference are what make an action 'rational' or not; not whichever crazy mechanism produced the preference in the first place. If there is a hole in the current formulation of economic theory, it's that it doesn't (cannot) take the non-financial preferences into account; i.e. it by-and-large assumes everyone is playing the same game, which may well not be the case (we've all played games where one person had more fun expressing themselves within the framework [e.g. kamikaze moves] than by playing to win; e.g. some speculate that Soros did it for kicks more than for financial gain).