anti-inductive situations: forum for week of 17 October

I found today's lecture on the shortfalls of induction to be particularly interesting. I am specifically interested in exploring in what circumstances are we allowed to induce beliefs from incomplete knowledge? Suppose it were possible to produce everything we believe induction, and that induction is simple a matter of convenience (a highly improbable position in my opinion). This simplification is still obviously necessary – if nothing else, waiting for simple, deductive knowledge takes too much time. Case in point, people knew about the symptoms of diabetes long before they came to know its underlying physiological causes. Inductive beliefs in this case can be formed by doctors who know just of the symptoms and cases, and but of the underlying framework. So, induction serves as a sort of bridge between ignorance and knowledge (sound rather like the Daimon in the Symposium…).

So in what cases are we right in drawing such conclusions from such evidence? In class, we were talking about chartists and the markets. Sure, there seems to be certain patterns which dominate in certain assets – e.g. Fibonacci retracement. We hardly know why it should be the case that 61.8 or 38.2 should be support levels in stocks, but that seems to have been the case for quite some time. The pattern has some predictive powers, though, of course, it’s not very accurate by any measure. But the fact that through all the transactions that take place in the market place every second of every day, there exist a (weak) pattern is still pretty extraordinary – it’d very strange if it were just a quirk that disappeared tomorrow. Perhaps there are certain immutable laws about trading and exchanges that lead to these patterns, and it’s simply the case that we haven’t quite figured them out yet – that’s one view. On the other hand, there’s LTCM and the spectacular failure of algorithms and pattern seeking. But where is it that we actually draw the deciding line?