"Instead of being always right, financial markets are always wrong. They have the ability, however, both to correct themselves and occasionally to make their mistakes come true by a reflexive process of self-validation. That is how they can appear to be always right."
Some other key points are:
a) While the methods of scientific enquiry have proven successful in physical sciences, it is incorrect to use them to the same extent in social sciences - because of inherent uncertainty and indeterminacy.
I recently read an article arguing that financial markets collapsed because economics doesn't use some superior mathematical techniques that have been used elsewhere for sometime. Many people argue that better risk management techniques will make sure a crisis like this doesn't recur. That is incorrect - because risk is also a function of leverage, and not merely the volatility of the underlying cashflows. An inherently more risky financial product is less risky when bought with no leverage. Conversely, if people believe in "a great moderation in the volatility of inflation" (Greenspan and Bernanke) - and by implication cashflows, they will lever everything up more - so the system ends up with more risk. Uncertainty in participants actions makes the future indeterminate in social sciences. Light, on the other hand, doesn't change its speed.
b) The Enlightment tradition focuses solely on the cognitive function and not on the manipulative function. "The philosphers of Enlightment put their faith in reason, they saw reality as something seperate and independent of reason, and they expected reason to provide a full and accurate picture of reality". This philosophy has served physical sciences well for centuries. In security analysis, this is equivalent to constructing a DCF and a WACC to figure out the price.
"The postmodern approach goes the other extreme - by focusing on the manipulative function and treating reality as collection of often conflicting narratives, it fails to give sufficient weight to the objective weight of reality". In security analysis, this is the equivalent of saying all DCF is non-sensical. Or something similar to what the world leaders are attempting right now - "lets talk our way out of this recession."
The truth lies somewhere in between. Sometimes, financial theories work - and sometimes they do not. The trick lies in identifying the turns. Because "the behavior of markets is best regarded as a historical process". A long-term investor might not get the same prices when theories start reworking as when they stopped working. The movement of prices during the chaotic period might have irreversibly changed the fundamentals - like US regulators panicking on seeing financial stocks falling and seizing WaMu etc.
Where I disagree with Soros is when he extends this theory to politics. Soros gives the example of the Bush administration as following the dangerous post-modern philospohy, and how it not merely recognized that truth can be manipulated, but promoted the manipulation of truth as a superior approach. But haven't politicans done this throughout history and in all countries. Goebbels did it in WWII. Even De Gaulle was manipulating when he kept insisting "I am France (Je suis la France)" - when France was under German occupation and De Gaulle was living in UK.
The difference between politics and economics is - while it well accepted that politicians manipulate, it is not recognized that markets can be manipulated as well. Because the manipulators are we ourselves. The manipulation might be passive and not active because no one person controls it. But to the extent that the biases of thousands are able to turn imagination into reality through the market mechanism, it indeed is manipulation.