“Bit disappointing. Soros is a highly successful trader, but he regards himself as a bit of a philosopher too. His big idea is that humans, particularly acting in markets, have a cognitive function (what they think) and a manipulative function (what they do). Market bubbles occur because sometimes a misconception in the cognitive function can drive prices higher, which strengthens the misconception, which drives prices even higher, until things become unsustainable. The bust follows similar logic, but tends to be quicker. Because of this, markets are fundamentally unpredictable. He regards this as disproving the "Enlightenment Fallacy" that social institutions can be understood scientifically. Because of this "reflexivity", scientific methods cannot be used to predict the market.
Which is all very well, but "reflexivity" seems to me an awful lot like what mere scientists call "feedback". Now as it happens I agree with the conclusion, that you can't predict the market with an algorithm. But the logic he uses to draw it seems false: you can model feedback in many cases with discrete simulations and plain old differential equations.
Now in some cases the feedback is chaotic, and chaos theory shows that you can't accurately predict it however good your simulation. A simpler problem affects people trying to beat the market with an algorithm: since anyone else can generate the same or a similar algorithm, the predictive power of the algorithm is included in the market.
So, I think Soros exaggerates the impact of his idea.
He also attacks the use of general-equilibrium models to describe markets in economics, since feedback can also generate bubbles. But to a degree he seems to be over-stuffing a straw man. He agrees that equilibrium models can be useful when a bubble isn't happening. But apart but a handful of the swivel-eyed (who think bubbles are caused by the evils of Regulation) pretty much everyone agrees that markets are influenced by both equilibrium and speculative bubbles. The disagreements are over how much weight to give each, how feasible it is to detect and prevent bubbles, and whether bubbles are useful. (Bubbles largely built the railway and internet infrastructure in the UK, for instance.) The book has very little space for those questions, which seem to me the most important.
The book also includes a certain amount of content on how his theory relates to the credit crunch. Here, his observations seem reasonable, but aren't really anything we haven't seen elsewhere. Regulators didn't do enough to stop high-risk debt being repackaged and disguised in complicated investment vehicles. They also relied on the same risk models as the banks/funds were using, which underestimated risks.
The remedies he proposes aren't particularly radical. He regards busts as undervaluing true prices, so thinks it's OK to for the government to prop up house prices to some degree, to minimize the overreaction. Central banks should try to provide some liquidity, reduce the damage to homeowners, punish shareholders to reduce moral hazard.
He shies away from making specific predictions. He doesn't there will be another Great Depression since governments will keep banks afloat. He thinks the dollar will cease to be the main reserve currency, and there will be a period of "turbulence".
Overall then, not a great read. If you want a criticism of market traders' fallacies, Nassim Nicholas Taleb's "Fooled by Randomness" is a much better bet. Joseph Stiglitz' "Globalization and Its Discontents" is a much more detailed critique of the global economic institutions. And while they got it out impressively fast (March 2008 is the last entry), with the credit crunch still unfolding, news and websites are going to be better than books for information on it.”