To me stuff like this actually shows how some hard-science people apply autistic-like literal interpretation even to things that do not lend themselves to literal, narrow, interpretations. I'm sure, though, that many hard science people can expand their horizons enough to absorb different kinds of knowledge, too.
But the author claims that something is equivalent to P=NP! Now, P=NP does have a very "literal, narrow interpretation" (read: it has a very precise meaning), in fact, this is the only interpretation in which it makes sense at all.
So if you claim that something is equivalent to it, you better be prepared to define what exactly "something" is, and give a rigorous argument as to why it's equivalent. Some hand-waving with vaguely defined and ambiguous terms just doesn't cut it.
I have to agree. The author hasn't dealt with all the unsolved problems with bounded rationality, which quite frequently involves the inability to compute the future (the case where even if perfect information is available). Then, somehow, he "models" the cost as computational complexity, of a particular sort. But that's why we have peer reviews.
He interprets the definition of an efficient market as one that immediately represents all that is theoretically knowable about a security by the present time. Using this literal interpretation, everything that is knowable includes solutions to NP-hard problems. In fact he could have used it to show much more. If the price represents all that is knowable by the present time, than that means than not only all information is processed in zero time, information must travel instantly. He could have just as well claimed that for markets to be efficient information must travel faster than the speed of light.
The definition of an efficient market actually means that that the price of a security at time T represents all that market players can deduce about its history by time T, using information travel and processing speeds that market players possess. It really just means that everything that is historically known about the security can and will be used to determine its price as fast as possible. If it isn't possible - it won't determine the price.
