Sure, but significant facts come out of it:

* Economic systems are optimization algorithms on an NP hard problem. Hence markets have no special efficiency capabilities, they are simply a choice of optimization algorithm, there may be better ones out there with more favorable properties. Any optimization algorithm with similar resources could have effectively equivalent efficiency. This destroys the economic calculation problem.

* That you are wrong about the small epsilon. You can't make even that kind of guarantee in the face of NP hardness. What the market has a state may be wildly far away from the optimum, the market could be stuck in a local minima, and without P=NP you can't even know how far from optimum you are. Without a complexity class for markets you can't even begin to discuss it's properties, and that's what this paper is an attempt at.

It is your cleverness that misses the point, the author is trying to attach 21st century computational theory to economics (he may be wrong, but economic theories aren't going to help disprove it) and it's going to have some consequences.

>In economics, the difference between "efficient market" and "epsilon away from efficient" is very little.

In economics, yes. In real life, not so much.

Markets don't need to be 99.9999% efficient either. Central planning can have a wide variance in efficiency. It could be substantially more efficient or extremely inefficient. If markets are able to reliably produce more efficiently year after year, even at the cost of lower peak-efficiency, then markets can still triumph in the long run.