"Busting" trades creates moral hazard because it acts like an insurance policy against mistakes. This means that automated trading ends up being less expensive than it would be otherwise, which suggests that there's likely an overproduction of automated trading.
(Read more: Who is tracking trading errors?)
The exchanges, of course, benefit from this situation because they generate fees from the automated trading. So they're incentivized to bailout the big customers like Goldman Sachs when the customers make mistakes.
Kid Dynamite, the former Wall Street trader who turned his poker blog into a must-read financial commentary site, agrees with Scholes.
I have been and continue to be a defender of the right of programmers to take advantage of their abilities to trade quickly. I'm not afraid of high frequency trading and I don't think it should be banned.
But if you're going to allow automated trading—high frequency or not—you have to allow the natural consequences of that trading. That means that when the code screws up – and make no mistake: what that means is that the people who WROTE the code screwed up—there are consequences.
You don't get a mulligan. You don't get a do-over. You don't get to erase all of the money losing trades that came as a result of your screw-up.
The current rules employed by the exchanges all have the SEC's stamp of approval. So any change along the lines Scholes and Dynamite are advocating would have to begin with the regulators.