Whenever I read opinion pieces by almost any macroeconomist— Keynesian, monetarist, new Classical, Austrian, etc, there is almost invariably a point where alarm bells go off. At some point the economist will make an assertion that seems to me to be in conflict with the EMH. And after that point I have trouble taking anything they say seriously. I keep thinking “If you’re so smart . . . ”What Sumner implies is more likely is that the market is working from the same data as you, and it has priced in the probability of your predictions already. It's more than a little hubristic to think that you've figured out something that no-one else can understand.
Also, if you think you're aware of a weakness in financial markets, be it regulatory or just some failure of the market to price something correctly - why aren't you rich?! If you're totally confident of your model, before you tell anyone (or even after, if you don't think they'll believe you), use it to make money! Whether or not you actually do this, I think this is a good internal confidence check, and perhaps you should use the results to determine how strongly you advocate for your desired remedy.
Update: Arnold Kling responds, saying that Sumner presents a false dichotomy between believing in the EMH and being rich:
But one can believe that the markets are wrong and still not get rich. The markets can be right. Or, the markets can be wrong in ways that you did not expect. As an investor, the prudent approach may very well be to act as if markets are efficient. Load up on those stock index funds and those inflation-indexed Treasury securities, and be done with it.Surely if you consistently bet that the markets are wrong and it turns out they're right, you should reconsider the EMH! If the market is wrong in ways you did not expect, then unless someone expected it, that isn't necessarily an argument against the EMH. It may have been that whatever information the market needed to make the correct call just wasn't there. The EMH doesn't mean that markets are omniscient, simply that they price all available information. If Barack Obama abolished lending at interest, the markets would plunge. This wouldn't be evidence against the EMH.
I also don't think that the market has to be perfectly efficient for Sumner's point to stand. If the market were arbitrarily or randomly inefficient, demonstrating this still wouldn't help you make money off it.
Here's a thought: if it became common knowledge that the market was informationally inefficient in some systematic way (animal spirits, etc), would the market price that and cancel it out? Is the only thing precluding us from a more efficient market a lack of scholarship?