The op-ed by soon-to-be-ex Goldman Sachs employee, Greg Smith, is getting plenty of attention on the internets this morning. Called, “Why I Am Leaving Goldman Sachs,” the letter is a fine exemplar of the genre of bridge-burning parting shots by exiting employees. See also here and here. Smith excoriates the firm for what he says is its shift toward viewing clients only as profit centers, with the success of the client’s investments a distinctly, well not even secondary goal. Clients, it turns out, were often called “muppets.” On Twitter, Andy Borowitz went on a tear, tweeting: “BREAKING Goldman Reassures Clients: ‘All the Other Assholes are Staying.’”

I want to draw attention to one paragraph that captures the essence of the problem here.

What are three quick ways to become a leader? a) Execute on the firm’s “axes,” which is Goldman-speak for persuading your clients to invest in the stocks or other products that we are trying to get rid of because they are not seen as having a lot of potential profit. b) “Hunt Elephants.” In English: get your clients — some of whom are sophisticated, and some of whom aren’t — to trade whatever will bring the biggest profit to Goldman. Call me old-fashioned, but I don’t like selling my clients a product that is wrong for them. c) Find yourself sitting in a seat where your job is to trade any illiquid, opaque product with a three-letter acronym.

In simple terms, Goldman seeks to profit by encouraging clients to enter deals on bad information it provides.

The very reason we like private transactions is that we think people who trade with one another each want what the other has. They’ll be better off if they trade, and, therefore, we’ll collectively be better off. When you buy a car, you value the car more than the money you give to the dealer, and the dealer values the money more than the car. Simple. Here, Smith says, Goldman’s doing two things. It’s conveying information to clients in order to make them value a trade erroneously (erroneously form Goldman’s perspective). Then it’s transacting with them in a way that enriches Goldman but, presumably, impoverishes the client. And there’s no guarantee that Goldman’s gains even outweigh the client’s losses. You’d think the mortgage crisis never happened.

Economists may well suggest that we’re getting an efficient level of bad deals here. That is, clients don’t invest in obtaining more information on their own, because the marginal increase in deal quality that better information would give them is not worth the cost of acquiring it. Maybe, but this is a problem that repeats itself, and one wonders if this is really the best we can do. Your mechanic tells you that you need a new engine. The doctor tells you that you need a surgical procedure. The air conditioning repairman says that you need a new compressor. What do you do?

The simple fact is that our counter-party in each of these scenarios is wearing two hats. There is not one transaction, but two. On the one hand, they are our information agents, meaning that we are relying on them, and often paying them, to produce the information needed to decide whether to engage in a transaction. On the other hand, they are our prospective partners in that very transaction.

The case for regulation, for imposing legal duties on parties that both provide information and profit from our use of the information, rather than allowing the free market to do its thing is greater as the costs of getting a second opinion (in time, effort, and money) go up and as the cost on the agent of acquiring the information it’s withholding or distorting go down. (For a classic discussion of this, see this article by Richard Posner, especially page 21 about fraudulent nondisclosure.) In all of the above cases, there’s significant hassle in acquiring additional information — so much so that people often prefer investing in online systems to distribute information about trustworthiness (like yelp) to shopping engaging in more than one transaction to acquire information.

Now, there are of course costs that would owe to regulation here, perhaps even large ones to the extent we do more than punish outright, conventional fraud. And I’m not trying to make the case for any particular response. (I’m just kind of scratching around in this post.) Rather, I just found this to be an example of one of the many ways that free markets are not really free. They may, or may not, be freer than the alternative. But we are constantly boxed in by the state of our own knowledge, at the mercy of others whether we like it or not.