Notes for a blog post or something on The Principle of Exquisite Fairness

I like to talk about the Principle of Exquisite Fairness which, technically, in full generality, is the Shapley Value of a cooperative game. But looking that up will make it seem a million times more complicated than it is in normal circumstances. So this is a more normal-person-friendly version.

You know about the idea of BATNAs in negotiation? I made a sidebar about it in case you don’t! Mostly the Principle of Exquisite Fairness is just this: Collaboratively figure out each side’s BATNA and then go with the average of them. (Which is in fact a special case of the Shapley value. And is also pretty common sense.)

The Shapley value is beautiful and important, but 99% of the time you don’t need its full generality. The special case of “midpoint of the BATNAs” suffices.

If you’re just here for the Principle of Exquisite Fairness, you can go home now.

What follows is a little spin-off essay applying the Principle of Exquisite Fairness to the question of how to decide how much money to charge for things.

Why Pay-What-You-Want Sucks

First, answering the question “what am I comfortable/happy/willing to pay?” can be paralyzingly agonizing. If you ask me that question I’ll be paranoid about paying less than what’s fair, especially less than the seller’s cost of providing the thing. I’d generally rather not buy something at all than risk being unfair to the seller when they’re explicitly trusting me to be fair. But there’s a bigger problem. Pay-what-you-can pricing means recruiting the potential customer’s help in working out what’s socially efficient but with insufficient information to be able to do so.

Here’s a heuristic I made up, for pricing in general: Always give the buyer a circumscribed optimization problem to solve. A fixed price is the simplest case. The buyer’s optimization problem is simply “is this worth it to me?”. Having student discounts or scholarships to apply for or different tiers, those all still yield a straightforward optimization problem: “What do I qualify for? What hoops are worth my time to jump through? Are these perks worth this extra cost?” And of course auctions are also circumscribed optimization problems.

But “how much would I be happy to pay?” is ill-posed. There are two superficial/extreme answers. One is $0, maximizing one’s own utility, and the other is the threshold, $MAX, beyond which you’d prefer to not have the good or service. There’s no such thing as a specific price that makes you happy to get the thing. It’s a continuum of decreasing happiness between $0 and $MAX. So how do you pick? The Exquisitely Fair answer, as explained above, is the average of $MAX and the marginal cost of what you’re getting, because those are the BATNAs.

My BATNA is doing without the thing, for $0 of utility. That’s also my utility for paying $MAX for it (I pay $MAX and get $MAX worth of utility, for a net of $0). The seller’s BATNA is keeping the thing, for $0 of net utility. That’s also their utility for selling it at cost.

So the range of possible prices start at the seller’s cost and go up to the buyer’s value. Those extremes are the buyer’s and seller’s BATNAs and so the Exquisitely Fair price is exactly between them.

If you want the buyer’s help in figuring that out, you need to give them more information. It’s not enough to ask what they want to pay. In fact, it’s literally not fair to ask what they want to pay.

Related Reading