Ben Thompson’s Inequality Paradox

Access and decency in Uberfied world

Ben Thompson body-checks Paul Graham:

…Paul Graham’s weekend piece Economic Inequality ring hollow. Graham’s defense of the broad-based gains that accrue from new technology is absolutely correct….But given that much of those efficiency gains also contribute to winner-take-all dynamics, it is reasonable to expect that those winners — and their investors — pay commensurately more.

As usual, Ben is right. I want to spend a moment unpacking the idea that the winners should pay more. But before I do, it’s important to point out that Ben doesn’t go nearly far enough in his criticism. Graham doesn’t just drop the ball before he gets to the end of his line of reasoning. His reasoning is wrong.


Paul Graham likes to stacks observations one atop the next until he reaches some surprising insight. This rheotical simplicity is essential to his persuasiveness and charm. When he’s right, it allows him to quickly reach startling and non-obvious conclusions. The most notable success, of course, is the reasoning behind Y Combinator itself, which by any reasonable measure has been wildly successful.

But this structure also makes his arguments brittle — break one link in the chain, and the whole thing collapses. In the case of his inequality essay — which he helpfully re-issued in an explicit link-by-link form — the problem is here:

But startups also increase economic inequality. A founder whose startup succeeds will end up with stock worth a lot of money.

No, startups do not need to increase economic inequality. Founders (and investors) might end up with a lot of stock, but they’ll be the first to point out that stock doesn’t buy groceries. You need to sell it first. And that triggers taxes.

As a society, we could decide to raise taxes (on capital gains, among others) to a level that would halt or even reverse the current trend of increasing inequality. Importantly, there’s every reason to think you could do this without destroying the incentive to start companies. After all, I can think of a few companies started when the top capital gains rate was nearly 40%, the top estate tax rate was above 50%, and the top marginal tax rate was above 70%.

The point is that if startups don’t need to increase inequality, Graham loses his foundation for the claim that inequality per se isn’t bad, and therefore loses his reason for objecting to explicit efforts to reduce inequality.

Also, extreme inequality actually is bad.


Back to Ben Thompson. He says that the winners are going to need to pay more. What, exactly, should this look like?

Recall the central thought experiment of Ben’s piece. It’s New Year’s Eve, and Uber is in full surge-pricing mode. A poor pregnant woman goes into labor and needs a ride to the hospital. Competing for the Uber she wants to hail is some rich guy who’s too lazy to walk a few blocks. And, per Thompson:

It very well may be that the latter’s ability-to-pay will trump the former’s willingness-to-pay.

And so, the piece argues, our new politics should

focus on ensuring that everyone has the same opportunity to signal their preference.

This is such a perfect crucible for the political question at hand that I think Ben deserves to put his name on it. So let’s call it Ben Thompson’s Inequality Paradox. How do you preserve the market mechanisms that make Uber successful yet ensure that we have a society that, by default, treats the poor with basic human decency?

Thompson’s Paradox is useful because it goes beyond Uber. Indeed, it’s worth pointing out that Uber isn’t even profitable yet. But there are plenty of other companies—Apple, Comcast, Exxon-Mobile, Citigroup—that are wildly profitable and whose services (or those of their competitors) are already essential to functioning in the modern economy. How do you receive your wages without a checking account? How do you apply for a job without internet access? How do you communicate without a mobile phone? (Excercise left to the reader: for each of the above-mentioned companies, create a parable that maps onto Thompson’s Paradox.)

For now, let’s stick with Ben’s original framing: Uber, the rich guy, and the poor woman in labor.

We can begin by admitting that there’s no way to unwind this problem that doesn’t involve redistribution. The basic incentive structure of Uber falls apart if we take away surge pricing. So the Paradox requires that when the woman goes into labor, the price of her Uber is more than she could afford from her own income. Somehow, at that critical moment, events need to transpire that make it possible for her to pay.

How can we make this happen? My first instinct—and Ben’s, judging by the rest of the piece—is to jump to public policy:

  • Society can reduce inequality such that the disparities among the rich and poor are small enough that the rich guy won’t be rich enough that he’s willing to pay more than the poor woman in labor. Probably a pretty narrow band. So, communism.
  • Society can establish a floor below which members of society cannot fall, and set the level of this floor sufficiently high that if a pregnant woman goes into labor on New Year’s Eve and can’t afford an Uber, it was because she was demonstrably profligate. This might be something like a guaranteed minimum wage, which would require substantial tax increases.

But is public policy the only option? The Paradox arises out of the rational actions of three actors: Uber, a rich person, and a poor person. If we’re going to add actors to solve this problem, we can choose which actors to add. The federal government is just one of many possible actors to introduce, and while it has the benefit of being able to prove uniform, enforceable solutions, it has obvious drawbacks, not the least of which is that the most likely scenario is that it does nothing.

So let’s jump to the opposite extreme. What if we invite no new actors to the stage? Suppose Uber knows the woman is pregnant and poor. (There are myriad ways this could be accomplished, all with interesting implications and edge cases.) What if Uber created a mechanism for:

  • Opt-in decency. Public transit has always had signs flagging the easiest-to-reach seats as “reserved for the elderly and those with disabilities.” Uber could invite the rich guy to demur.
  • Imposed decency. Uber could prioritize her ride request and subidize her fare. Uber could potentially recover some or all of this subsidy through higher fares to other riders or at other times.

By the way, in the long run, these mechanisms might increase shareholder return. Uber’s long-run competitive advantage will come from a network effect. To ensure fast, inexpensive service, Uber needs to have the most drivers. To have the most drivers, Uber needs the most riders. And riders go to where the service is fast and inexpensive. What better way to get more riders than with subsidies? Today, when Uber enters a market, they flood it with ephemeral subsidies for everyone. Why not consider permanent subsidies for those that can’t afford to use the service regularly?


We’ve looked at two limiting categories of solutions to Thompson’s Paradox. One is pure public policy. The other is permissionless innovation. But maybe you like low taxes and you don’t think companies will take this problem seriously. In that case, consider a hybrid solution: publicly mandated innovation.

The FCC’s mandate to expand telecom access is a useful example here. The story goes something like this: installing communications infrastructure (first phone, then cable, now broadband) is expensive, so the telecom companies’ incentive is to get infrastructure to as many people as are required to make that infrastructure successful. That basically means cities. Left unregulated, a largely rural country like the U.S. would have vast tracts completely disconnected. So we subsidize rural communications infrastructure. We use public policy to make sure that telecom companies solve the problem.

In other words, we decided as a society that we weren’t willing to allow a third of our citizens to go without access to modern communication. Is telecom the only infrastructure that meets this bar? Why shouldn’t we think it’s important to guarantee access to transit, or to mobile devices, or to an internet connection?

Maybe we need to broaden the idea of a common carrier to include any consolidated industry that controls the tools for an essential area of human activity. And then we could require that all these common carriers treat those who need their services but can’t afford them with some basic level of decency.

Or maybe not. Personally, I’d be happy to see a substantial rise in the capital gains rate, along with the rates on dividends, estates, and the top income bracket. It seems like the simplest mechanism, and it could do good things for our politics overall.

Either way, we’d be a better society if untangling Thompson’s Inequality Paradox was a problem our politics and Silicon Valley’s royalty was willing to tackle head-on.

[EDIT: Removed some gratuitous snark.]

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.