Public Goods and Provisioning

Nathan Robinson proposed a thought experiment: an "American Free Diner" where anyone can show up and eat for free. Healthy, varied, no questions asked — breakfast, lunch, dinner, with a buzzing neon sign and great music. Not fancy, but nutritionally complete. The public option wouldn't replace private restaurants, just as public schools don't prevent private schools from existing. Anyone who wants a high-quality meal can simply walk in.¹

Scott Alexander's response is one of the sharpest defenses of voucher-style provisioning over direct public provision. His objections are practical rather than ideological, and they're worth taking seriously precisely because he agrees with Robinson's premises: capitalism produces unhealthy food, companies lie about nutrition, and the current system is genuinely broken.

The Class Sorting Problem

Alexander's first objection: any public cafeteria built to serve rich and poor alike would be 90%+ poor within a year, no matter how good the food. Rich people would feel uncomfortable for signaling reasons (they don't want to look like someone who can't afford better) and safety reasons (the same anxiety that keeps them off public buses). The least tolerant rich people leave first, then the slightly more tolerant, cascading until it's exclusively poor people. Going to the cafeterias becomes stigmatizing — "your family eats at the cafeterias" becomes a schoolyard taunt. And any service used only by poor people quickly deteriorates because its users lack the political power to demand maintenance.¹

Food stamps avoid this. Poor people go to the same air-conditioned supermarkets as rich people. They blend in. And if the supermarket deteriorates, rich customers raise hell, which fixes it for everyone. The poor free-ride on the political voice of the rich — a feature, not a bug.

The Autonomy Problem

A public cafeteria decides what you eat. If you love soda and have thought carefully about the health tradeoffs, tough luck. If your religion has complicated dietary requirements shared by three people in the United States, nobody's building a kitchen line for you. Kosher food will probably be available — Jews have political clout. Halal? Depends on who's in the White House this year.¹

Vouchers with taxes and subsidies, by contrast, let everyone make their own choices while nudging the price signals. Tax soda to make it more expensive; subsidize vegetables to make them cheaper. The soda-lover can still have soda — just at a slightly higher price. The person with exotic dietary needs can find their own supplier.

Alexander's third objection is about time and dignity. Under a public option, rich people eat in their own kitchen, on their own schedule. Poor people travel to a centralized cafeteria — adding transport time to already overloaded schedules, forcing intimacy with strangers for three meals a day. "Did anyone except the top-ranking bully enjoy high school lunch?" he asks. "Would they have enjoyed it more if it were limited to poor people?"

The Actually Existing Failure

The twist is that Alexander doesn't think the voucher system actually works well in practice. It's great in principle. In reality, the US government subsidizes the least healthy foods and restricts the most healthy ones. Between 1995 and 2009, the USDA distributed over $246 billion in subsidies, overwhelmingly to commodity crops — corn, wheat, soybeans — that feed into meat production and high-fructose corn syrup. Fresh fruits and vegetables are classified as "specialty crops" and receive no subsidies. Farmers participating in commodity programs are prohibited from growing fruits and vegetables on their subsidized land.¹

So we pay federal taxes to make Mountain Dew cheaper than it should be, then we're scolded by public health authorities for buying it, then New York proposes another tax to discourage us from buying it. As one journalist noted: "This is nuts."

The USDA also runs a "dairy checkoff program" that spent $202 million in 2011 promoting cheese consumption — including explicitly promoting pizza. For every $1 spent increasing cheese demand, farmers get $4.43 in additional revenue. The government is simultaneously telling you to eat less cheese and spending millions to sell you more pizza.

UBI as an Alternative Frame

The universal basic income conversation in Australia during COVID crystallized similar tensions. When JobKeeper payments hit A$39,000 annualized — more than many part-time and some full-time wages — the objection was predictable: people won't seek jobs if the government pays more than employers. The response from UBI proponents was that the data from every UBI experiment worldwide shows people continue to work; individual health improves; education participation increases.²

The more interesting argument was about big-UBI versus small-UBI. A big-UBI at $2,000/month lifts everyone out of poverty but requires serious tax reform. A small-UBI at $1,000/month is cheaper but sits below existing welfare payments, meaning it only helps if it replaces means-testing bureaucracy. The economists worried that UBI would cause inflation in supply-constrained goods — housing in zoned areas, medical care from guild-approved doctors — while having no effect on food or manufactured goods. This connects directly to the housing bottleneck: UBI without housing reform just hands money to landlords.

Higher Education as Public Good Failure

The COVID pandemic offered a natural experiment in what happens when a public good — higher education — loses its funding model. Universities in the post-2008 era had made up funding shortfalls by treating international students as a profit center. COVID's travel restrictions eliminated that revenue stream overnight.³

The prediction was grim: 2-3 years without an academic job market, followed by years of absurdly competitive hiring as a PhD backlog cleared, followed by a demographic shock as the smaller birth cohorts of 2008 reached college age. The practical advice — "learn to code," emphasize methods for the industry market — was also a tacit admission that the public good of academic research had been so thoroughly defunded that it could no longer sustain the people who produced it.

The deeper point is that the mode of provisioning matters enormously. Public schools, public food, public research — these all face versions of the same dilemma: direct provision risks class sorting and bureaucratic calcification; market provision risks capture by profit-seeking firms that optimize for the wrong thing. Neither "public" nor "private" is inherently better. What matters is the specific institutional design, the incentive structure, and — critically — who has the political power to demand that the thing works.