## The experiments that keep succeeding
Finland tried it. 2,000 people, two years. Participants were happier, less stressed, slightly more likely to find work. Parliament did not extend the program.
Stockton tried it. 125 residents, $500 a month. Full-time employment rose from 28 to 40 percent. They spent the money on food, car repairs, utility bills. Congress did nothing.
Kenya tried it. GiveDirectly's long-running cash transfers showed durable gains in consumption and well-being. Kenya's government did not adopt the model.
The experiments keep succeeding. The legislatures keep declining to scale them. At some point, the pattern becomes the finding.
## Why the math works and the politics doesn't
The idea polls above 50 percent in the abstract. It collapses the moment the question shifts from "should people receive a basic income" to "should your taxes pay for it." A family earning $65,000 does not experience a neighbor's $12,000 annual payment as a policy innovation. They experience it as a $4,800 line item. Every funding mechanism — VAT, wealth tax, carbon levy — runs into a different coalition that blocks it. The welfare system everyone hates has proven immune to replacement for sixty years because voters prefer a bad system they know over a good one that costs them something visible.
The evidence is settled and the political vehicle does not exist. Twenty-seven years of Cherokee casino dividends produced no decline in labor force participation and a 40 percent drop in childhood behavioral disorders — data that cite and Congress ignores. But evidence is not the only thing that matters. When the Chevy plant closed in Lordstown, the damage was not primarily economic — it was the men who set alarms for thirty years and suddenly had nowhere to go, a loss that say no check can replace. Meanwhile, every funding model — VAT, wealth tax, carbon levy — runs into a different blocking coalition, which is why want strings attached: not because strings are good policy, but because strings are sellable policy.
And while the legislature argued about the size of the check during COVID, neighborhood fridges in the South Bronx were feeding everyone who showed up — no application, no caseworker. The networks and disaster-relief kitchens stopped waiting for permission and built the safety net themselves.
## The automation clock
AI is compressing service rates in sectors where human labor was assumed permanent. If the machines do the work, UBI stops being theoretical and becomes a question about whether a species that tied survival to employment for ten thousand years can untie the knot before the employment disappears. Everyone agrees the current welfare system is broken. The argument is about what fills the holes — a universal check, a conditional floor, a voluntary network, or the stubborn insistence that the answer is still a job, even if the jobs are leaving.
Where do you stand?
Join the conversation →