Better Alternatives

In January 2026, the Internal Revenue Service told state officials that Direct File would not be available for the filing season. The program let people file a federal return directly with the agency, for free, without a commercial product in between. About 300,000 people used it in 2025. Asked about the decision, the Treasury Secretary said the government had “better alternatives.”1

The better alternative is paying a company to do what the agency had been doing for nothing.

That line reads as a joke about a bureaucrat. It works better as a description of a market. A company that sells navigation of a maze has an interest in the maze. Intuit did not build the tax code, and its product is real: for a lot of people TurboTax is faster and less frightening than the forms. Companies defend revenue, and that needs no explanation. What needs one is why the alternative kept losing.

The record is not secret, and it is not one-sided. Under a memorandum with an industry group, the companies offered free filing to most Americans and the agency promised not to build its own system. The arrangement ran for two decades. The Free File Alliance said 70 percent of taxpayers were eligible. About 3 percent used it each year, and ProPublica reported that the companies had deliberately underpromoted their own free editions, hiding them from search engines.2

In 2019 Congress moved to make the arrangement permanent. The Taxpayer First Act, with bipartisan sponsors, contained a provision that would have barred the agency from creating its own online filing system. The language had appeared before in an industry-backed bill called the Free File Permanence Act. Intuit and H&R Block had spent a combined $6.6 million on lobbying the year before.2

Then the provision came out. ProPublica’s reporting produced an outcry, the sponsors disputed the interpretation, and in June the House and Senate passed the bill without it.3 That reversal matters here. The arrangement was contingent. It was defended, reported on, and partly withdrawn, by people who could be moved.

The memorandum expired in 2021. The agency then built the thing it had promised not to build. Direct File ran as a pilot in 2024, expanded to 25 states in 2025, and was gone in 2026.

So the elimination was available the whole time. The agency already holds most of the numbers on a wage earner’s return; ProPublica was reporting in 2013 that it could send people pre-filled forms.4 Capability was never the missing piece. A reason for anyone with money to want it was.

The asymmetry is in who can be paid. A company can charge for getting someone through a denial, a form, an audit, a maze—the commercial engine of the repair economy. Nobody can charge for making the maze shorter, because the saving is diffuse, arrives later, and lands on whoever happens to hold the budget. The workaround has an invoice and a customer. The elimination has a public benefit and no payer. So the workaround gets built, financed, sold and improved, and the elimination gets a pilot.

The founder’s part is the uncomfortable one. Intuit’s founders did not set out to defend a maze. They set out to make taxes less bad, and they largely did. The company became valuable by being good at the difficulty. Then every step that made it stronger made the difficulty more valuable to it: a durable brand, recurring revenue, a market that grows when the code gets more complicated. None of those decisions requires anyone to prefer the maze. The organization becomes less able to imagine a future in which it is not needed, and the person who started it can stay sincere while losing the freedom to finish.

No villain is required for that. It is what success does to an institution built to solve a problem it can now only manage.

None of this makes the maze pure waste. Filing a return is not an obstacle invented to sell software. The code encodes choices a legislature made, and a pre-filled form still asks a person to check it. A public filer can be underfunded, captured, or worse than the private one. The claim is narrower: where an alternative was buildable, was built, worked, and was stopped, the stopping is the finding.

An innovator can ask whether the thing is valuable. The harder question is who would pay if it worked perfectly. A product that dissolves its own market needs a payer with an interest in the dissolution, and almost nobody has one. What would have to change for that to be false?