The Gig Economy's 'Reckoning' Keeps Getting Reckoned With — and Keeps Winning.

In 2019 California passed a law that, on paper, ended the gig economy. AB5 codified the strict 'ABC test' — to call a worker a contractor, you had to prove they were free from your control, doing work outside your core business, and running an independent trade. A driver ferrying riders for a rideshare company fails the middle prong instantly: driving people around is the core business. By that logic, Uber and Lyft and DoorDash had to make their drivers employees. The reckoning had arrived.⁶ One year later, the same drivers were contractors again — and the platforms had spent more money making it so than anyone had ever spent on a ballot measure in the state's history.

The story everyone tells is that regulation has finally caught up with the gig platforms. The more honest story is a seesaw. Every serious attempt to force these companies to employ their workers has been reversed by a ballot, shelved by the next administration, or quietly carved out by lobbying. The pendulum swings; the cost advantage stays.

Here is the thesis a smart friend could repeat: the gig-economy reckoning is real, loud, and chronically incomplete — and as of mid-2026 the structural labor-cost advantage of the contractor model remains largely intact, because the platforms learned to fight the rule that comes after the rule.

The single most expensive way to overrule a law

When AB5 made the platforms' drivers into employees, the platforms did not appeal it to death or comply with it. They wrote their own law and bought a yes vote. Proposition 22, on the 2020 California ballot, exempted app-based drivers from AB5 in exchange for a softer package of benefits — and the companies funding the Yes campaign poured in $205.7 million: Uber alone gave $59.5 million, DoorDash $52.1 million, Lyft $49.0 million, Instacart $31.6 million, Postmates $13.3 million. The opposition raised $18.9 million.⁵ It became the most expensive ballot-measure campaign in California history. The math of it is brutal and clean: spend a couple hundred million once, and lock in a contractor cost structure across the largest state economy in the country indefinitely.

Then came the part that mattered for whether it would last. Labor challenged Prop 22 as unconstitutional, and on July 25, 2024 the California Supreme Court unanimously upheld it.⁴ The headlines read like a permanent victory. The opinion was narrower than that: the court expressly left the Legislature room to pass new workers'-compensation laws for app-based drivers down the line. And because Prop 22 was a voter initiative, amending it now requires a seven-eighths supermajority of each legislative chamber — a threshold so high it functions as a lock.⁹ Durable, then. Not unconditional. The platforms didn't win the argument forever; they raised the price of reopening it to something close to politically impossible.

Why the federal rule keeps changing its mind

The same pattern plays out in slow motion at the federal level, where the relevant question is which test the Department of Labor uses to decide who counts as an employee under wage-and-hour law. On January 10, 2024 the Biden DOL published a final rule, effective that March, replacing a 2021 Trump-era standard with a six-factor 'totality of the circumstances' economic-reality test — a framework that, by weighing more factors, tends to sweep more workers into employee status.¹ A genuine tightening. It lasted about fourteen months as enforced policy. On May 1, 2025 the DOL's Wage and Hour Division issued a bulletin announcing it would simply stop enforcing the 2024 rule in its own investigations, reverting to a 2008 standard.² Then, on February 26, 2026, the DOL went further and proposed to formally rescind the rule and replace it with something close to the 2021 version — re-elevating 'control' and 'opportunity for profit or loss' as the core factors, the two factors most favorable to calling a gig worker a contractor.³

The mechanism here is the why behind the why. A regulation is not a wall; it is a setting on a dial that whoever runs the agency can turn back. Court rulings are more durable, but they decide constitutionality, not policy — they tell you the platforms may use the contractor model, not that anyone must stop them. And ballot initiatives, once passed, sit behind supermajority locks. So the platforms learned to route every reckoning toward the most reversible venue available and to raise the cost of the irreversible ones. A rule that took fourteen months to write can be un-enforced with a single bulletin. That asymmetry — slow to impose, fast to suspend — is the whole game.

Isn't the reckoning real where AB5 still stands?

The fair objection is that the seesaw framing is too cynical — that the reckoning has genuinely landed in places. It has, and the piece would be dishonest not to say so. AB5 survived a federal preemption challenge: on June 10, 2024 the Ninth Circuit upheld it against Uber and Postmates, and the ABC test still governs classification in California and other states that adopted it, like New Jersey and Massachusetts.⁶¹⁰ The 2024 DOL rule, even unenforced by the agency, technically still binds in private civil litigation, so a worker can still sue under it.⁸ These are not nothing.

But notice what each carve-out preserves. Prop 22 exempted the exact workers — app-based drivers — for whom the cost stakes are largest, leaving AB5 intact mostly for everyone else. The federal rule still applies only to wage-and-hour law and never adopted California's strict ABC test in the first place; it used a softer multi-factor test even at its toughest.⁸ The reckoning lands hardest precisely where the platforms chose not to spend, and softest where they did. And the workforce numbers suggest the structural pressure was never as large as the rhetoric: the most rigorous government count, the BLS Contingent Worker Supplement, found about 10.2% of the labor force — roughly 15.5 million people — in alternative arrangements as their primary job in 2023, essentially flat since 2017.⁷ The '36% of the workforce' figure you see everywhere — drawn from surveys like McKinsey's American Opportunity Survey, which explicitly counts anyone with any side income as an independent worker — runs well above what the BLS's primary-job measure captures.¹¹ The reckoning was loud partly because the number was inflated.

The gig economy's reckoning is real. It just keeps getting reckoned with. AB5 was the wall; Prop 22 was the door cut into it for the people who paid. The Biden rule was the tightening; a single 2025 bulletin was the loosening, and a 2026 proposal is the unwind. Every cycle produces a headline about regulation finally catching up, and every cycle ends with the contractor model still cheaper than the alternative across most of the work that matters to the platforms' margins. The reckoning didn't fail. It just discovered that a law you can reverse is a law the people it targets will outspend, outwait, or out-venue — and that the only reckonings that stick are the ones nobody can afford to undo.