Uber Didn't Win the Driver Fight. It Moved the Fight Somewhere Courts Couldn't Reach.

In 2013, a California driver named Douglas O'Connor sued Uber, arguing the obvious: that a company which set his fares, rated his work, and could shut off his access wasn't a 'technology platform' he happened to use — it was his employer.¹ He was right enough that a federal judge agreed the question deserved a jury.³ He was also about to discover that being right is not the same as being heard, because Uber had already done something quieter and far more decisive than win an argument. It had made sure most drivers like him would never get to make the argument in the same room at the same time.

The popular story is that Uber lost the fight over whether its drivers are employees. The truth is the reverse — and stranger. Uber has not won the legal argument; courts keep telling it the contractor label is a fiction. What Uber won is the right to never have the argument decided against it at scale. It converted a labor-law liability into a regulatory machine: fragment the lawsuits, rewrite the statute, redefine the category. The strategy works almost everywhere it controls the venue — and collapses the instant a court it can't restructure simply looks at what Uber actually does.

How you dissolve a class action without ever winning it

A class action is dangerous to a company for one reason: leverage. Tens of thousands of small, individually-unaffordable claims fuse into one large, existential one. The judge in O'Connor refused to grant Uber summary judgment in 2015, holding that whether drivers were employees was a question for a jury — a genuine threat.³ So Uber did not fight that fight head-on. It attacked the fusion itself. Buried in the driver agreement was an arbitration clause, and in 2018 the Ninth Circuit ruled it enforceable and reversed the class certification entirely.¹ In one stroke, the existential class evaporated back into tens of thousands of solo claims, each one to be heard alone, in private, with no precedent and no collective leverage. When Uber finally reached a settlement agreement for $20 million,⁹ the settlement reached only the roughly 14,000 drivers who happened not to be bound by arbitration — and even they stayed classified as independent contractors, with payments issued in 2020.² The vast majority were already gone, scattered into individual proceedings that no longer kept anyone at headquarters awake.

Notice what arbitration did and didn't do. It never disproved the misclassification claim. It made the claim un-poolable. That is the whole move — and it is why 'Uber lost O'Connor' is the most common and most misleading thing said about its labor record.

When you can't win the law, buy the law

Arbitration manages lawsuits one at a time. It does nothing about a statute. In 2019 California passed a law tightening who counts as an employee, aimed squarely at gig companies — and Uber's response was not to litigate it forever but to go over the legislature's head to the voters. The Yes on Proposition 22 campaign became the most expensive ballot-measure effort in California history, raising $205.7 million.⁴ Uber alone put in $59.5 million; DoorDash, Lyft, Instacart and Postmates supplied the rest, against a labor-backed opposition that scraped together $18.9 million.⁴ Uber even used its own product as a billboard, pushing in-app messages that asked drivers to click 'Yes on Prop 22' or 'OK' before they could work.⁵ On November 3, 2020, it passed. App-based drivers were now defined, by the voters of California, as independent contractors whom state employment law did not cover.⁵ The lawsuit Uber couldn't be sure of winning had been made moot by a statute it wrote.

The genius and the rot of Prop 22 sit in the same clause. The campaign sold a guaranteed earnings floor of 120% of minimum wage. But the floor applied only to 'engaged time' — the minutes a driver is actually on a trip or driving to a pickup — and ignored the hours spent waiting between fares. A UC Berkeley Labor Center analysis found that once that idle time is counted, the celebrated wage floor works out to roughly $5.64 an hour.⁶ The headline number was true. It just measured a sliver of the workday and called it the whole shift — the same trick, in wage form, that the contractor label is in legal form: define the boundary narrowly enough and the obligation nearly disappears.

The one room Uber couldn't redesign

Then there is Britain, where none of the levers worked. There was no arbitration clause to fragment the claimants and no ballot box to rewrite the statute. There was only a court, and the court did the one thing Uber's entire strategy is built to prevent: it ignored the contract language and looked at the relationship underneath. On 19 February 2021 the UK Supreme Court ruled unanimously that Uber drivers are 'workers,' not self-employed.⁷ What's devastating is the reasoning. The judges enumerated the control: Uber sets the fares, so it controls pay; drivers can't negotiate terms; a passenger ratings system can terminate a driver's access; declining rides carries penalties; and Uber restricts contact between driver and passenger.⁷ Read that list back. It is a description of an employer. The 'platform' framing — the same posture Uber adopted in litigation after court records show it had once marketed itself as 'Everyone's Private Driver'³ — simply did not survive contact with the facts. Uber had lost at every prior British stage since the original 2016 tribunal,⁷ and the highest court agreed unanimously — six justices, after one was unable to sit due to illness.¹⁰

Watch how Uber metabolized the loss. It immediately reframed a unanimous Supreme Court judgment as a narrow finding about 25 named drivers' use of the app in 2016, and promised to 'consult' the rest.⁸ That is the playbook in defensive mode: when you can't fragment the plaintiffs or rewrite the statute, fragment the precedent — shrink the ruling's reach in the public mind until it sounds like an edge case rather than a verdict on the model. It is the same instinct that built the arbitration clause, pointed backward.

Isn't this just a company that keeps losing in court?

The fair objection is that this reads as a winning strategy only if you squint — Uber was denied summary judgment in California, lost every British stage including the Supreme Court, and paid out a settlement. Doesn't a string of courtroom defeats just mean it's losing slowly? It's a real challenge, and the honest answer is that it depends on what 'winning' means to a company at this scale. Uber is not trying to win the abstract legal argument; on the merits, where courts look closely, it tends to lose. It is trying to keep the contractor classification operationally intact across millions of drivers for as long as possible, because the alternative — reclassification at scale, with payroll taxes, benefits, and waiting-time wages — is a different and far more expensive business. By that measure, arbitration and Prop 22 are wins precisely because they postpone the merits indefinitely, and the UK ruling is a contained loss precisely because Uber could quarantine its reach. The strategy isn't 'never lose.' It's 'never lose everywhere at once.' The vulnerability, which the British judges exposed, is that this only works where Uber gets to pick the room. Each forum it can't restructure is a crack — and courts that look past the contract are multiplying, not disappearing.

Uber's labor record is usually filed as a series of scandals and setbacks. It is better understood as a single, coherent answer to an inconvenient fact. The contractor label cannot survive a judge who studies the relationship — so Uber spent a decade making sure, as often as it could, that no such judge would ever be the one to decide. It fragmented the plaintiffs, then bought the statute, then shrank the precedent. The strategy is genuinely brilliant and genuinely fragile for the same reason: it doesn't depend on Uber being right. It depends on Uber never having to find out. And every court that simply looks — past the app, past the tagline, at who sets the fare and who can be switched off — keeps arriving at the same verdict the fine print was built to avoid.