The Boss in Your Pocket

Logged Out

One of the workers interviewed by The Guardian in January described losing access to a delivery platform over a matter of minutes. Another said facial-recognition checks started failing because his beard had grown. A father in Lincoln watched half his income disappear when an app suddenly stopped giving him orders, then received conflicting explanations weeks later about why his account had been revoked. There was no manager to argue with, no office to walk into, no clear accusation to answer. Just a software system making decisions and a support channel that felt like shouting into padded walls.

That kind of story gets told as a glitch, an unfortunate edge case in a young industry. It’s a labor regime.

In February, Eurofound reported on a survey of almost 4,000 online platform workers across 15 EU member states. The numbers are ugly. Seventy-eight percent faced time tracking. Sixty-seven percent faced communications monitoring. Fifty-three percent were subject to screen surveillance through screenshots or keystroke logging. More than 40% worked under warning systems tied to performance thresholds, and a sizeable minority reported immediate termination or suspension when they fell short. This is the modern boss in a compact form: assigning, measuring, ranking, punishing.

Control Without Friction

Capital has always wanted labor that can be watched closely, paid by the task, and replaced without much fuss. Digital platforms arrived with a cleaner interface and sharper instruments. The old piece-rate wage fits very nicely inside a phone.

A supervisor costs money. A piece of software that allocates jobs, adjusts pay, tracks location, rates performance and issues penalties can be deployed at scale. It doesn’t get tired, it doesn’t explain itself unless forced, and it doesn’t build solidarity with the workforce. From the point of view of an investor, this is beautiful machinery.

The language around these systems is always softened: matching, optimization, trust and safety, quality assurance. Read worker accounts for ten minutes and the picture sharpens. The Guardian piece describes couriers trying to guess why the app gives a job to one rider and ignores another standing a few meters away. James Farrar watched one offer swing from GBP 14.74 to GBP 12.30 and up to GBP 16.08 in less than half an hour. The worker takes the gamble while the platform keeps the flexibility, and that matters when rent is due and the app is the one moving the numbers.

The ranking systems matter as much as the surveillance. Eurofound found that leaderboards and points systems are nearly everywhere in online platform work, including highly skilled remote work. Software developers and AI services workers were among the groups most likely to be under “comprehensive control.” That fantasy that this kind of management belongs only to couriers and drivers should be dead by now. The same logic travels upward. Once labor can be chopped into measurable units, management starts dreaming of total visibility.

The Point Is Discipline

People sometimes talk about algorithmic management as if the real problem were opacity. Opacity matters, and workers should know how decisions about pay and dismissal are being made; they should also be able to challenge them without begging a chatbot for mercy. Still, the black box serves another purpose. It disciplines people.

If your pay changes from one job to the next with no stable logic, you learn to accept volatility as normal. If access to work can disappear after a low rating, a delayed pickup, or a facial scan that fails under bad light, you become careful in a particular way. You stop taking risks. You stop arguing. You stop acting like a person with leverage. A workforce that can’t see the system clearly has a harder time fighting it together.

Look at what has happened with Glovo in Spain. After years of legal and political pressure over bogus self-employment, the company moved to an all-salaried model in 2025. Then came a new round of layoffs. In March, CCOO (🦐) accused Glovo of carrying out a disguised mass dismissal through disciplinary sanctions, citing terminations tied to alleged excessive delivery times, “no shows,” and even absences on days with red weather alerts. Formal employment helps, and union pressure helps, but the software logic doesn’t disappear because the contract changes. Management still wants a workforce that can be expanded, cut, and intimidated at low cost.

That’s why I don’t buy the line that these systems are neutral tools which sometimes get used badly. They’re being built for a labor market where insecurity is profitable. You can see it in the design choices. Constant tracking. Reputation scores. Dynamic pay. Automatic warnings. Thin support. Subcontracting piled on subcontracting. None of this is accidental.

The State Is Late, and Selective

The European Parliament was forced to admit there is a problem when it adopted the Platform Work Directive in 2024. The directive says platform workers can’t be fired by an algorithm alone. It requires human oversight for important decisions and restricts some categories of personal data processing. Those protections matter, and people fought for them.

I have no interest in romanticizing the state here. European governments didn’t discover platform abuse yesterday. They spent years tolerating business models built on false self-employment, wage suppression and outsourced risk. Regulation arrived after the sector had already matured, after millions of people had absorbed the cost in lost pay, arbitrary deactivations and legal limbo. Even now, enforcement depends on national governments that are perfectly capable of praising innovation in public while leaving labor inspectorates understaffed.

States also like these systems more than they admit. A workforce managed through apps is legible. It produces data. It can be monitored, taxed, nudged and, when useful, disciplined through the same digital channels that corporations use. There are conflicts between governments and platforms, certainly. There is also a shared appetite for centralization, traceability and control. Workers stand on the receiving end of both.

A Technology Worth Defending

I’m for technology. That has to be said clearly because too much criticism of platform labor gets flattened into nostalgia. Workers are asking for useful tools, and for those tools to stop treating them as disposable inputs. Route planning can reduce waste. Scheduling software can spare people a lot of dead time. Translation tools, logistics systems, cooperative dispatch software, public digital infrastructure, all of that can make work easier and safer.

The real fight is over ownership and control. A dispatch algorithm run by workers or a public service wouldn’t have to maximize the permanent vulnerability of the people using it. A rating system designed for safety could be transparent and limited instead of hanging over every shift like a threat, and productivity gains could shorten working time or stabilize income. Under capitalism they usually travel upward.

That is the class question sitting underneath all this. Platform owners use software to remove friction from management and bargaining power from labor. The interface looks clean, but the social relation underneath it is old: some people own the platform, while other people wait for it to blink.

I don’t think this gets fixed by asking nicely for better ethics statements from delivery apps. Workers need rights they can enforce, access to the logic that governs their income, union protection, real appeal mechanisms, and the ability to shut these systems down when they are used as weapons. More than that, they need power over the infrastructure itself. Otherwise the pattern will keep spreading from couriers to coders, translators, designers, teachers, care workers, anyone whose labor can be measured, sliced up and sold back to them as flexibility.

The boss now arrives as software, customer ratings, automated emails and prices that move when you are too tired to keep track. Supposedly this is modern management, but workers know the feeling already. Those above can see everything, while those below are left guessing. The software is new. The class relation is old.