Enshittification, page 14
But that dream shrank. As Big Tech consolidated its grip on the tech sector, venture capitalists and founders understood that the most likely “exit” for any startup was acquisition by a Big Tech company. This accelerated after the dot-com crash in the early 2000s, when the stock market entered a prolonged doldrum.
The new dream for tech workers became “Work for a big, lumbering company for a few years, then strike out on your own and do a ‘startup’ with the intention of being rehired by your old employer or one of the other giant tech firms.” These transactions are called acqui-hires, because the purchaser isn’t really interested in the target’s products, which are typically shuttered immediately following the acquisition. Those products don’t exist to be used by people; they’re really just a kind of postgraduate thesis project for techies, undertaken to prove that you can conceive, create, and ship a product. The company’s purchase price is divvied up among the early employees of the startup according to their shares, with a hefty payout to the venture capitalists—basically functioning as a hiring bonus and a finder’s fee.
This is an extraordinarily wasteful way to run a corporate recruiting system! The workers who “win” the acqui-hire game spend years working on a fake product, pulling all-nighters, and neglecting their families. The customers who love that product sink money and attention into it, not suspecting that it only exists as a way to demonstrate the team’s willingness and ability to wreck their lives in service to shipping code. The “investors” in these acqui-hires spend millions to get one successful acquisition, at very high risk.
But still, for tech workers, acqui-hires represented a shot at winning one of the medium-sized prizes in the Silicon Valley lotto: a lump sum of cash and a chunk of stock in a Big Tech company, and a job that you could do until you got bored and started another “company” in the hopes of getting your old boss to “rehire” you with another cash/stock bonus.
As acqui-hires slowed down, techies’ dreams shrank further. Young engineering grads started to strive for an entry-level job at a gigantic company that provided free massages, a laundry service, and a company kitchen with a complimentary assortment of every kind of kombucha known to humankind. If you hit every one of your key performance indicators, this would be a job for life of the sort that middle-class people once took for granted. (The corollary of the mid-century adage “Nobody ever got fired for buying IBM” was “Nobody ever gets laid off from IBM.”)
Then, in 2023, the US tech sector laid off 260,000 workers. In the first half of 2024, it fired another 100,000 workers.
Tech workers’ dreams have shrunk to pinpricks. Now the aspiration is “Get a $300,000 engineering degree, get an $80,000-a-year job at a tech company, and pay off as much student-loan debt as you can before they fire your ass in the same year you hit every one of your performance metrics while they’re making record profits.”
At this point, you’ve learned about all the ways that companies wriggled out from under the systems that disciplined them and prevented them from enshittifying. Even as they slipped free of those bonds, tech workers held the line.
Even after acquisitions, predatory pricing, preferential discounting, exclusivity deals, and other antitrust violations allowed the tech sector to sew up the market and eliminate competition, tech workers held the line.
Even after the newly consolidated industry captured its regulators and convinced them that violating our labor, consumer, and privacy rights was okay so long as it was done with an app, tech workers held the line.
Even after those captured regulators were pressed into service to smash startups, tinkerers, hackers, and hobbyists whose plug-ins, mods, and aftermarket parts disenshittified the products their bosses managed to enshittify, tech workers held the line.
But today, after devastating layoffs, the line once held by tech workers has broken.
Tech bosses know this, and they’re thrilled with it. In the records of the Delaware Chancery Court where the lawsuit to force Elon Musk to make good on his binding promise to buy Twitter played out, we find evidence of it. Musk and his friend, the “investor” Jason Calacanis, had a chummy back-and-forth by text message in which Calacanis relished the possibility of firing Twitter employees as a means of enticing the survivors to put in longer hours and toe the line on Musk’s corporate strategy: “2 day a week office requirement = 20% voluntary departures. Day zero … sharpen your blades boys.”
All through 2024 and into 2025, we’ve been treated to daily triumphant blasts from tech bosses about how they’ll replace workers with AI. Tech companies are ordering work-from-home coders back to the office and delighting in the “voluntary” departures of those workers. It’s a hard time to be a tech worker.
Tech Rights Are Worker Rights: Para and Tuyul Apps
As any chickenized reverse-centaur could tell you, enshittified tech can transform any job into a nightmare. But in the absence of enshittification, in a world where workers can seize the means of computation, digital technology is a powerful tool for clawing back power from the bosses directly, even when the regulators and lawmakers who are supposed to have the workers’ back fall down on the job.
That’s because twiddling is a double-edged sword. While twiddling can erode gig workers’ pay (through algorithmic wage discrimination and other, even blunter scams, as we’ll see in a moment), counter-twiddling can push pay back up again, pitting algorithm against algorithm.
To understand why this is viable, we first need to spend a moment on security theory.
It’s useful to conceptualize security matters as a contest between defenders (anyone who wants to keep the status quo) and attackers (those who want to change it). This is a bedrock of security thinking. You may have heard of military exercises that pit red teams (attackers) against blue teams (defenders).
Broadly speaking, attackers and defenders can be assumed to be in possession of similar knowledge and techniques: the same earthmoving machines used to build the ramparts for a fortress can be deployed by a besieging army to tear those ramparts down. However, despite the two equally matched sets of technological possibilities, the red team usually prevails. For the blue team to successfully defend the status quo (that is, a fortress with walls that are still upright and acting as barriers to entry for the city within), it has to build walls with no flaws. For the red team to undermine those walls, it only has to find a single mistake that the blue team made, and the walls will fall.
This is the attacker’s advantage. Under enshittificatory conditions, workers are on the blue team (hoping to defend their wages) and bosses are on the red team (trying to twiddle those wages down).
But when workers can twiddle back—when they can avail themselves of technology that counters the bossware that turns them into reverse-centaurs and other tormented beings—the teams switch sides. They become the attackers, the red team, and the advantage is theirs.
Here’s an example: If you drive for DoorDash, you have to contend with an extremely shabby wage-theft trick. DoorDash tells drivers that they are their own bosses, that they get to pick and choose which jobs make sense for themselves. In practice, though, DoorDash hides a key detail from its drivers (whom it calls Dashers): how much the customer has committed to tipping. Tips are a hotly contested item in the gig economy, and multiple companies, including DoorDash, have been caught stealing their workers’ tips, a practice they ended only under legal duress. Since these tips can constitute the majority of a Dasher’s compensation, Dashers are required to commit to making runs that are often so poorly compensated that the driver can actually lose money on them, after factoring in fuel and vehicle wear and tear.
In hiding the tip from Dashers, DoorDash can entice drivers to commit to money-losing jobs. Why would the company do this? Because cheap jobs are enticing to customers, who love a bargain. DoorDash started out offering money-losing deliveries that it subsidized out of its investors’ capital. (Stage one of enshittification: Allocate surpluses to end users.)
Now DoorDash wants to recoup that investment by removing its subsidy. If it can trick Dashers into clocking in on money-losing jobs, it can shift the subsidy costs to them. (Stage three: Take surpluses away from business customers and hand them to shareholders.)
Meanwhile, DoorDash strongly suggests to its customers that they should tip heavily for better service, hinting that they won’t get their deliveries in good time unless they tip (all this on top of creeping junk fees that get piled up at checkout). Some customers take the bait and offer big tips. (Stage two: Make things worse for end users to make them better for business customers.)
The fact that some Dashers can be tricked into taking jobs on a money-losing basis turns ordering from DoorDash into a kind of slot machine: as a customer, you put a dollar in the slot and pull the handle, and if you get lucky, you get a below-cost delivery. DoorDash can tweak the odds here by occasionally stepping in to offer a subsidy when no driver rises to the bait (giving out giant teddy bears).
For Dashers, the fact that some customers will pony up for big tips that you don’t find out about until after you make the delivery turns driving into a casino game, too. Put your dollar in the machine (clock on for a job), pull the lever, and if you’re lucky, you get an extra $10 in tips. Again, DoorDash can improve the odds by adding its own subsidy here, giving out giant teddy bears to drivers, too.
This is a grubby little con game, and its technical execution is very clumsy. It turns out that when DoorDash sends a job to the app on a Dasher’s phone, the job listing includes the tip amount, but the Dasher’s app just hides that information from the Dasher. This is some pretty bad blue teaming: “I sent my attacker a secret, but I put it in an envelope marked ‘Secret: Do not open,’ so I’m pretty sure they won’t learn my secret.”
That’s where Para, a small startup, enters the story. Para noticed that this extremely useful information was being sent to Dashers but hidden from them, so it created an app that revealed the tip, the secret number that a Dasher needed to know in order to figure out whether a job was worth taking.
DoorDash lost its mind. The company sent a threatening letter (your basic “felony contempt of business model” threat) and smeared Para with nonsensical claims that its app could expose Dashers to identity theft by stealing their Social Security numbers (implying that the DoorDash app stored Dashers SSNs in an insecure state, which is quite a weird flex, but whatever).1
To its credit, Para refused to back down. DoorDash poured resources into fighting Para. I had a conversation with the company founders in 2022 during which they told me that a DoorDash insider told them that there were forty engineers assigned to blocking Para from uncovering Dashers’ full compensation.
Today, Para has expanded its offerings, with an “autodecline” feature that lets gig workers automatically decline any job offer that is below a certain level. This is part of a “multiapp” strategy that aims to let gig workers set up accounts with multiple services—Uber and Lyft, DoorDash and Grubhub, and so on—and get a dashboard showing them which service is offering the highest payout from moment to moment, playing the gig companies off against one another.
Gig workers are ground zero for this kind of counter-twiddling, because they’re the first class of workers whose boss is an app. When your boss is an app, you live with algorithmic wage discrimination and other arbitrary forms of abuse, like getting suspended without any explanation or appeal. Apps do all kinds of sneaky things.
The industry has thought up all kinds of ways to break the law and get away with it. When your boss is an app and you are misclassified as an independent contractor, you have no one to argue with and you’ve got little legal recourse.
Of course, this strategy isn’t perfect. Instawork is an app that dispatches people to work as low-wage temps. In July 2023, when members of the UNITE HERE Local 11 who worked at Orange County’s Laguna Cliffs Marriott Resort & Spa, owned by the University of California and managed by Aimbridge Hospitality, tried to bargain for a contract with their employer, the boss refused to deal, triggering a strike.
So Marriott/Aimbridge turned to Instawork to drum up an army of robo-scabs, dispatching gig workers to cross the picket line in a bid to break the strike. Any worker who refused to cross the picket line was permanently blacklisted from Instawork, which is the primary source of temp work throughout Southern California. Firing an employee for refusing to cross a picket line is extremely illegal, but by muddying the waters about whether workers were contractors or employees, Instawork was able to offer this scab-as-a-service deal to the region’s major employers, who were only too happy to take advantage of it. No one called them on it—until they tried to make Thomas Bradley a robo-scab.
Thomas Bradley was an unemployed culinary worker who lived in his car. He refused to cross the UNITE HERE picket line at the Marriott and was blackballed by Instawork. The union took his case to the National Labor Review Board, fundraising to get him shelter and then getting him a real, full-time job at a nearby union hotel, the Westin Bonaventure Hotel & Suites in downtown L.A.
There’s something genuinely wonderful about workers who counter-twiddle their bosses’ apps and escape reverse-centaurism. For example, drivers for Amazon Flex—a gig delivery platform that’s even more exploitative than Amazon’s Delivery Service Partner scam (discussed on page 120)—figured out that they could get more jobs if they bought burner phones and hid them in trees near Amazon warehouses. The drivers then installed remote-control software on their main phones that let them see the burner phones’ screens and send taps and clicks to them.
This worked by tricking the Amazon dispatch algorithm into thinking that the drivers were at the warehouse’s doorstep, which got them priority for new delivery jobs.
Sure, such counter-twiddling is a cat-and-mouse game that Amazon can eventually win by throwing more programmers at its driver dispatch app, but it’s also the only way that the drivers can “bargain” with Amazon. Cell phone–festooned trees might be the only path Amazon has to find out how its delivery prioritization scheme affects the actual delivery workers.
Automation supercharges the ability of workers to push back against their employers. In the Lehigh Valley in Pennsylvania, the community of Dashers is small enough that the drivers all know one another. Dashers there were able to use a hashtag, #DeclineNow, to agree among themselves to reject lowball app offers, which pushed wages higher for all drivers. Before long, the #DeclineNow forum had forty thousand users, and wages were rising for Dashers everywhere as a result.
Apps like Para can automate #DeclineNow, and even coordinate among workers to raise the threshold for declining an offer, pushing it up and up.
But to get that kind of counter-automation, startups will need lots of risk-tolerant capital: enough money to pay programmers to bust through a forty-engineer-strong anti-counter-twiddling team at DoorDash, while fending off DMCA, CFAA, copyright, patent, trademark, and other legal attacks.
Other countries dangle a tantalizing hint of how well this can work. In Indonesia, motorbike riders find gig work offering both taxi and delivery services. Early on, these riders formed cooperatively run clubhouses to offer shelter, repair, and social services for members. These co-ops became worker hubs where riders discussed the problems they had with their robot bosses, and that led to the creation of tuyul2 apps: apps that modify the functionality of the gig platforms’ apps.
Tuyul apps offer a wide range of functions to riders. One popular tuyul app increases the font size used by the official dispatch app, which allows older riders who struggle with tiny type to read information about the next job without finding their reading glasses. (As a person who is currently typing on a laptop screen that’s turned up to 200 percent magnification, I approve.)
Other tuyul apps are far more ambitious, like the app that allows riders to spoof their location. This is widely used by riders who are hoping to pick up taxi fares when commuter trains pull into busy train stations. Under normal conditions, the dispatch apps will not assign a train-station pickup to a driver unless that driver is right in front of the station. This creates incredibly dangerous traffic jams, as riders, family members, hawkers, and others press up against the station as the train pulls in. So smart riders install the tuyul app and wait around the corner, while spoofing their location so that, as far as the app knows, they are sitting in pole position to pick up a taxi rider. This is a higher-tech, easier, and more reliable version of Amazon Flex drivers hanging burner phones from trees in front of Amazon warehouses, but it shares the same essential characteristic: by allowing workers to seize the means of computation, these tactics override the shortsighted, dishonest, or just foolish choices of their bosses.
Tuyul apps are a hint of what a fair automation-centaur fight would look like, one in which bosses didn’t get a government-backed veto over how their workers’ technology worked.
As with privacy blockers plugging the gap for a long-overdue update to privacy law, tuyul apps aren’t a substitute for better labor law. But even so, they represent a significant improvement over a world of weak labor laws and no adversarial interoperability self-help measures for workers.
What’s more, a worker whose wages and working conditions are being (imperfectly) protected by counter-twiddling is a worker who has more time and money to apply to the project of improving labor law and labor protections.
And finally, rational bosses who know that workers can counter-twiddle are incentivized to treat their workers more fairly in the first place. Just as a printer company has to factor in the possibility that hiking ink prices will drive users to figure out where to get third-party ink (forever), just as web companies have to worry that making their ads more intrusive will trigger their users’ installation of permanent ad blockers, so will bosses using an app to cheat workers have to consider the possibility that this abuse will trigger workers to create and install counter-apps that make it impossible for the bosses to ever know whether they are getting accurate data from their apps again.
