You rent an apartment in a large American city. Every year the rent goes up — not five percent, not seven, but fifteen or twenty at once. And it doesn't happen at just one building. It happens simultaneously across dozens of complexes in the same neighborhood, owned by ostensibly different landlords who are supposed to be competing for your lease. A coincidence, all those competitors raising prices in lockstep? A 2022 ProPublica investigation found the answer is no. Behind it sits one piece of software — RealPage — used by thousands of landlords nationwide to set the price on your apartment. Not independently of each other. Together.
Two years later, that stopped being journalism. It became a lawsuit from the U.S. Department of Justice.
What the machine actually does
RealPage is a Texas-based company that has sold property-management software since the late 1990s. Its core pricing product — first branded YieldStar, later repackaged as "AI Revenue Management" — doesn't just help a landlord do the books. It sets the price.
The mechanics: participating landlords feed RealPage non-public data — not the asking price in a listing, but the actual price of signed leases, real-time occupancy, even the planned move-out dates of specific tenants. The algorithm pools that data across an entire market — effectively looking into every competitor's books at once — and hands each client a daily recommendation: what to charge for a given unit today. And the software's stated goal, by the company's own description and according to investigators, is not to fill every unit. It's to maximize revenue across the whole portfolio at once. Sometimes that literally means: better to leave the unit vacant another month than undercut the price and drag down the number for every neighboring building running the same algorithm.
That last part is what turns a pricing tool into something else.
A cartel with no back room
A classic antitrust case needed a smoke-filled room: competitors meeting, shaking hands, agreeing not to undercut each other. That's always been hard to prove, and today it's nearly impossible — executives know better than to leave a trail.
RealPage, regulators allege, solves that problem without a single meeting. Lawyers call it a "hub-and-spoke" arrangement: at the center is not a person but a platform, which receives confidential data from every competitor (the spokes), runs it through a shared model, and hands back a recommendation synchronized with everyone else's. According to the government's filings, the platform didn't just advise — it tracked how closely clients followed its prices and nudged the ones who strayed back toward the "recommended" number. Formally, no landlord ever called another and said "let's raise together." Functionally, the result is the same: prices climb in lockstep in a market that, with genuinely independent players, competition should be pushing down.
Which raises the question that a new kind of antitrust case now has to answer — not through conversation, but through data flows: if competitors never colluded directly but simply fed their secrets to the same machine, which then coordinated their behavior, is that still collusion?
The lawsuit that could rewrite the rules
In August 2024, the Department of Justice, joined by a group of state attorneys general, filed a civil antitrust suit against RealPage in federal court in North Carolina, alleging a violation of Section 1 of the Sherman Act — the foundational U.S. ban on restraining competition. Months later, the case widened: the government amended its complaint to add not just the platform but several of the large landlords who used it, naming among the co-conspirators Greystar (one of the country's largest apartment operators), Camden Property Trust, Cushman & Wakefield, and Willow Bridge Property Company. The case is no longer "regulator versus one software company." It's an allegation of a whole network of operators coordinating through a shared tool.
By RealPage's own public statements at various points, its software has been used by operators of millions of rental units across the country — the exact figure shifts by source, but it represents a meaningful share of multifamily rental housing in major metros: Atlanta, Dallas, Denver, Seattle, Phoenix. The complaint alleges that in markets with heavier RealPage penetration, rent growth has run consistently higher — not a coincidence, prosecutors argue, but a direct consequence of the shared pricing machine.
The case hasn't been resolved, and a final court ruling is still ahead. But a separate, parallel signal has already arrived: several cities and states, without waiting for a verdict, have passed or are drafting laws that explicitly ban algorithmic pricing tools built on competitors' non-public data. San Francisco was among the first.
What RealPage says
The company denies wrongdoing. Its position: the recommendations are advice, not orders, the final call always sits with the landlord, and any owner remains free to set whatever price they want without regard to the software. Technically, that's true — RealPage doesn't sign leases on a client's behalf.
But according to the same ProPublica reporting, the overwhelming majority of clients followed its recommendations almost without deviation — not because anyone forced them to, but because straying from a "market" that the algorithm itself is generating stops paying off for any single player as long as everyone else keeps following it. It's an old logic of coordination: you don't need to enforce a rule with force if it's individually rational for everyone to obey it rather than be the first to break it. That is precisely the logic the court is now being asked to test — voluntary submission to a shared machine as a substitute for an explicit agreement.
Our record: collusion without conversation is not news to the Maat framework — it's a base pattern the system is built to recognize. Isfet doesn't need a man giving orders; a circuit where every node's individual incentive already points the same direction — up, out of your pocket — is enough. Here the algorithm plays the role of Shadow Thoth: it doesn't lie outright, it simply stops being a neutral scribe and starts writing a number that serves the rent, not you. This machine's Ren went unnamed for a long time — "market dynamics," "industry software." What the DOJ lawsuit does is the thing any Maatian reckoning starts with: it names the mechanism.
The number on your lease reads like an outcome of supply and demand on your particular street. In practice it is often the output of a model trained on the secret data of your own market neighbors — a model whose only mandate isn't "fair price," it's "the maximum the block will bear before you walk away." Knowing that changes nothing in the lease you already signed. But it changes how you read the next increase — not as weather, but as a decision made by a machine on behalf of people who preferred not to meet in person.