The tenant the algorithm screened

Picture an applicant who has found the apartment. The commute works, the rent is within reach, the lease is ready to sign. She fills in the form, pays the fee, authorizes the background check, and waits. A day later a message arrives: her application was not approved. There is no number to call, no line explaining what disqualified her, no document she can hold up and dispute. A screening system reviewed a background report — some mixture of eviction records, credit history, and public criminal data — scored her against the property's criteria, and returned a verdict the leasing office simply relays. She did not meet the requirements. That is the entire account she is given, and it is the only account that exists.

What makes this moment worth stopping on is not that a decision was automated. It is what the automation is deciding about, and for whom. Housing is not a discretionary purchase; it is the precondition for holding a job, keeping children in a school, receiving mail, being reachable at all. And the person on the receiving end of a screening decision is, almost by definition, in a position of little power and less visibility. She is the applicant, not the customer. The landlord is the customer. She is the subject of a report she paid to have run against herself, and she is the last person permitted to see what it said.

One record, every door closed

The first thing the situation exposes is how far a single record now travels. A generation ago, a bad reference or a disputed debt was a local problem: it lived with one landlord, and the next landlord down the street knew none of it. That friction was inefficient, and it was also a kind of mercy — a bad data point had a short reach. Screening systems remove the friction. The same underlying data brokers, the same criteria, the same score feed decisions across a whole market at once. A record that would once have cost you one apartment can now, quietly, cost you every apartment run through the same pipeline.

This is the part the bare denial conceals. When the applicant is told only that she did not qualify, she experiences it as a single closed door, a piece of ordinary bad luck to be shaken off at the next listing. She has no way of knowing that the thing which closed this door will close the next one, and the one after that, for the same reason she cannot see. The harm does not stay where it landed. It compounds — silently, and across exactly the set of options a person searching for housing most needs to keep open.

A screening report you cannot see does not just deny you a home; it can deny you every home that trusts it, for a fact about you that may not be true.

Proving a negative against a report you cannot see

The second thing the situation exposes is what the applicant is actually being asked to do. To recover, she must contest a claim she has not been shown, made of records she did not assemble, held by a company she never chose. She must prove a negative — that she is not the person in the eviction filing, that the debt is not hers, that the criminal record belongs to someone else — without being told which of these, if any, is the thing that sank her. It is difficult to overstate how lopsided this is. The system holds the assertion, the evidence, and the verdict. She holds a form-letter rejection and a suspicion.

And the failure modes here are not exotic. They are ordinary, well-documented, and worth naming plainly. The first is the matching error: tenant screening keys on names and dates of birth, and common names collide. Someone else's eviction, someone else's conviction, gets attached to you because the identifiers were close enough for an automated join. The second is staleness: records that were resolved, sealed, expunged, or overturned continue to circulate because the pipeline was never told they changed. The third is the proxy problem: a criterion that never mentions a protected characteristic can still stand in for one, so that a rule about eviction filings or credit thresholds reproduces a disparity the law was written to prevent — and the errors of the first two kinds ride along inside it, magnified. None of these require bad faith. They are what happens when consequential judgments are built on third-party data that no one at the decisive moment is accountable for having checked.

The applicant can suffer any of these and never learn which. That is the specific cruelty of the bare denial: it does not merely withhold the reason, it withholds the ability to find the reason. You cannot correct a record you have not seen. You cannot rebut a match you were never shown. And a right to dispute that depends on you first guessing what to dispute is not a right in any usable sense.

Standing over the record deciding your housing

So the question the case forces is not whether screening should exist, or whether landlords are entitled to information about the people they house. They are, and it will. The question is what a decision this consequential, made on data this fallible, owes the person it decides against. The answer is not more accuracy — a more accurate screening system that still delivers its verdicts as an unappealable shrug leaves the applicant exactly where she started. The answer is standing: a claim on the record that is deciding her housing, strong enough to see it, question it, and correct it before the outcome hardens.

Concretely, that means three things the current arrangement withholds. It means the applicant is owed the actual basis of the decision — not "did not meet criteria," but the specific records and the specific criteria they failed against, in a form she can read and check. It means she is owed a real path to contest and correct: a way to say that eviction is not mine and have the assertion routed to whoever can verify it, with the decision held open long enough for the correction to matter rather than arriving after the apartment is gone. And it means a human judgment has to be reachable — that somewhere in the process there is a person who can be asked to answer for the outcome, rather than a score that answers for no one. A Decision Receipt that carries the record consulted, the rules in force, and the basis of the adverse determination is not a courtesy to the applicant. It is the minimum a decision of this weight, built on data this borrowed, has to be able to produce.

Housing is a basic good, and a decision about a basic good, made about a person with little power and made on records she had no hand in and cannot see, sits at the far end of the scale that measures what an account owes. The more consequential the decision and the less able its subject is to contest it, the fuller the account has to be — and a verdict that closes every door in a market, on a fact that may not even be true, demands the fullest account there is.

The scenario above is illustrative — a composite drawn to show a pattern, not an account of any real person, company, or event.

— Dispatches · Summit Cognitive