The labor of being wronged

Consider what happens the moment a system gets you wrong. A flag lands on your file that should not be there. A mark appears on your record that belongs to someone else, or to no one. A benefit you are owed is denied by a rule misapplied to facts that were never yours. From the institution's side, the event is over — a value was written, a case was closed, the queue moved on. From your side, it has only just begun. Because the error, once made, does not correct itself, and the party that made it does not, as a matter of course, come find you to make it right. Instead the work of undoing it is transferred, silently and completely, to you. You did not make the mistake. You will do all the labor of fixing it.

This transfer is so ordinary that we rarely name it, but it is the quiet center of what it feels like to be on the receiving end of a machine decision. The apparatus that decides against you keeps the power to decide and sheds the responsibility to be right — or rather, it keeps the appearance of standing behind its decisions while offloading the entire cost of its errors onto the people those errors fall on. The result is a second job you never applied for, in a field you did not choose, for an employer who will not acknowledge that you work for them.

The second job you did not apply for

The work is real work, even if it produces nothing and is paid nothing. It begins with discovery, which is itself unpaid — you have to figure out that something is wrong, then figure out what, then figure out who to tell. From there the tasks are numbingly familiar to anyone who has done them. You gather documents to establish facts that were already true: statements, letters, identity papers, records of a history you have been living all along. You wait on hold, because the correction cannot begin until a person is reached, and the person is reached only after the hold music has extracted its toll in time. You fill out a form, and then, because the form went to the wrong department or the case was reopened under a new number, you fill out the form again.

And underneath all of it runs the strangest task of all — you are asked to prove a negative. Not that you did something, but that you did not; not that a fact is true, but that a conclusion drawn about you is false. Proving a negative is famously hard, and the difficulty is not incidental here. It is the shape of the burden. The system arrived at its conclusion by some path you cannot see, and you are made to disprove the conclusion without access to the path — to argue against a verdict whose reasoning was never shown to you. So you assemble evidence against a claim you can only guess at, and you send it into a process that will tell you, eventually, whether your guess was close enough.

Worst of all, the work does not stay in one place. A mistaken mark rarely lives in a single system; it has usually already traveled — copied into a downstream record, cited by a second institution, cached in a file you did not know existed. So even a successful correction at the source does not end the job. You must now chase the same fix across a landscape of systems that do not talk to each other, correcting the error again at every stop, never sure you have found them all. The decision that was made once, cheaply, must be unmade many times, expensively, by you.

The cost that shows on no ledger but yours

Here is the asymmetry that makes the whole arrangement stable: the cost of being wrong is nearly invisible to the institution and heavy on the person. When a system makes an error and lets the affected party clean it up, the error costs the system almost nothing. No line in its ledger records the hours you spent, the calls you made, the sleep you lost, the pay you forfeited to sit on hold during working hours. Those costs are real, but they fall entirely outside the institution's accounting — which means, from the inside, being wrong is close to free. And a cost that is free to bear is a cost there is little pressure to avoid.

On your side, the same event is expensive in exactly the currencies you can least afford to spend. It costs time, which is finite and unrecoverable. It costs attention, drawn away from work and family and everything else, toward a fight you did not pick. It costs a particular kind of stress — the grinding, low-grade dread of an unresolved thing hanging over your record — that does not photograph and does not appear in any complaint statistic. This is, in the plainest sense, a regressive tax. It falls hardest on the people with the least slack to absorb it: the ones who cannot take an afternoon off to wait on hold, cannot afford to hire someone to do the chasing, cannot easily produce the documents a smoother life keeps filed and ready. The people best able to endure a system's error are the ones least likely to suffer one; the people least able are the ones on whom the cost lands with full weight.

A system that makes the error and hands you the cleanup has discovered how to be wrong for free.

Once you see it as a discovery, you cannot unsee the incentive. An institution that bore the true cost of its own mistakes would invest in not making them, and in finding and fixing the ones it made. An institution that has externalized that cost onto the wronged has no such pressure. It can be accurate enough to satisfy an auditor and still leave a long tail of errors uncorrected, because the tail is carried by people whose suffering never reaches its books. The offloading is not a bug in the process. For the party doing the offloading, it is the cheapest available way to run.

Accountability means bearing your own errors

All of which brings the question back to what accountability actually is. We are used to measuring an accountable system by whether it lets you appeal — whether there is a channel, a form, a review. But a channel to appeal is not the same as accountability, and in this light you can see why. An appeal channel is precisely the mechanism through which the labor gets transferred: it is the door marked "your problem now." A system can have an immaculate appeals process and still be one that never bears the cost of being wrong, because the entire point of the process is that you bear it. To offer an appeal while keeping the burden of proof, the burden of discovery, and the burden of chasing the fix across systems is to keep the authority to decide while shedding the responsibility to make it right.

Real accountability includes the cost of one's own errors. It means the party that made the mistake does the work of finding it and unmaking it — proactively, at the source, and across the downstream systems the error reached — rather than waiting for the wronged to notice, prove, and pursue. The measure of an accountable system is therefore not just whether it lets you contest a decision, but who does the work when the decision is wrong. A record built for the affected individual points in this direction: a Decision Receipt that carries the inputs the decision rested on, the rules that were active, and enough provenance to trace where a conclusion came from, is a record that makes an error legible to the institution that produced it — legible enough to find and fix without conscripting the victim as its unpaid investigator. Standing to contest is the floor. The ceiling is a system that carries its own errors so you do not have to.

Until then, the labor stays where it lands. When a system gets you wrong, notice the transfer for what it is: not a neutral event awaiting your correction, but a cost that was moved onto you the instant the mistake was made. The error was theirs. The hours are yours. And a system that has arranged things so that its being wrong is free to it and expensive to you has not built accountability at all — it has built the opposite, and dressed it in a form for you to fill out.

— Dispatches · Summit Cognitive