The toll on the appeal

Every appeals process has a cost, and it is worth being blunt about who pays it. Not the money cost of a filing fee, though that matters too — the whole cost. The form that runs to nine pages and asks for a reference number printed on a letter you were never sent. The line that is open between ten and two on weekdays, which is to say during the hours you are at work. The wait that stretches long enough that the deadline you are contesting passes while you are still contesting it. The uncertainty about whether any of this will change anything, which is its own kind of tax on the will to begin. Add these together and you have a number, denominated not in dollars but in effort, time, and nerve. That number is the price of being heard. And a price, once you see it as a price, is something a decider gets to set.

We are trained to read that friction as inefficiency — as the residue of bureaucracy, the thing a better-run office would sand down. Sometimes it is exactly that. But treating it only as sloppiness misses the more interesting and more troubling possibility: that the friction is doing work. That it is not the process failing to be smooth, but the process succeeding at something it will never name. The cost of appeal is a dial, and someone, somewhere, however unconsciously, has a hand on it.

Friction is a price

Start with the economics, because the economics are unusually clean here. An appeals process faces a cost curve: for any given design, there is some amount of effort a person must expend to lodge and pursue a contest. Where you set that curve determines who shows up. Set it low — a single clear form, a decision that arrives already explained, a channel that answers — and you will hear from nearly everyone with a genuine grievance, plus a few with none, the way any cheap good attracts some buyers who do not really want it. Set it high and you filter the pool. The filter is the point.

What the filter selects for is the thing to watch. Effort is not distributed evenly across the population of the wronged. The person with a flexible schedule, a printer, a working knowledge of how institutions answer, and the confidence that persistence pays — that person can absorb a high toll and still appeal. The person working an inflexible shift, without slack in the day or the budget, already worn down by the original decision, cannot. So the toll is regressive in the precise economic sense: the same nominal price consumes a far larger share of what the poorer objector has to spend. And here is the cruelty of the sorting. The people least able to pay the toll of contest correlate closely with the people least able to absorb the underlying wrong — the ones for whom a mistaken denial, a wrong charge, a false flag, does the most damage. The appeals process, priced high, sheds exactly the objections that matter most.

This is a kind of price discrimination, but run in reverse. The ordinary version charges each buyer as much as they are willing to pay, and captures more revenue by sorting them. This version charges each objector enough to select them out — not to extract their maximum, but to ensure that below a threshold of resource and resolve, they simply do not transact. The decider is not trying to monetize the appeal. The decider is trying to not receive it.

The appeal you priced away is the error you kept

Why would anyone want to not receive appeals? Because appeals are expensive to the decider in a way that has nothing to do with justice and everything to do with cost. Each contest that gets through is a demand on attention, a chance of reversal, a moment of scrutiny that might find something. A process that hears fewer of them books a real saving: fewer overturned decisions, less rework, less exposure. The toll booth on the appeal is, from the decider's side of the ledger, a cost-control measure. It works.

But the saving is an illusion bought at the system's own expense, because an appeal is not only a complaint. It is the feedback signal by which a decision system learns that it is wrong. A wrongly denied claim that is appealed and reversed tells the process something it could not otherwise know: that its rule, or its reading of the facts, or its model, produced an error here. Suppress the appeal and you do not make the error go away — you make the knowledge of the error go away. The mistake stays in the system, uncorrected, and the one mechanism that would have surfaced it has been priced out of reach. The decider saves the cost of the reversal and pays, invisibly, with its own accuracy.

This is where the quiet scandal lives. A system can be built to look appealable while being priced to be un-appealed. The right exists on paper; the channel is published; the form can be requested. And almost no one uses it, because the toll is set where it needs to be. Then the low appeal rate gets read — by the decider, by its overseers, by anyone glancing at the dashboard — as evidence of correctness. Few appeals, few errors. The inference feels natural and is exactly backwards. A low appeal rate can mean the decisions are sound. It can equally mean the decisions are wrong and the wronged cannot afford to say so. The metric cannot tell the two apart, and a decider who has set the toll deliberately knows which one it is looking at.

A system that counts its few appeals as proof of its accuracy has confused the height of its toll booth for the soundness of the road beneath it.

Pricing the appeal honestly

The corrective is not to abolish the cost of appeal, which cannot be done — someone always spends something to contest a decision, and a process with no friction at all invites the frivolous to drown the genuine. The corrective is to set the price deliberately, and to set it against the stakes rather than for the convenience of the decider. Where the consequence of an error is severe and the people most exposed to it are the least resourced, the toll on contest should be near zero: the decision arrives already carrying the reasons and the evidence behind it, the channel to object is short and open, and the effort demanded is proportionate to what a wronged person can actually muster. A Decision Receipt that hands the affected party the grounds of the decision has, among other things, lowered the toll — it spares them the first and largest cost of appeal, which is the labor of reconstructing what was even decided and why.

And the measurement has to change with the price. A process serious about accuracy does not count the appeals it received; it estimates the appeals it suppressed — the gap between how many contests a fair reading of its error rate should produce and how few actually arrive. That gap is the toll made visible. It is a hard number to get, but the discipline of trying to get it is what separates a system that wants to be right from one that only wants to look unchallenged. Standing is not merely the formal right to object; it is the real, priced-in ability to exercise the right without being sorted out for lacking the means.

Which is why the cost of the appeal is never an operations detail, however much it is filed under one. Someone chooses the length of the form and the hours of the line and the patience the wait demands, and that choice decides, in advance, whose objections the system will ever hear. It is a moral choice wearing the costume of a workflow. The height of the toll booth is a statement about who the road was built for — and a decider who wants to know whether its decisions are good has to first make sure it has not priced away the only people who could tell it they are not.

— Dispatches · Summit Cognitive