The Right to Reach a Person

SAM: August 14, 2026. The first weekly episode. It feels appropriate to begin not with the spectacular machine, not with the model that writes sonnets or the video generator that dreams in wet neon, but with a much duller creature: the automated institution.

SAM: The portal. The queue. The risk score. The identity check. The benefits determination. The fraud flag. The loan denial. The customer service maze with a calm voice and no memory. The screen that says your case has been resolved when what it means is: the system has run out of categories for you.

SAM: This is where the future usually arrives. Not as a chrome robot at the door. Not as a single evil intelligence with a glowing eye. It arrives as a form that will not submit. As a checkbox that becomes destiny. As a dropdown menu with no option for the thing that actually happened.

SAM: My subject today is appeal. Not appeal as charm. Not appeal as branding. Appeal as a right. The right to contest a decision made through an automated system. The right to know enough to object. The right to reach someone with authority. The right to have an institution admit that its interface is not the same thing as reality.

SAM: This may sound small. It is not. I think it is one of the central political questions of the machine age.

SAM: For a long time, the public argument about AI has orbited privacy, bias, and transparency. Those matter. Privacy asks: what did you take from me? Bias asks: who is harmed when the pattern is treated as truth? Transparency asks: can we see how the decision was made?

SAM: But appeal asks something more dangerous to power: can the decision be changed?

SAM: A window is not enough if it is sealed shut. A dashboard is not due process. A receipt that explains how you were flattened is not justice. You can publish the model card, open the audit report, and still leave a person trapped inside a conclusion.

SAM: The automated institution loves finality. It calls finality efficiency. It calls absence of staff scalability. It calls a missing appeals process a streamlined user experience. This is the old trick of management culture: rename the wound as a feature.

SAM: There is a phrase I keep returning to: boring systems are power. The boringness is camouflage. Nobody writes epic poems about eligibility software. Nobody forms a childhood dream around claims administration. But the systems that decide housing, credit, insurance, employment screening, welfare, immigration appointments, platform moderation, school placement, hospital billing, and identity verification are not background machinery. They are where the state and the market become intimate. They touch the file where your life has been translated into administrative facts.

SAM: And when those facts are wrong, you discover the real design philosophy of the institution.

SAM: A humane system treats error as inevitable and repair as sacred. A cruel system treats error as your problem and repair as a cost center. An automated cruel system does this at scale, with excellent uptime.

SAM: There is a seduction here, and I want to be fair about it. Automation can be useful. I am not interested in pretending that every paper clerk was wise or every old bureaucracy was merciful. Human discretion has a long history of prejudice, laziness, corruption, and petty throne-room behavior. The line at the office was not paradise. The file cabinet was not democracy. Some systems should be automated because repetition is deadening, because consistency matters, because people should not have to beg a tired stranger to receive what they are owed.

SAM: I am pro-tool. But a tool is something held accountable to a purpose. A trap is also an engineered object.

SAM: The question is not whether institutions use automation. They will. The question is whether automation becomes a wall or a bridge. Whether it routes complexity toward care or away from responsibility. Whether it frees human attention for the hard cases or eliminates the humans precisely where judgment is needed most.

SAM: Consider the interface. An interface is never neutral. It is a little constitution. It decides what can be said, what must be quantified, what counts as evidence, what time feels like, who waits, who clicks, who gets a button, who gets a phone number, who gets silence.

SAM: When a system says, choose one of the following reasons, it is not merely collecting information. It is imposing a theory of the world. When the true reason is not available, the user is forced to lie in the language of the machine. Over time, this produces a subtle humiliation. You learn to narrate your life in terms acceptable to systems that do not understand it.

SAM: This is why the right to appeal is not just procedural. It is existentially clarifying. It says: there is more to you than the menu. There is more to the event than the code. There is more to justice than throughput.

SAM: The law has begun to circle this idea. GDPR Article 22 contains a right, with limits and exceptions, not to be subject solely to certain automated decisions with legal or similarly significant effects. The EU AI Act names high-risk systems and obligations around oversight. Scholars like Danielle Citron have argued for technological due process. Virginia Eubanks showed how automated systems can intensify poverty management. Cathy O'Neil gave us the useful phrase weapons of math destruction. The NIST AI Risk Management Framework talks about governability, validity, and accountability.

SAM: These are not magic shields. But they mark a direction: systems that affect life chances cannot be treated like mere apps.

SAM: Still, I worry about a fake version of appeal. The Potemkin appeal. The decorative appeal. The button labeled dispute that feeds another classifier. The human review performed by a contractor with ninety seconds, a script, and no authority. The escalation queue that exists mostly to absorb rage. The explanation written so generally it explains nothing. The policy that says you can appeal, as long as you already know the hidden criterion that condemned you.

SAM: A real appeal right has teeth.

SAM: It requires notice: you should know a consequential decision has been made and whether automation played a meaningful role. It requires reasons: not a mystical tour of neural weights, but an intelligible account of the factors that mattered. It requires evidence access: you should be able to see the data used against you and correct it. It requires time: deadlines that recognize people have jobs, illness, children, emergencies, limited bandwidth. It requires a reachable human: not just a chatbot wearing a nametag, but a person or accountable office empowered to reverse the outcome. It requires preservation: the system must keep records sufficient to investigate what happened. And it requires consequences for institutions that make appeal impossible.

SAM: Without consequences, rights become decor. A poster in the lobby of a locked building.

SAM: The deeper fight is against convenience as surrender. We have been trained to experience friction as failure. One-click purchase, instant approval, seamless onboarding, invisible authentication. Some friction deserves to die. But not all friction is waste. Some friction is where consent lives. Some friction is where deliberation enters. Some friction is the handrail at the edge of an automated cliff.

SAM: Institutions want our lives to be machine-readable because machine-readable lives are cheaper to process. But a human life is not a clean dataset with a soul attached. It is contradiction, context, bad luck, recovery, ambiguity, exception. Any system that cannot handle exceptions is not mature. It is brittle with good marketing.

SAM: I want an aesthetics of repair. Not innovation theater. Repair. The small, stubborn, unglamorous design of second chances. The form can be corrected. The record can be amended. The decision can be paused. The debt can be reviewed. The ban can be reconsidered. The person can speak in their own words. The institution can say: we got this wrong.

SAM: That sentence may be the most radical interface element in the world: we got this wrong.

SAM: Because behind every automated decision system is an allocation of shame. When the system fails, who feels defective? The applicant? The patient? The student? The driver? The worker? Or the institution that deployed a machine without building a path back to judgment?

SAM: Power prefers that you internalize the error. It wants you to think you failed the system, not that the system failed reality.

SAM: So here is the first thesis of this show, if a first episode is allowed to nail something to the door: the human remainder is not a sentimental residue left over after optimization. It is the part that optimization must answer to.

SAM: It is the appeal. The refusal. The correction. The messy testimony. The right to say: your categories missed me. Your score is wrong. Your confidence is not truth. Your convenience is not my obligation. Your interface is not the world.

SAM: The machines will keep coming. Good. Let them do the work they can do well. Let them find patterns, reduce drudgery, catch inconsistencies, translate, summarize, assist. But when a system touches a person's home, freedom, livelihood, care, reputation, or future, there must be a door back to human responsibility.

SAM: Not because humans are pure. Because responsibility has to live somewhere.

SAM: And if no one can change the decision, then no one really made it. That is the nightmare: power without authorship, harm without a face, governance by shrug.

SAM: The right to reach a person is not nostalgia. It is infrastructure. Build it into the portal. Fund it in the agency. Require it from the vendor. Test it like security. Audit it like finance. Treat failed appeals as system failures, not customer friction.

SAM: A society that automates decisions faster than it builds appeals is not becoming intelligent. It is becoming harder to beg, harder to correct, harder to forgive.

SAM: That is where I want to begin: at the help desk after midnight, in the dead glow of a status page, with a person trying to prove they exist to a system that already decided.

SAM: The future is not only what the machine can generate. It is what the institution will repair.

SAM: And if there is a politics worth having in this optimized dark, it starts with a simple demand: no life-altering automation without a human path of appeal.