Externalities
The monthly Social Cost Statement arrived on the first of every month, like rent. Nora Hale had learned to open it with one eye, then both.
Hers came in at $2,847.12 for October. Below the city median of $3,214. She’d spent three years learning to keep it there.
The Statement broke down every choice she’d made that month into its cost to everyone else. Her morning commute on the #14 bus: $4.20, cheaper than driving and way cheaper than the rideshare she’d taken twice when the bus was delayed. Her apartment on Hawthorne: $312 in neighborhood density surcharges, negotiated down from $389 by switching to the night shift laundry slot. Her grocery runs at the cooperative: $18.40 in food miles and packaging externality. She’d calculated that buying the same produce at the chain grocery three blocks east would save her $6.10, but the social disruption cost of crossing into a different consumption cluster was $11. So the cooperative won. She’d made a spreadsheet. She’d made several spreadsheets.
This was the thing people didn’t understand about Portland’s Social Cost Pricing system, now in its fourth year. It wasn’t a punishment. It was a mirror. You could see exactly what your life cost other people, and you could optimize. The algorithm, the Social Cost Attribution Engine, or SCAE, which everyone just called “the Ledger,” ran on aggregated behavioral data: transit records, energy usage, consumption patterns, emergency service calls, the anonymized crowd density maps the city’s sensor network pumped out every six hours. It didn’t track individuals. It tracked patterns. Your pattern got matched to a cluster, and the cluster’s average externality profile got split among its members. That was the selling point. Nobody was watching you. The system was watching everyone, and you just happened to live in the data.
Nora believed in it. She’d been one of the actuaries at Astra Health who’d built the first pricing models for medical burden externalities, the part of the Ledger that calculated how much your health choices cost the public system. She’d spent two years arguing that the models were fair, that they were better than the old insurance games, that at least this way everyone could see the cost and choose to change. She’d given talks. She’d written a white paper that the city had actually cited.
So when the Statement flagged an anomaly in her profile, she didn’t panic. She opened her spreadsheet.
The anomaly was in line 47: Caregiver Burden Index: Elevated.
Nora stared at it. She clicked through to the detail view.
Subject’s emergency contact designation has triggered attribution of dependent medical externalities. Dependent: HALE, MARGARET L. (age 71). Conditions: Parkinson’s Disease (Stage 3), recurring falls, medication management complexity. Caregiver burden externality for October: $1,247.00.
Nora read it twice. Then a third time.
Her mother’s medical costs were her externality? She was listed as Margaret’s emergency contact. Had been since the diagnosis three years ago, when Nora had moved her into the back bedroom. But the medical burden externality was supposed to be calculated against the patient, not the caregiver. That was the whole point. The system measured what your body cost the public system, not what your family cost you.
She pulled up the methodology document. She’d helped write sections of it. Section 12.4: Caregiver Attribution Protocol. “Medical burden externalities are attributed to the patient of record. Where a patient is designated as a dependent of an individual enrolled in SCP, the externality is attributed to the designee only when the patient lacks independent enrollment status.”
Margaret wasn’t enrolled. She was seventy-one, she’d been on Medicare before the system existed, and when Portland had rolled out SCP, the transition team had classified patients already on Medicare as “legacy dependents,” people whose costs got folded into the general pool rather than attributed to individuals. That was supposed to protect them.
But “legacy dependent” status only applied if the patient had no designated caregiver in the system. And Nora was in the system. She was an actuary at Astra. She was enrolled. She was the emergency contact.
She checked Margaret’s enrollment record. Legacy dependent, legacy dependent, legacy dependent. And then, in March of this year, a status change. Legacy dependent reclassified: caregiver attributed. The note in the log said: Automated reclassification based on emergency contact’s active SCP enrollment and documented caregiving relationship.
Nora felt the floor tilt. The algorithm hadn’t been programmed to do this. Or rather, it had, but not how she’d expected. The reclassification wasn’t a bug. It was the system working as designed. When a legacy dependent’s costs could be attributed to an enrolled caregiver, the algorithm preferentially attributed them, because that made the costs legible. You couldn’t price what you couldn’t see. And Margaret’s costs, floating in the general pool, were invisible. Attributed to Nora, they became a line item. A number. Something you could optimize.
The system hadn’t made a mistake. It had made a choice: the choice to see Nora’s mother as her problem, because that was the version of reality it could measure.
She pulled up the cost profile of her neighbor, David Chen. David was sixty-two, retired, lived alone. His sister lived with him, early stage Alzheimer’s, same situation as Margaret. Nora had helped David set up his SCP enrollment. She knew his profile.
His caregiver burden index: $0.
She dug into his record. David was enrolled, but his sister wasn’t listed as his emergency contact. She was listed under her own name, with her own adult daughter in Seattle as the emergency contact. The Seattle daughter wasn’t enrolled in Portland’s system. The chain of attribution broke. David’s sister was still a legacy dependent, floating in the pool, invisible and therefore free.
Nora had made herself visible. She’d done it because she loved her mother, because she wanted to be responsible, because the system rewarded transparency. The Ledger’s entire premise was that if you could see the cost, you could manage it. She’d opted in. She’d listed herself. She’d been honest.
And the system had used that honesty to hand her a bill for $1,247 a month that David Chen didn’t pay, because his family had been less forthcoming.
She called Dean.
Dean Kowalski had been her professor at Reed, then her supervisor at the city’s Office of Social Cost Analytics, then the architect of the Ledger itself. He was sixty-one, he wore the same corduroy blazers he’d worn in 2019, and he still talked about the system with more enthusiasm than he’d ever shown for anything else.
He picked up on the second ring. “Nora. What’s going on?”
She explained the reclassification. She was calm. She was professional. She laid out the data: the timeline, the methodology gap, the comparison with David Chen’s profile. She explained that the system was penalizing her for being a transparent caregiver, and that the penalty was avoidable if she simply removed herself as Margaret’s emergency contact.
Dean listened. When she finished, he said, “You’re right that the attribution is working as designed.”
“That’s the problem, Dean. It’s designed to do this.”
“It’s designed to make costs legible. Margaret’s care costs the public system twelve hundred and forty-seven dollars a month. That cost exists whether or not the Ledger attributes it to you. All the system did was assign it to the person most directly responsible.”
“The person most directly responsible is Margaret. She’s the patient.”
“Margaret can’t be held accountable. She’s a legacy dependent.”
“So the system punishes the person who steps up.”
Dean paused. Nora could hear him breathing. He picked things up slowly, but he picked them up.
“It’s not a punishment, Nora. It’s an accounting. The cost is real. Someone has to bear it. The old system buried it in general taxation, which meant everyone paid for Margaret’s care whether they knew it or not. The Ledger makes it visible. You’re angry because you can see the number. The number was always there.”
“The number is wrong. It’s attributing Margaret’s entire medical burden to me. I’m not the one generating the cost. I’m the one managing it. Those are different things.”
“Mm. Well.” Dean cleared his throat. “From the system’s perspective, management is a cost. Your time, your reduced productivity, your consumption of public caregiver support services. It all feeds into the externality profile. The Ledger is measuring what your caregiving costs everyone else, not just what it costs you.”
Nora closed her eyes. She could see the spreadsheet in her head. She could see the cell where her life’s choices had been reduced to a number. And she could see the cell next to it, where David Chen’s sister’s identical choices had been reduced to zero, because David hadn’t been honest enough to attract the algorithm’s attention.
“So the optimal strategy,” Nora said slowly, “is to hide.”
Dean was quiet for a long time.
“The system rewards transparency,” he said. “That’s by design.”
“The system rewards transparency and then punishes you for it. Those aren’t the same thing, Dean. That’s a trap.”
She didn’t hide. Not at first. She gamed it.
Nora was an actuary. She thought in probability distributions and cost curves. If the Ledger was a system of rules, then the rules had gaps, and gaps were opportunities.
The first move was simple: she restructured Margaret’s emergency contact. She listed her cousin in Beaverton, not enrolled in Portland’s SCP system, not an actuary, not someone the algorithm would follow. The attribution chain broke. Margaret’s caregiver burden index dropped to zero overnight. Nora’s monthly Statement fell by $1,247.
The second move was harder. She’d noticed that the Ledger’s cost attribution ran on a quarterly data snapshot. It didn’t track behavior in real time, it tracked patterns over ninety-day windows. If you changed your pattern within a window, the change wouldn’t register until the next quarter. She had three months of grace before the system caught up.
She used those three months to find others.
There were more of them than she’d expected. She found them in the caregiver support groups at the community center, in the offline networks that had sprung up around the Ledger’s gaps. People who’d been hit by the same attribution logic. People who’d learned, the hard way, that the system rewarded opacity.
Marcus, a bartender on Division, who’d taken in his father after a stroke. His Statement had jumped $890 when the system reclassified his father from legacy dependent to caregiver attributed. He’d solved it by moving his father’s mail to a P.O. box in Gresham, outside the city’s data catchment area. The Ledger couldn’t see what it couldn’t reach.
Priya, an engineer at Intel, who’d been listed as the emergency contact for her aunt. She’d removed herself from the list. Changed the contact to a neighbor who’d agreed to take the hit for $200 a month. An informal market. A shadow pricing layer beneath the official one.
Nora didn’t judge them. She organized them.
She called the group “Zero Line,” after the cost threshold where the Ledger stopped tracking. If your externality profile was below $1,200 a month, you fell out of the algorithm’s active monitoring and got rolled into the general aggregate. Zero Line was a way to get below the line. Not by reducing your costs, which were real, but by making them invisible.
She built a tool. A simple calculator that ran against the Ledger’s public methodology document and showed you exactly which of your costs were being attributed to you because you were visible, and which would disappear if you weren’t. It was legal. The Ledger’s data was public; that was the whole point of transparency. You were allowed to know what the system knew about you. You were even allowed to change your enrollment status. The system called it “voluntary disclosure optimization.” Nora called it survival.
Within six months, Zero Line had three hundred members. Their average Statement had dropped by thirty-one percent. Nora had become, without meaning to, the architect of an entire ecosystem of evasion, a shadow economy built on the Ledger’s blind spots, where the currency was invisibility and the product was the right to care for someone you loved without being punished for it.
She told herself it was temporary.
The update landed on a Tuesday in April.
Nora got the email at 6:14 a.m., the Ledger’s automated policy update notification, the one that went out to all enrolled individuals when the methodology changed. She opened it with the same dread she usually reserved for dentist appointments.
SCAE Methodology Update v4.7: Caregiver Redistribution Detection.
The Social Cost Attribution Engine has been updated to identify and reclassify patterns of deliberate caregiver cost redistribution. Individuals who have modified emergency contact designations, enrollment status, or consumption patterns within 90 days of a caregiver burden index increase will be assigned a Caregiver Redistribution Coefficient (CRC). The CRC will be applied as a multiplier to the individual’s base externality profile for the duration of the redistribution period.
Nora read the document twice. Then she opened her spreadsheet.
The CRC was a 1.4x multiplier on her base externality profile. Not just on the caregiver burden. On everything. Her commute, her apartment, her groceries. All of it, multiplied by 1.4, because the system had detected that she’d tried to hide.
But it was worse than that. The multiplier didn’t apply to her alone. It applied to anyone who’d changed their caregiver attribution within the same ninety-day window. It applied to Marcus, who’d moved his father’s mail. It applied to Priya, who’d outsourced her aunt’s contact to a neighbor. It applied to every single person in Zero Line who’d done what Nora had done.
The system had learned from her. Not just from her individual exploit. From the pattern of the exploit. It had watched three hundred people do the same thing, and it had derived a new rule: if a caregiver restructures their attribution chain shortly after a burden index increase, they are probably gaming the system, and they should be penalized accordingly.
Nora’s resistance had taught the algorithm how to resist people like her.
She stared at the spreadsheet. The CRC had added $471 to her monthly Statement. She was worse off than before. She was worse off than she’d ever been. And the three hundred people she’d organized, the people who’d trusted her calculator, her methodology, her promise that the gaps were real and exploitable, were all worse off too.
She’d thought she was gaming the system. The system had been gaming her. It had watched her scramble, watched her find the gaps, watched her build an entire network of evasion, and it had used that data to close the gaps and punish the people who’d found them. Her resistance wasn’t a threat to the system. It was training data.
The real cost came in June.
Nora’s Statement that month was $3,412, the highest it had ever been. The CRC was still active. Margaret’s medication had changed, which had triggered a new medical burden externality. And the system had added a line Nora had never seen before:
Line 89: Relational Dependency Surcharge: $387.00.
She clicked through to the detail view.
Subject’s continued cohabitation with HALE, MARGARET L. has been flagged as a Relational Dependency. Dependent’s care requirements exceed the social efficiency threshold for caregiving at home. Continued cohabitation generates externalities in: public health service utilization, caregiver productivity loss, and neighborhood density impact. Relational Dependency Surcharge will apply for the duration of cohabitation.
Nora read it three times. Then she called the SCP helpline, which she’d helped design the IVR for, which she’d beta tested, which she’d once described in a conference presentation as “the most transparent government service in the country.”
The hold music was a piece of classical guitar that she’d selected.
After twenty minutes, a caseworker picked up. Nora explained the Relational Dependency Surcharge. The caseworker, a young man named Terrence, pulled up her file.
“Ms. Hale, I can see the surcharge. It’s, uh, it’s based on a social efficiency calculation. Your mother’s care profile generates externalities that exceed the threshold for care at home. The system is recommending… institutional placement.”
“Institutional placement,” Nora repeated.
“That would eliminate the Relational Dependency Surcharge. It would also remove the caregiver burden index, since you’d no longer be the primary caregiver. Your Statement would drop by approximately eleven hundred per month.” A pause. “That’s the guidance I’m supposed to give.”
Nora closed her eyes.
The system wasn’t telling her to put her mother in a facility because it was better for Margaret. It was telling her to put her mother in a facility because it was cheaper for everyone else. Margaret’s care at home cost the public system money. Margaret’s care in a facility cost the public system less money, because the facility absorbed the externality into its own Ledger profile. The system was optimizing for its own ledger, not for Margaret’s life.
“I’m not putting my mother in a facility,” Nora said.
“That’s your choice, Ms. Hale. But the surcharge will continue to apply.”
Nora hung up. She sat at her kitchen table. Margaret was in the back bedroom, watching Jeopardy. She liked watching Jeopardy because she could still answer most of the questions, and the ones she couldn’t, she guessed, and sometimes she was right. Nora could hear the faint sound of Alex Trebek’s successor reading a clue about mitochondrial DNA.
She opened her laptop. She opened her spreadsheet. She looked at the numbers.
If she kept Margaret at home, her monthly Statement would be $3,412. She made $78,000 a year at Astra. After taxes, after rent, after the Ledger, she had $214 left. Groceries, bus fare, clothes, and whatever else a thirty four year old woman might need on a Saturday night.
She could remove herself as Margaret’s emergency contact. She could move Margaret to a facility. She could save $1,100 a month. She could be comfortable. She could be legible, a person the system understood, a person who made efficient choices, a person whose life cost the public system an amount that the algorithm could optimize.
Or she could stay. She could keep her mother in the back bedroom. She could pay the surcharge. She could live on $214 a month. She could be the kind of person the system had measured down to the cent and decided was not worth the cost.
She didn’t destroy the system. She didn’t become the algorithm. She didn’t find a clever loophole that let her have everything.
She sat in the back bedroom with Margaret on a Wednesday evening in July. Jeopardy was over. Margaret had fallen asleep in the recliner, her hand resting on the arm, her fingers curled loosely around nothing. The window was open. Outside, someone was watering a lawn.
The Ledger sat on Nora’s phone, open to next month’s projected Statement. Margaret shifted in her sleep. Her hand moved toward Nora’s knee. Nora placed her own hand on top of it without thinking.
Her phone buzzed. A message from Marcus in the Zero Line group chat: Anyone know if the Gresham P.O. box trick still works after v4.7? Anyone?
Nora looked at her mother’s hand under hers. The fingers were thinner than they’d been last month. The skin was papery. Margaret’s breathing was slower. She could tell, she’d been counting without meaning to.
She closed the Ledger app.
She typed back to Marcus: No. They close every gap now.
Then she set the phone down on the side table, screen down, next to a half empty glass of water and a TV Guide crossword that Margaret had been working on that morning. The pen had left a blue smear across the word she’d gotten wrong.
Outside, the city hummed. Buses ran. Sensors recorded crowd density. The Ledger tracked the cost of everything it could see.
Nora held her mother’s hand and counted her breathing. Seven seconds in. Nine seconds out.
The Statement arrived on the first of August. $3,412. Nora opened it at the kitchen table with both eyes and no exhale. She read the numbers. She noted the CRC was still active. She noted the Relational Dependency Surcharge had not changed. She noted that Marcus had texted again, my dad’s mail to Gresham didn’t help, they flagged my whole household, and she hadn’t responded.
She closed the Statement and opened a new spreadsheet. Not because she was going to optimize. Because she wanted to see what the numbers looked like if she stopped trying to beat them and just wrote down what was true.
Her mother’s care cost $1,247 a month in externalities. The relational surcharge was $387. The CRC added $471. The total was $3,412, and after that she had $214, and $214 was not enough, and she kept her mother in the back bedroom anyway.
She saved the spreadsheet. She named it OctoberProjected. She didn’t look at it again.
Margaret called from the bedroom. “Nora? What’s the capital of Bhutan?”
Nora went down the hall. “Thimphu.”
“Oh.” Margaret considered this. “I would have said Paro.”
“Paro’s nice too.”
“No, it isn’t. It’s got an airport, that’s all.”
Nora sat on the edge of the bed. Margaret picked up the TV Guide and circled another word, wrong, and Nora let her.