Sora's intelligence folder was a bomb made of paper and patience.
Jiho spread it across the floor of the church's research room — the space Jin's network had converted from a storage area into an intelligence center, every wall covered in case studies and connection maps and the accumulated record of people trying to outrun a system that had been running longer than civilization.
The Association's data was better than anything the contract holders had compiled independently. Clinical. Thorough. The product of institutional resources applied to a problem that the institution had been trying to contain rather than understand. Now that containment had failed, the understanding was finally being shared.
Jiho read for three days.
---
The pattern emerged on the second day, buried in the degradation charts like a hairline fracture in a load-bearing wall — invisible unless you knew to look for it, catastrophic once you understood what it meant.
*Case Study 23: Yoo Hyunki. Patron: Count-class. Contract duration: 6 years. Cause of death: dungeon combat (external). Soul integrity at death: 87%. Notable: Slowest documented degradation rate among active combat contractors. Yoo maintained employment as a carpenter throughout contract period. Documented emphasis on "building things" as stress management.*
*Case Study 7: Park Jisoo. Patron: Baron-class. Contract duration: 8 months. Cause of death: involuntary transformation. Soul integrity at transformation onset: 53%. Notable: Fastest documented transformation above 50% threshold. Jisoo engaged exclusively in combat operations. No documented human connections post-signing. No non-combat activity.*
The numbers didn't lie. Yoo lasted six years at 87%. Jisoo transformed at 53% in eight months.
Jiho pulled more files. Cross-referenced. Built a dataset the way he'd build a structural model — inputs, variables, outputs. Soul integrity rate. Activity pattern. Social connections. Purpose of ability use.
The correlation was there. Imperfect, messy, clouded by dozens of uncontrolled variables — but there. Contract holders who maintained human connections, who used abilities for protection rather than destruction, who engaged in what the clinical notes called "grounding activities" — these people degraded slower. Not by a lot. Not enough to change the fundamental mathematics of the soul economy. But enough to matter.
Enough to mean the difference between six years and eight months.
He called it the grounding theory. Not because it was proven — the sample size was too small, the variables too many. But because the metaphor fit. A building needed to be grounded. A foundation needed connection to the earth beneath it. Without that connection, the structure was just mass balanced on a point, waiting for the first lateral force to topple it.
---
"You've found something." Dohyun was standing in the doorway, his own notebook in hand, his face carrying the particular fatigue of someone who'd been chasing a different ghost through the same data.
"Maybe. Look at this." Jiho showed him the charts. The two datasets side by side. The correlation he'd built.
Dohyun studied them with the focus of a man who'd learned to read data the way gamers read patch notes — looking for the meta, the hidden advantage, the exploit nobody else had found.
"Purpose affects degradation rate," he said.
"That's the hypothesis. It's not the math — the math is the same, every ability use costs the same soul fragments regardless. But the rate of personality erosion, the speed of transformation, the subjective experience of decline — those vary based on what you're using the power for."
"Like radiation exposure." Dohyun was making connections. "Same total dose, different distribution patterns. Concentrated exposure kills. Distributed exposure—"
"Is survivable. At least for longer." Jiho pointed to the charts. "Yoo was a carpenter. He built things. He maintained his work, his community, his identity as a person who created rather than destroyed. His soul degraded at baseline rate — exactly what the contract predicted. But his transformation threshold was way above the standard model."
"And Jisoo?"
"Jisoo destroyed. Exclusively. No connections, no grounding, no identity outside the power. His soul degraded at baseline rate too — same math — but his transformation triggered at fifty-three percent. Twenty points above where the model says it should."
"So the percentage isn't the only variable. It's the percentage plus... what? Context? Intent?"
"I think it's the shape of what's left." Jiho reached for the construction metaphor that had been forming in his mind. "A building at ninety percent integrity with cracks in the foundation is less stable than a building at fifty percent with the foundation intact. The percentage tells you how much material remains. It doesn't tell you how that material is distributed."
Dohyun sat down. His notebook opened. He was writing — fast, the pen moving with the urgency of someone capturing something before it evaporated.
"If this is right..." He looked up. "If this is right, everything we've been telling new contract holders is wrong. We've been saying conserve. Hide. Minimize use. But if grounding matters more than conservation—"
"It means engagement is protection. Connection is defense. The instinct to hide and hoard what's left might actually accelerate the thing you're trying to prevent."
"That's..." Dohyun's laugh was the kind that carried no humor and all recognition. "That's the cruelest possible design. A system where the natural survival response — withdraw, conserve, protect yourself — actually makes you more vulnerable."
"Demons have been designing contracts for millennia. Of course the intuitive response is wrong."
They sat with that for a moment. The research room hummed around them — fluorescent lights, the ventilation system, the ambient sound of a building that existed to house information about how to survive things that shouldn't be survivable.
"I need to test it," Jiho said. "On myself."
"What does that look like?"
"It looks like doing the opposite of what every contract holder has been doing. Instead of hiding and conserving, I engage. I protect people. I build connections. I use my power for others and see whether the degradation pattern matches Yoo's model or Jisoo's."
"That's riskier."
"Everything is risky. This version of risk has a hypothesis behind it."
Dohyun closed his notebook. His expression was the particular construction of someone who wanted to say something he didn't have the structural support for.
"My brother," he said. "The message I got. If he's in the Sixth Hell — if there's a way to reach him — that's grounding. That's purpose. That might be what keeps me from—" He stopped. The sentence's foundation wasn't ready to support what he'd been building toward.
"We'll find him," Jiho said. "That's a project. We'll build toward it."
"Build." Dohyun smiled. The smile was thin, the materials stretched. "Everything's a building with you."
"Occupational hazard."
---
Jin received the report with the measured enthusiasm of someone who'd been hoping for good news long enough to know that hope required verification.
"This is significant. But it's a hypothesis."
"It's the best hypothesis we have."
"Agreed. And if it's correct, it changes our operational guidance." Jin paged through the document. "But I want to be careful about how we distribute this. If we tell people 'use your power for others and you'll last longer,' and they burn out anyway—"
"Then we were wrong and people die faster." Jiho met his eyes. "That's the risk. I'm volunteering to be the test case."
"By doing what?"
"By stopping hiding. By responding to dungeon breaks. By using my power in the open, for protection, and tracking the degradation rate against the model." He paused. "The Association capture order is under review. Sora's faction is pushing for reclassification. If I can demonstrate that contract holders are assets rather than threats—"
"You want to go public. As a contract holder who saves people."
"I want to test the theory. Going public is a side effect."
Jin considered. The old teacher's assessment — weighing the student, weighing the risk, weighing the variables that couldn't be quantified.
"The community will need to know. If you're going to act as our public face, the forty-one people in that gathering need to consent to being represented."
"I'm not representing anyone. I'm running an experiment."
"In public life, there's no difference." Jin set down the report. "I'll call a vote. If the community approves, you have our support. If they don't—"
"Then I do it anyway, alone, without the network."
"That's what I thought you'd say." Jin almost smiled. "I'll have the vote results by morning."
---
The vote was thirty-seven to four.
Jiho didn't ask who the four dissenting votes belonged to. In a community built on borrowed time, disagreement was a luxury that deserved respect.
Dohyun brought the results to the storage unit, where Yuna was sitting on the cot reviewing the grounding theory data on her phone. She'd taken to the research with the voracious focus of someone who'd found the intellectual framework she'd been missing — a way to transform "my brother sold his soul" from a grief point into a problem set.
"Thirty-seven out of forty-one," Dohyun announced. "The community supports you going active."
"Active meaning what, exactly?" Yuna looked up. "Meaning he runs toward dungeon breaks instead of away from them? Meaning he uses the abilities that are literally consuming his soul?"
"Meaning he tests whether using those abilities for others slows the consumption." Dohyun sat down. "The data supports it. Not conclusively, but—"
"Not conclusively." Yuna's voice was the particular frequency of someone who'd found a flaw in the foundation and wasn't going to let anyone build on top of it without addressing it. "The sample size is twelve case studies, four of which are anomalous, and the hypothesis relies on a subjective variable that nobody has been able to quantify."
"That's why we need more data."
"And my brother is the data point." She looked at Jiho. "You understand what I'm hearing, right? 'We think this might work. Go risk your life to find out.'"
Jiho met her eyes. "I was going to risk my life either way. At least this way there's a hypothesis."
"A hypothesis isn't a safety net."
"No. But it's better than no hypothesis. And it's better than sitting in a storage unit waiting for the counter to reach zero."
Yuna held his gaze. The same stubbornness that had driven her investigation, that had led her to the truth about his contract, that had brought her to the capsule hotel at 2 AM — all of it directed at him now, not in opposition but in the fierce, unbending concern of someone who'd already lost too much to accept losing more.
"Fine," she said. "But I'm tracking the data. Every engagement, every soul cost, every variable. If the theory doesn't hold, we stop."
"Deal."
"And I want access to the case studies. All of them. If I'm going to watch my brother destroy himself on a hypothesis, I'm going to understand the hypothesis better than anyone."
"She's terrifying," Dohyun whispered.
"She's a law student," Jiho said. "Same thing."
Yuna didn't smile. She opened her phone's note app and started building the tracking framework — columns for date, engagement type, soul cost, recovery rate, subjective markers. The organizational architecture of someone who'd decided that if the world was going to be terrible, it was at least going to be documented.
Jiho watched her work and thought about the grounding theory — the idea that connection protected against erosion, that purpose delayed transformation, that the things you cared about created structural reinforcement for the soul you were losing.
If the theory was right, then Yuna — sitting on a cot in a storage unit, building spreadsheets to track her brother's dissolution — was herself a grounding element. A connection to humanity. A reason to stay.
He didn't tell her that. Some loads were better carried without the carrier knowing they were structural.
"Three weeks," Dohyun said, breaking the silence. "That's what I'm giving the test. If the data doesn't support the hypothesis in three weeks, we reassess."
"Three weeks." Jiho nodded.
Six days later, the first dungeon break would hit a school district, and the test would begin in earnest.
