Chapter 78: Fragment Work

Fourteen days wasn't nothing, but it wasn't the ninety days they'd been planning around.

The expedited review board would convene in thirteen days now, and Acharya was working through the documentation verification with the methodical urgency of someone who understood that her case had just changed and the change was both better and more dangerous. Better because the Sera Vance documentation was, if authenticated, devastating. More dangerous because filing it meant a direct attack on the Association's historical framework—which meant the Association's response would be proportionate to the threat.

He ran the fragment acquisition in parallel.

Not because the expedited review timeline gave him more free time. Because stopping fragment work during a crisis was the error pattern Gareth had been training him out of: the tendency to go narrow when everything was urgent, to stop doing the foundational work that kept the Harmony growing. The Fragment Harmony at 101 was better than at 100. At 102 it would be better still. Marginal gains that compounded.

Yuki had flagged a practitioner with an unusual class two weeks ago—before the Perfect One's ceasefire, before the expedited review, before any of it. She'd flagged it as a potential contact, not a fragment opportunity. The practitioner was a Level 38 Space-Time Convergence Mage named Eo, operating out of the eastern Second District, unaffiliated, somewhat reclusive, known in the information market as someone who occasionally consulted on spatial-temporal anomaly problems for private clients.

He looked at the class name and thought about what Gareth had found in the Harmony's decompression work—the superposition mechanism, the evaluation cycle's dual-state operation. He thought about what the Perfect One's analysis had said about the superposition's control architecture.

Space-Time Convergence was a hybrid class between Space Mage and Chronomancer. Not either of those—something synthesized. The Fragment Harmony already held fragments of both source classes. Adding a convergence fragment would either integrate smoothly with the existing architecture or—

Or it would create pressure at the integration points.

He'd talked to Gareth about the risk before leaving.

"The Chronomancer-Space Mage junction," Gareth had said. He'd looked at the oscilloscope with the measuring expression. "The orientation interface you developed runs along that junction. Adding a fragment of a class that synthesizes both source classes might enhance the junction's architecture—or it might create competition for the regulation layer's integration resources." He'd paused. "There's no way to know which until the fragment is retained."

"Acceptable risk?"

"Manageable risk." He'd met Damien's eyes. "Different. The fragment's retention is irreversible. If the competition is significant, the orientation interface could experience interference until the Harmony resolves the integration conflict."

"How long to resolve."

"Unknown. I've never seen the Harmony integrate a convergence fragment of a junction class." He'd paused. "If the prior integration pattern holds—two to four days."

He'd gone anyway.

---

Eo lived in an apartment building above a Fourth District tea shop.

Not the Fourth District core—three streets over, in the residential section that had developed around the core's presence, the kind of quiet neighborhood that forms when people are comfortable near something old and inert. The building was unremarkable. The apartment door opened before he knocked.

"Yuki sent you," Eo said. Not a question.

"Yes."

She was older than he'd expected—mid-fifties, with the calm posture of someone who'd been living with an unusual ability for long enough that unusual wasn't the operative word anymore. Her field was—complex. Not the clean single-frequency signature of most class practitioners. It had multiple layers, the Space component and the Time component running in a braid that didn't quite resolve into either source class.

"You want the fragment," she said.

"If you're willing."

"Yuki said you'd be useful to know." She studied him with the calm attention of someone reading field signatures. "Fragment Harmony. I've read about the theoretical architecture. I didn't know one existed." She looked at his mana field. "That's—substantial."

"Thank you."

"That wasn't a compliment." She kept looking at his field. "It was an observation." She stepped back from the door. "Come in."

He came in.

The apartment had the comfortable disorganization of someone who'd been working on the same problem for decades: stacked papers at every horizontal surface, a worktable with spatial diagram overlays spread across it, a bookshelf that had given up organizing itself and was now just containing things. It smelled like tea and old paper.

She made tea—actually made it, put the kettle on, got cups—and he sat at the cleared end of the worktable and looked at the spatial diagrams while the kettle ran.

"The convergence problem," she said from the kitchen end. "Space-Time Convergence is the synthesis of two classes that shouldn't coexist. Space magic operates on spatial probability arrays. Chronomancy operates on temporal probability arrays. The two arrays overlap at certain decision-point nodes, and that overlap is where Convergence operates." She poured the tea. "The problem is that the overlap is small and the resolution is unstable. Convergence practitioners spend most of their careers just trying to keep the synthesis coherent."

"What do you use it for."

"Consulting, primarily. When someone has a spatial-temporal anomaly problem that pure Space Mage or pure Chronomancer approaches can't address—they come to me." She brought the cups. "The anomaly problems are interesting. The clients are usually not." She sat across from him. "Why do you want a fragment of a class this unstable."

"The Harmony has both source class fragments," he said. "The synthesis might help with a specific architectural function I'm developing."

"Or it might create a conflict at your existing integration junctions." She looked at his field. "If your Chronomancer and Space Mage fragments are running in a junction—and a convergence fragment comes in—the junction's regulation architecture has to decide which integration paradigm takes precedence." She paused. "Convergence's regulation paradigm isn't compatible with either source class's individual paradigm. It's its own thing."

"I know."

She looked at him. "And you're doing it anyway."

"Yes."

She picked up her tea. "The reason I'm willing to talk to Yuki's people at all," she said, "is that I've been working a spatial-temporal superposition problem for six years. A client originally. The client is gone—the problem is still there." She held the cup. "I've gotten close. The superposition function requires a dual-state stability mechanism that I don't have in the convergence class's baseline architecture." She looked at his field. "But a Fragment Harmony that contains a Convergence fragment, a Space fragment, and a Chronomancer fragment—and that has a Harmony-level integration layer—"

He looked at her.

"You can read the Harmony's architecture," he said.

"Superficially. The surface presentation." She met his eyes. "I can't read the deep architecture. But the orientation interface function you've developed at the Chronomancer-Space Mage junction—if the junction holds a Convergence fragment—the interface's access to the superposition window might extend." She paused. "Which is useful for your work and for mine."

He thought about the Perfect One's analysis. The evaluation cycle. The superposition's millisecond duration and the control architecture that determined its length.

"The dual-state stability mechanism," he said.

"Yes." She looked at his field. "What I've developed—" She turned and pulled a set of diagrams from the worktable. "This is the stability framework I've been building. It doesn't work in the Convergence class's native architecture because the class can't hold both states simultaneously at the required frequency. The conflict resolution is too slow." She spread the diagrams. "But a Fragment Harmony with a proper integration layer—the Harmony's regulation architecture runs the conflict resolution passively. It doesn't need the class's native mechanism."

He looked at the diagrams.

The dual-state stability mechanism. Six years of work by someone who'd been trying to solve the same problem from a different direction.

"Share the framework," he said. "The full technical documentation. And I share the Harmony survey data—what we know about the superposition mechanism from the inside."

She looked at him.

"You're using the fragment as the medium," she said. "The junction architecture carries information in both directions."

"That's the theory."

She looked at the diagrams. She looked at his field. She looked at the cup in her hands.

"The class shift," she said. "It's voluntary. I have to agree to the shift."

"Yes."

"And what I keep—"

"Ten percent. The regulation layer takes the fragment at shift completion. The base class level resets."

She looked at the diagrams. Six years.

"If the framework works," she said, "what happens to the problem you're solving."

"If the framework works," he said carefully, "a significant number of people whose classes were lost to an absorption event might have a pathway to restoration."

She looked at him.

He held her eyes. He wasn't going to oversell it—the mechanism was still theoretical, the study was still in progress, there were multiple steps between now and anything functional. But it was real. The direction was real.

She put down the cup.

"Ready when you are," she said.

---

The shift lasted four seconds.

He came out of it with the Convergence fragment settling into the Harmony's architecture and the regulation layer—running faster than it had in any previous shift—working the integration with the immediate focused intensity of a system encountering a foreign element that didn't behave the way the existing integration patterns expected.

"Gareth," he said. His earpiece was on.

"I see it," Gareth said. He was at the monitoring station. "The fragment is integrating. The regulation layer load is—" A pause. "Elevated. Significantly." Another pause. "The Chronomancer-Space Mage junction is showing competition pressure. The Convergence fragment's integration paradigm is creating a conflict with both source class fragments simultaneously."

"The orientation interface."

"Running at—" A pause. "Eighty-three percent of nominal function. The competition is drawing regulation resources from the interface's operation." He was quiet. "This isn't critical. It's manageable. But the interface is not running clean."

He sat in Eo's apartment and felt the Harmony doing the work of resolving something it had never encountered before. Eo was looking at his field with the professional attention of someone watching an interesting problem solve itself.

"The conflict resolution timeline," he said.

"At current regulation load—" Gareth calculated. "I don't have a confident estimate. The pattern doesn't match prior integrations." He paused. "It might be two days. It might be more. The convergence integration is—different."

"Is the interface still functional."

"At eighty-three percent—functional. Reduced. The activation under pressure will still occur. The precision is degraded." A pause. "If an engagement were required in the next forty-eight hours—the orientation read would still run, but the two-point-four second window might be shorter."

He looked at Eo. She was sketching something on a notepad—his field's response pattern, probably. Documenting.

"Keep monitoring," he said to Gareth.

"Yes." A pause. "Come back when you can. I want the full read before the integration settles."

---

He spent two hours with Eo sharing the Harmony survey data—the translation function, the evaluation cycle, the superposition mechanism, what the Perfect One had identified about the control architecture. She shared her stability framework documentation in return. The exchange was efficient in the way efficient exchanges between technically fluent people usually are: specific questions, specific answers, no time spent on framing.

At the end of two hours she said: "The Harmony's internal decompression key—the evaluation cycle's control mechanism. That's the element my stability framework is missing. If the Fragment Harmony's evaluation cycle can be directed to run at an extended frequency—sustained superposition rather than millisecond evaluation—the dual-state stability problem resolves." She looked at the diagrams. "The Convergence fragment, properly integrated, should give you access to the extended frequency function. Once the integration conflict resolves."

"Two to four days."

"Maybe more." She looked at his field. "The conflict is complex. The Harmony is running the resolution, but it's not running it fast." She paused. "It's being careful."

He looked at her.

"The Harmony is careful?" he said.

She pointed at the surface field presentation. "Watch the regulation layer's oscillation pattern." He looked. She was right—the layer wasn't driving through the conflict resolution. It was probing. Testing. Running small integration attempts and backing off and trying different approaches. "It knows this is complicated," she said. "It's not forcing it."

He looked at his own Harmony doing something that looked, in the specific way that mechanical processes occasionally did, like caution.

"The Fragment Harmony learns," he said.

"Yes." She made a note. "Which means it's already working on the solution." She looked at him. "Give it time."

---

Acharya's verification came at four PM.

*The override code authenticates. Director Marcus Holt. The code is in the Association's historical registry under his credentials.* A pause. *The modification timestamp matches the date range from the original investigation. This is real.*

He read the message twice.

*The forty-year incident's original investigation concluded Class Absorption entity as responsible,* Acharya continued. *That conclusion was overwritten by a Director override forty years ago. I have the authentication.* A longer pause. *This changes the case. The monitoring provision's evidentiary basis includes the forty-year incident as precedent for the "demonstrated harm" standard. If the forty-year incident was misattributed—the demonstrated harm standard loses its primary historical anchor.* She paused again. *I need to file this before the expedited review board convenes. I can have the counter-documentation ready in five days.*

He sent back: *Five days. Do it.*

Then he sent to Maya: *Verification confirmed. Five days.*

Her response was immediate: *Good. How's the fragment.*

He looked at the Harmony's regulation layer running its careful resolution work at the integration junction.

*The Convergence fragment is in. There's a conflict. Manageable.*

*Gareth says "manageable" or you say "manageable?"*

He thought about that. *Both.*

A pause. *Come home.*

He came home.

---

At nine PM, Gareth ran the updated survey.

The conflict was real. The Chronomancer-Space Mage junction was running at elevated load. The orientation interface was at eighty-one percent of nominal—down two points since the afternoon. The regulation layer's resolution pattern was patient and methodical.

"It's not getting worse," Gareth said. He was reading the oscilloscope. "But it's not getting better quickly either. The Convergence fragment's integration requires the Harmony to develop a new paradigm for the junction—not just accommodate the fragment, but revise the existing junction architecture to support the three-source synthesis."

"How long."

"Days." He paused. "The interface will be reduced for the duration." He looked at Damien. "If the expedited review requires an engagement response—"

"The interface runs at eighty percent. That's still functional."

"Functional. Not optimal." Gareth met his eyes. "The precision degradation at eighty percent—the two-point-four second window may reduce to two seconds. In a fast engagement—"

"I know."

Gareth looked at him. "Was the Convergence fragment worth the conflict window."

He thought about the dual-state stability framework. About Eo's six years of work and what it contributed to the restoration mechanism. About what the Perfect One's analysis would do with the combined data.

"Yes," he said.

Gareth held his eyes for a moment. Then: "The extended frequency function. When the integration resolves and the Convergence fragment's synthesis becomes stable—what will it do to the Harmony's superposition capability?"

"Extended evaluation window," he said. "Sustained dual-state operation. The Harmony could hold both an active class configuration and a decompressed alternative simultaneously—for longer than milliseconds."

Gareth looked at the oscilloscope display.

"How much longer," he said.

"Eo thinks the stability framework could support duration at the second scale. Possibly longer with the Harmony's adaptive architecture." He met Gareth's eyes. "If we can direct the superposition—if the Perfect One's analysis identifies the control mechanism—"

"You'd be able to run both an active fragment set and a decompressed class configuration simultaneously." He paused. "What would that feel like."

He thought about it. "Like having two instruments you can play at once, instead of switching between them." He looked at the oscilloscope. "Not simultaneously in the sense of fragmenting—coherently. Both fully operational."

Gareth was very still.

"That's not a generalist improvement," he said slowly. "That's something else entirely."

"Yes."

"That's not someone who can do everything adequately." He met Damien's eyes. "That's someone who can be two things completely at once."

He looked at the oscilloscope. At the Harmony doing its careful work at the junction. At the counter running in the corner of the display.

[Fragments: 102 / 1000]

[Fragment Harmony: OPERATIONAL — Junction Integration Conflict / Orientation Interface: 81%]

[Translation Architecture: UNDER STUDY — Day 8 — Dual-state stability framework acquired]