Running the Diagnostic

The sequencing review came back at 0647, forty minutes before Greel's debrief.

Vasht's message was two sentences: Third candidate moved to first extraction slot. Operational rationale accepted. Revised timeline attached.

I read it twice. Then I read the revised timeline. The geographic reshuffling pushed the first candidate back by three days and put the third candidate's extraction window at day eight instead of day twelve. Eight days to move, versus Dominion contact projected at nine to eleven days out, depending on which routing estimate Vasht's intelligence team trusted.

One to three days of margin. That was what a specific ask, framed correctly, had bought.

The hub core hummed. I pulled up the candidates' files again and started cross-referencing the revised timeline against Sonn's development projection data, not because I expected the numbers to change but because walking through the math was the only way to identify gaps before someone else identified them in the middle of an operation.

The third candidate's profile said nineteen years old, rural southern hemisphere, biological survey compatibility score noted as exceptional and the assessor's language around it had been careful in the same way Sonn's language was careful when the data exceeded the notation system's comfortable range. The development rate projection was Sonn's conservative estimate, and Sonn built conservative estimates by halving the observed rate and adding a correction factor for unknown variables. Even halved and corrected, the projection was faster than mine had been at any point in the first four months.

Rovan came back from the perimeter check at 0700. He had not slept but he did not look like he was working against it; he was one of those people whose physical baseline was high enough that a missed night registered as fatigue rather than impairment. He checked the relay panel, logged the surveillance signal update, and then looked at me over the equipment rack.

"She moved it north overnight," he said. "Three kilometers east, about two hundred meters north of last position. Still passive."

"She is mapping the perimeter from outside useful range."

"Yes." He pulled a field ration from the storage ledge and opened it without particular enthusiasm. "She will not push again before she has a complete picture."

"No." I filed the surveillance position update in the monitoring log and sent it to Vasht's relay automatically. "She has time. She knows the convoy reinforced us. She is not losing the initiative by waiting."

Rovan ate standing up, watching the relay panel. "The first extraction," he said. "You are not on that one."

"Correct."

"Standard Compact contact team."

"With Sonn remote monitoring."

He nodded slowly. The first candidate was seventeen, urban, compatibility score in the high-range but not the exceptional range. Standard contact procedure was designed for exactly that profile: extract before Dominion contact, calibrate, begin supervised integration in a controlled environment. The procedure worked. Sonn had seen it work across multiple species integration cases.

The third candidate was not that profile, and standard procedure was not designed for someone whose development rate would mean measurable integration progress inside the first forty-eight hours, in conditions that could not be fully controlled, with a Dominion operation that had already been moving toward the contact zone.

"You will need to explain the integration to them before they decide," Rovan said. It was not a question. He had been on the relay when Denavi had made the same observation the previous night, and unlike Hetch, who had processed information quickly and moved on, Rovan tended to come back to a thread when he had thought it through to a conclusion.

"Yes."

"You are the only person in the Compact's current operational structure who can do that from inside the experience."

"Vasht can present the technical case. Sonn can present the safety parameters. Those are both useful." I set down the candidates' files. "But Vasht has never run the predictive extension through a three-hour assault and Sonn has never run it at all. The part that is hard to explain from the outside is what it is and what it does when the system is under full load and there is no option to step back from it."

Rovan considered this. "And what does it do?"

I looked at him.

"I am not going to brief against you," he said, practical about it. "I am asking because if you are going to explain it to a nineteen-year-old who has never heard of the Aeluun system, you will need to explain it in language that does not require existing context. And it is useful to know whether you can do that before you are standing in front of them."

That was a fair point, delivered with the same quiet precision he brought to everything operational. I thought about it for a moment.

"It is like hunting," I said. "When you have been in the field long enough and you know the terrain well enough, you stop consciously tracking every variable and the picture just builds. You know the deer is going to move left before it moves left because you know the wind and the cover and the way it was holding its weight. You are not running a calculation. The information is just there. The integration amplifies that. It takes whatever your brain was already doing with spatial and movement data and it runs it faster and further forward. And after a certain point, it does not stop running. It is on whether you want it or not. That has costs."

Rovan was quiet for a moment. "That is a good explanation."

"It is the honest one."

He finished the field ration and folded the packaging with the same methodical precision he brought to the relay panel. "The cost. Will you tell them that part first or last?"

"First," I said. "Vasht led with the necessity when she briefed me. It was the correct order strategically and it was the wrong order for getting the full picture. I did not understand what I was agreeing to until I was already committed. There is not a way to completely avoid that sequence, because some of what the integration costs you only becomes visible through the experience of it. But you can tell someone the costs you can name, clearly, before you ask them to make a decision. That way the decision is theirs."

The debrief was at 0900. I had ninety minutes.

I pulled up Sonn's analysis and drafted the capability summary Rovan had asked about the night before: range, load limits, degradation indicators, the predictive extension's relationship to the baseline pattern recognition. No mechanism, no development projections, those stayed in Sonn's analysis chain. Just the operational parameters the unit needed to work alongside someone running this integration in a live engagement.

It took twenty minutes to write something that was accurate without requiring context that the unit did not have.

Rovan read it over my shoulder without asking permission, which was the kind of thing he did naturally and which did not bother me because his instinct for what was operationally relevant was good and his corrections were useful when he made them.

He made one: "The range note. You say fifty to sixty meters for reliable predictive modeling. The assault last night, you were calling movement vectors at what Denavi and I estimated as eighty meters, maybe more."

"That was with the extension at full load and under the interference from Eshet's signature. The sixty-meter figure is the clean reliable range without interference. Under full load it extends, but the accuracy degrades at the edges and I cannot always tell the difference in real time between a reliable read and a projection error."

"So sixty meters is the number the unit should plan around."

"Yes."

He went back to the relay panel. I finished the capability summary and flagged it to Greel's debrief relay and Vasht's communication chain simultaneously.

The tone in my right ear ran steady.

Outside the north access, the sky over the Caleth Corridor was the flat grey of a sector that processed heavy industrial material at a scale that had deposited a permanent particulate layer across the lower atmosphere. Not a natural grey. The color of a place that had been used hard for a long time. The convoy's vehicles were visible from the access point, parked in the node's exterior lot in the orderly rows of a supply operation that had just arrived and not yet settled into the rhythms of a sustained position.

New soldiers at the perimeter. People who had not been here for the seventeen hours of the assault and who were reading the position fresh.

I walked the perimeter once before the debrief, not for surveillance purposes but because the habit was established and because the extension's passive monitoring gave me a clear picture of the sector's current state when I was moving through it rather than stationary. The surveillance signal was still east, still passive, about two kilometers now, Eshet having moved it overnight in a slow arc that kept it at the edge of useful range from my position.

She was patient. She had eleven years of operational experience and she had gone up against someone with eighteen months and come out of the assault with her forces intact, her logistics intact, and her information-gathering operation still functional. The Caleth Corridor was still contested. The node's reinforcement made immediate seizure impractical but it did not remove the corridor from the operational equation, and Eshet would know that as well as anyone.

She was going to come back. The question was when and in what configuration.

Durrik was at the perimeter's east position, standing with one of the convoy's officers in a conversation that had the body language of a briefing rather than an exchange: the convoy officer was talking, Durrik was listening with his weight centered and his attention on the middle distance, the posture of someone receiving information they had already partially assessed and were comparing against a prior model. He saw me pass and nodded once. The convoy officer did not look up.

I went back inside and found Denavi already at the debrief table, her notes stacked in a specific order she used when she was preparing to give a structured account. She had slept four hours and it showed in the quality of her stillness: she was rested and ready, not running on managed fatigue.

Greel's medical team had set up a relay link. Greel himself was in the forward medical station four kilometers back, his condition stable and his arm in a cast that the medical officer had described to his relief team as a clean fracture, straightforward repair. He was not going to be operational for six weeks. The debrief was going to be the first time the unit heard from him since the assault's third hour.

The link opened at 0901. Greel's face on the relay panel was the color of someone who had been through a medical team's care, which was not a good color, but his voice was exact.

"Node status," he said.

Rovan gave it: supply connections intact, perimeter reinforced, surveillance signal at two kilometers east, passive, logged and relayed to command chain. Convoy integration proceeding per standard staging protocol.

"Casualties," Greel said.

Rovan gave that too. Greel listened without interrupting. When Rovan reached Hetch, Greel's face did not change but he was quiet for three seconds before he said, "Continue."

Denavi gave the tactical account. She was precise and complete and she attributed decisions to the people who made them, including two decisions that were mine, with the same flat informational tone she used for everything else.

When she finished, Greel was quiet for a moment. "Caulfield," he said.

"Here."

"The gap," he said. "The north breach. Walk me through your read."

I walked him through it: the predictive extension's coverage of the breach vector, Eshet's interference degrading the precision of the forward model, the decision to move through the gap rather than hold the rear position, and the reasoning behind "count six" as the relay to Denavi. He listened without interrupting.

"What did you see in her pattern that gave you the cycle timing?" he said.

"The gifted soldiers were repositioning on her signal, not on independent assessment. That means the cycle was her decision interval, not the terrain's natural movement pace. I could read her repositions through the extension even with the interference because her integration signature is dense enough to be visible at range even when it disrupts the precision of the predictive model. Every time she moved, they moved six seconds later. Once I had three data points on the pattern, the extension projected forward."

Greel was quiet again. Then: "The surveillance signal at two kilometers east. Yours to call on response protocol or mine?"

"Yours. But my operational recommendation is passive monitoring and log. She is gathering information, not committing to an approach. Any active response to a passive surveillance asset changes the information she collects."

"Agreed." He looked off-screen for a moment, at something in the medical station. "The unit's performance in the assault was exceptional given the force ratio and the duration of engagement. That assessment goes into the command chain. Every soldier present will have that noted." He looked back at the relay. "Caulfield. The Earth operation. You are staging from here?"

"On the revised timeline, yes. Eight days."

"Good." He was quiet for a moment. "Whatever you find there. Bring it back to something that helps."

That was the most Greel had said to me in personal terms across the entire time the unit had been operational. I took it as the equivalent of Durrik's two-sentence assessment: the highest register of expression that Greel's operational mode allowed.

"That is the intention," I said.

The link closed at 0943.

The rest of the morning ran through logistics: the capability summary distributed to unit briefing chain, Sonn's analysis flagged and relayed, the surveillance log updated as Eshet's signal continued its slow arc northward at the two-kilometer range. Durrik finished his conversation with the convoy officer and came inside at 1100 and poured himself water from the unit's filtered supply and stood at the equipment rack for a moment before he spoke.

"The convoy officer wants a formal assessment of the node's defensive geometry," he said. "For the sustained staging plan."

"Denavi has the terrain knowledge."

"I told him that. He wants it in the next four hours." He drank the water. "I will run the assessment with Denavi. You should spend the time running the full diagnostic before you are in transit to a contact operation."

He was right. The last full diagnostic had been thirty hours ago, before the assault, and the extension had run at sustained full load for six of the seventeen hours. Sonn's standing protocol was a comprehensive assessment within twelve hours of any sustained high-intensity use. I was outside that window already.

I set up the diagnostic interface at the equipment table while Denavi collected her terrain notes for Durrik's assessment. She left without comment, which was her version of respecting the process: she had been on enough diagnostic relays to know what they required and she did not fill the space with conversation.

The diagnostic ran in layers. The first layer was the baseline integration check: the junction points where the Aeluun system had wired into my nervous system in the original calibration, the eighteen months of subsequent development sitting on top of that original architecture like additions to a building that had been designed to bear them. Everything in this layer was stable. The junction points were clean. The original calibration architecture had not degraded.

The second layer was the predictive extension. Here the readings were denser because the extension was newer and because it was running continuously even at rest, the system's monitoring output processing the sector's environmental movement data at a low idle. The readings showed the sustained load from the assault clearly: the extension had run at 94% capacity for just over four hours during the worst of the interior engagement, which was the highest load it had recorded, higher than the retreat in Chapter 25 that had triggered the initial nosebleeds.

The readings also showed something Sonn had flagged in the analysis but that I had not seen in my own data before: the extension's baseline idle level had risen by approximately 12% since the assault. The system had recalibrated its resting state upward. The predictive model was running at a higher ambient level than it had been before the seventeen hours of the assault.

Not cascade instability. Not the failure mode Sonn had described from the Thuvari historical records. But movement: the integration developing, not toward a ceiling but toward a new floor.

I noted it in the diagnostic log and flagged it to Sonn's analysis chain with a timestamp.

The third layer was the physical integration markers. The right ear tone was the most measurable: it had settled at a frequency that Sonn's monitoring equipment could read as a specific carrier signal, the integration's operational frequency, and the tone was now approximately 15% louder at the new baseline than it had been at the old baseline. The assault had permanently shifted it upward by a margin that was not painful and was not interfering with environmental sound processing, but that would not be going back to the pre-assault level.

New baseline. New floor. The system running at a higher ambient level than it had been designed to run, or at least higher than any prior case on record had run, which was not quite the same thing.

The diagnostic took forty minutes. At the end of it the summary reading was the same as Sonn's assessment from Chapter 40: stable, developing, no mapped endpoint, no known risk parameters breached.

What the diagnostic did not measure, could not measure, was the thing Sonn had written about in the final section of the analysis: the question of what the system had been designed to find and whether it had found it. That was not a readings question. It was a different kind of question, and I had read Sonn's note on it twice and had not answered it and was not going to answer it in the next eight days before the extraction team moved.

Some questions ran in the background.

I sent the diagnostic summary to Sonn, to Vasht, and to the Compact medical relay that Greel's team maintained for the unit. Then I pulled up the third candidate's file and read through it again, this time cross-referencing the profile against the capability summary I had drafted for the unit briefing and against the specific ask Vasht had authorized.

First forty-eight hours. Direct integration monitoring authority. Sonn remote.

The first forty-eight hours after the initial contact were the period where the system would begin its read of whoever touched the device, and the person who touched it would have the most specific questions right then, not later, not after the calibration phase was complete. That was when they needed someone who could explain the experience from inside it rather than from notation and clinical assessment.

I thought about what I would have needed in those first forty-eight hours, which was not the question Vasht had asked or that the Compact's standard protocol was designed to answer. The standard protocol was built around subjects who had been briefed comprehensively in advance, who had signed whatever the Compact's legal documentation required, and who had been in the Compact's operational care for some period before the device contact. It assumed a controlled transition.

My transition had not been controlled. I had been pulled off a deer trail and handed information in a medical bay with my wrists loosely restrained, and the questions I had asked in the first forty-eight hours were operational questions, practical questions, the kind of questions a person asks when they are trying to understand what they have walked into well enough to navigate it.

The third candidate was not going to be pulled off a deer trail. The Compact's extraction procedure was better than that, more transparent, less abrupt. But they were nineteen years old and they were going to be told that the integration was permanent and developmental and that the cost structure included a tone in their right ear that would run for the rest of their life, and that information was going to land differently depending on who said it.

I drafted the opening for the explanation I was going to give. Not a briefing. Not a clinical assessment. The honest version, in the register of someone who had run the extension at 94% capacity for four hours in the interior of a building while a Dominion Blade-Commander tried to take it back.

The opening went through three drafts before it said what it needed to say without saying more than that.

Rovan came back at 1300 with two field rations, set one in front of me without comment, and went back to the relay panel. I ate without stopping the draft work.

Sonn's response to the diagnostic summary came in at 1342: four lines of notation confirming the readings I had flagged, a note on the extension's new baseline that said, in Sonn's formal assessment register, that the upward shift was within the developmental range they had been tracking and was not a cascade indicator, and a single final line outside the notation register: The first forty-eight hours. I will be on the relay for the full window. Whatever they ask that you cannot answer from experience, I will answer from data. We will cover the complete picture together.

I read that once, sent back a confirmation, and kept working.

Durrik and Denavi came back at 1530 with the node's defensive geometry assessment complete and handed to the convoy officer. Denavi went to the supply line to draw the unit's allocation for the next twenty-four hours. Durrik came to the equipment table and stood across from me and looked at the diagnostic summary I had left open on the secondary display.

He was quiet for a moment. Then: "The extension's baseline moved."

"Yes."

"Permanent?"

"Sonn says within developmental range, not a cascade indicator. Which means probably yes, this is the new floor."

He looked at the reading for another moment. "And the Earth operation. Eight days."

"Eight days."

He pulled a chair from the secondary equipment rack and sat down, which was unusual enough that I noted it: Durrik worked standing the majority of the time, and sitting was his register for a conversation he was planning to stay in rather than conduct in passing.

"What you said to Greel," he said. "About bringing back what you find there to something that helps." He waited a moment. "You mean that or you said it because it was the right thing to say?"

I thought about it honestly. "Both," I said. "In that order."

He nodded. That was the full extent of his expression on the matter, which was exactly correct, because Durrik's expressions were precise instruments and a nod in this register meant the answer was acceptable and the conversation could proceed.

"The three candidates," he said. "Once they are integrated. What does the unit look like?"

"Sonn's timeline has the first two operational at roughly twelve months under combat conditions. The third candidate at eight months, maybe less, based on the development rate projection."

"And you."

"I have eighteen months and whatever the assault added to the baseline." I paused. "Which is not the same as twelve months of standard integration. The trajectory is different."

"Because the system is doing something with you that it has not done with anyone else on record."

"Yes."

He was quiet for a moment, turning his water container in his hands, the one physical habit I had clocked on him across the entire duration of training and then in the field: he rotated objects when he was processing something, not rapidly, just the slow quarter-turn of a man thinking through a decision.

"The Compact is going to want to formalize your role," he said. "After the Earth operation. When there are four of you and the data on your development is in the command chain." He set the water container down. "They are going to call it a unit and they are going to want someone to run it."

"I know."

"What are you going to say?"

I looked at the diagnostic summary still open on the display. The readings were stable. The extension was running at its new baseline, the tone in my right ear marking it clearly. Eight days to the extraction window. A Dominion Blade-Commander running passive surveillance at two kilometers east, patient and methodical, gathering information for the next engagement.

The forty acres in Cocke County would be coming into spring. The bottomland by the creek would be soft, the kind of mud that took boot prints and held them. The deer trails would be running where they always ran, because deer moved the same routes when nothing pushed them off, and nothing had pushed them off the south ridge in the year and a half since the last person who knew those trails had been on them.

I thought about that for exactly as long as it took to think about it, and then I let it sit.

"I am going to say what the specific ask requires," I said. "Clear enough that it is a specific request rather than a general position. Because Vasht responds better to specific requests."

Durrik picked up the water container again. One quarter-turn. "Denavi."

"Yes."

"She teaches you things," he said. It was not a criticism. It was an observation with approximately the same register as "not bad," which meant it was the highest register he typically used.

I pulled up the message interface and opened a new draft to Vasht. The specific ask for the formalization question was going to need to be built carefully, because the Compact's instinct would be to create a command structure and fill it from existing military protocol, and a command structure built from existing military protocol was designed for soldiers, not for whatever the eight-day window was going to produce when it was done.

The ask needed to define the thing before the Compact defined it for me.

I started with the framework: what the unit's function was, what command structure it required and did not require, and what the decision authority looked like for someone running an integration that was, per Sonn's own notation, operating in territory that had no map.

The draft ran to four paragraphs. I cut it to two.

Rovan logged the surveillance signal's position update at 1602: two kilometers east, another fifty meters north, still passive. Eshet's arc was patient and slow and she was not going to resolve it tonight, or tomorrow, probably not in the next eight days.

But she was going to resolve it.

I sent the two-paragraph draft to Vasht and went back to the third candidate's file.

The nineteen-year-old in the southern hemisphere. The development rate that had made the Thuvari assessor careful with language. Forty miles from Cocke County, one of the other two candidates lived, and I had noted that and not said anything about it because there was nothing to say about it that changed the operational picture.

It was just a fact that sat in the file. I knew it was there, the way I knew the tone in my right ear was there, the way I knew the surveillance signal was at two kilometers east and the convoy was staged at the exterior lot and Greel was in the medical station with a clean fracture and Denavi was drawing supply allocation and Durrik was sitting across the table turning a water container in his hands.

All of it present, all of it running in the background with the patient thoroughness of a system that did not stop when I did.

I closed the third candidate's file. I did not call it done, because it was not done, because none of it was done; the extension's passive monitoring held the sector, Vasht's message interface showed her response to the two-paragraph draft as still composing, and the diagnostic summary sat on the secondary display with its stable readings and its new baselines and its notation of unmapped territory past the current position.

I opened the message interface instead and typed a single line to Vasht: Send me the candidates' files. All three. Complete versions, not the survey summaries.

The response came back in under a minute.

Three files. I opened them in sequence and started reading, and the integration ran in the background of everything, and the tone in my right ear tracked the interface's state with the steady frequency of something that had been built to last, and outside the north access the Caleth Corridor held its flat grey sky over the node's reinforced perimeter while Eshet moved her surveillance asset another fifty meters north in the slow patient arc of a commander who had been in this work for eleven years and understood that the next engagement was not tonight.

But it was coming.

I read the first file through. Then the second. Then the third again, the full version, every field Sonn had flagged and every field the Compact assessor had been careful around.

Durrik set down the water container and stood. "I am taking the 1800 perimeter watch."

"Rovan is logged for 1800."

"I know." He collected his kit. "I will be at the east position if Vasht's response needs an operational assessment."

He went out through the north access.

The hub core hummed. The system ran. I pulled up the opening draft of the explanation I was going to give the third candidate and read it through one more time, looking for the places where it was accurate but incomplete, the places where the technical frame was solid but the experience frame was missing.

There were two. I fixed them.

Then I drafted the next section: what the first forty-eight hours felt like, in language that did not require context, from inside the experience rather than from notation.

Vasht's response to the two-paragraph framework draft arrived at 1714: Under legal review. Seventy-two hour turnaround. Preliminary read: the framework is acceptable in scope, requires modification on the command authority language in section two. Specific revision suggestions attached.

I opened the attached revision suggestions, read them, and began working through the command authority language.

The extraction window was eight days out. The surveillance signal was at two kilometers east. The third candidate's file was open on the primary display with the development rate projection highlighted in Sonn's notation.

There was work to do.

I did it.