Mandatory Interval

The system notification arrived at 10:42 in the morning, two and a half hours into my third extended training block. It placed itself at the top of my status panel the way mandatory entries always did, without waiting for acknowledgment: MANDATORY REST INTERVAL. DURATION: FORTY MINUTES. ANNEX COURTYARD. ATTENDANCE REQUIRED FOR ALL ACTIVE CANDIDATES.

My pen went down on the open page. I looked at the cord.

Nothing. The cord had not moved in three sessions of testing. I'd worked through every status entry I had access to, one at a time, watching for a response the way I'd watched the disc. PARTIAL ATTUNEMENT STATUS: null. The inference log entries: null. INTERIM TRAINING ACCESS (LIMITED): null. Even AFFINITY: UNRESOLVED (ACTIVE), which had moved the disc, produced nothing from the cord. The braided fiber sat in its iron rings at middle tension, neither tighter nor looser than it had been on day one.

The stone tile was the same.

Two objects keyed to something I hadn't reached yet, or not keyed to classification state at all, or waiting on a change the current review window hadn't produced. I'd added a notation to the session log: methodology holds, cord and tile results negative, disc remains at eight degrees clockwise, no drift since last observation.

Forty minutes in the courtyard meant crossing the section of hallway where the labeled training halls were. I closed the session log, capped the ink, and left the lamp burning.

The corridor was already in motion when I stepped into it. Fass came out of his hall first, Coll half a step behind, both of them talking with the compressed efficiency of people who'd covered ninety percent of the sentence in their heads before the words came out. They nodded at me. Coll had chalk dust on his left forearm from whatever Kinetic work they'd been running.

Tobias's door was already open. He stood with his hands at his sides, reading something in his panel, and looked up when he heard footsteps. He gave me the same nod he always gave, the one that landed about a quarter second later than a casual acknowledgment would.

The courtyard opened at the far end of the corridor through a low archway. Late morning light came in from above the east wall, and the stone floor was still cold from the night, the kind of cold that works up through boot soles after the first six or seven minutes of standing.

Renne was already there, sitting on the steps at the far corner, eating something. She had a cloth-wrapped parcel in one hand and dark bread in the other, eating without hurrying, watching the rest of us file through the archway with the mild attention of someone who'd already dealt with the interesting part of her morning. Nobody else had found a provisions cache. Hers, apparently, had been there since day one.

Wynn came through the archway from the orientation wing, Petra two steps behind him. They had the slightly unfocused look of people who'd been sitting in documentation modules for three days straight, a look I recognized from early theory sessions before the index had given me anything useful. Wynn had his hands in his coat pockets. Petra carried a folded document she kept reading until she was fully into the courtyard and had to look up to avoid the steps.

Nadia arrived last, pulling her hall door closed with the quiet deliberateness she applied to most physical tasks.

The full cohort, all eight of us, in a space roughly thirty feet across. The mandatory interval didn't offer a stated agenda. It was the system's mechanism for pulling candidates out past the point of productive work, and it ran whether the group dynamic was ready for it or not.

Fass said something to Coll about their oscillation ratio, apparently up three points from the previous session, and Coll said he thought it would keep improving as long as the feedback loop stabilized, and Fass said the question was whether the ceiling was the loop or their individual output levels. They weren't particularly loud about it, just talking to each other in the self-contained register they always used, not aimed at the rest of the group but not private either.

Oren stood near the center of the courtyard with his arms crossed, attention angled down toward his panel, reading with the focused energy of someone who'd expected a specific number and wasn't satisfied with what he saw. He was the challenger the cohort dynamic had produced, not loudly but structurally. Skilled enough that anything less than first position read to him as a problem to be diagnosed, and Fass and Coll's linked amplification loop was, from his perspective, a structural advantage he couldn't match on a solo track.

He looked up from his panel, found me near the archway, and came over.

"I had a question," he said.

"Go ahead."

"Your track." He held his panel at an angle, like he might reference it. He didn't. "The theory track. You said at the cohort session you were on item fourteen. That was three days ago."

"I've moved past it," I said.

"How far past?"

The honest answer was that item count had stopped being a useful measurement around item twelve, when I'd started cross-referencing section headers against footnotes in sections I hadn't formally reached yet. But that wasn't what he was actually asking.

"Enough to have completed the conceptual mapping exercise," I said.

His attention sharpened in a way I'd seen before, the particular focus of someone updating a model rather than winning an argument. He wasn't baiting me. He was trying to figure out where I actually ranked.

"Tobias said you'd hit the ceiling," he said.

"I hit the ceiling the index gives me before the board convenes."

"Which means what, exactly?"

The question was direct enough that I gave him a direct answer. "The index requires demonstrated prior knowledge before it unlocks higher-level material. My classification status counts as prior knowledge for certain modules, but it's unresolved, so the index treats it as pending. Whatever those modules contain, I can't access them until my file is current."

Oren processed this. "So the system's stalling you."

"The system's doing what it documents. I read the documentation."

He looked at me for a beat longer than felt neutral. "That doesn't bother you."

It did, but in the specific way that problems with known structure bothered me, less like frustration and more like the awareness of a fixed variable in a problem I was still solving. I didn't say that.

"I've found other ways to work," I said.

Oren nodded once, filing something away not as settled but as noted. He drifted back toward the center of the courtyard, unhurried, and resumed reading his panel.

Tobias had been standing near the east wall the whole time, far enough that he couldn't have caught the conversation clearly, close enough that he'd been watching it. He raised his eyebrows when I came over.

"You're very calm for someone being interrogated," he said.

"He wasn't interrogating me."

"He was absolutely interrogating you." He glanced toward Oren with something that might have been amusement. "He's been doing it to everyone, actually. Mapping us. I'm the one who took too long on his calibration and then performed two tier levels above the expected threshold in the follow-up work." He said it plainly, no particular pride in it. "Apparently that's interesting to him."

"What did he ask you?"

"Whether I thought the calibration scoring was an accurate reflection of capacity or an artifact of methodology." Tobias paused. "I told him I didn't think it was inaccurate."

"Which is true."

"Which is exactly what I told him." He looked at the stone floor between his boots. "He doesn't like that I don't have a complaint to offer him. The people who find the calibration methodology limiting are easier to categorize. I just happened to do worse than I expected on the first pass and then better on the second, and I can explain both results without having to say the system failed."

Across the courtyard, Wynn had gotten drawn into something with Petra. She'd unfolded the document she'd been reading and was pointing at a line near the bottom, and Wynn was reading it with the tight focus of someone who didn't see the issue yet but was willing to be shown. Renne watched them from her corner, bread finished, the cloth wrapper folded in her lap.

"The cord and the tile still haven't moved," I said.

Tobias tilted his head slightly. "You're still testing."

"Methodology holds. One object at eight degrees clockwise. Two objects null across every status entry I can reach."

"You think they respond to something you can't reach from where you are."

"That's one explanation. Another is that they respond to state changes rather than static flags. The disc moved when my file updated. Null results on stable entries might be expected behavior."

Tobias looked toward the archway. "You'd need a state change to test that."

"Yes."

"And the next state change in your file would be."

"Board convenes at the fourteen-day mark. Classification confirmed or extended. Either way, something in the panel would update."

He was quiet, not uncomfortable with the silence. He was good at not filling space with noise. "Two days out," he said finally.

"Twelve days in. Fourteen-day window."

"Right." He turned his panel over in his hands, a habit. "Do you think it'll extend?"

I didn't know. The board had my conceptual mapping exercise, my logged inferences, and whatever Hael had or hadn't done with her observation period. What the board did with those inputs was outside my visibility. "I think either outcome gives me different data," I said.

"That's not what I asked."

He had the calm, level expression of someone asking a question he expected to be deflected and was making clear he knew the difference.

"I think it'll resolve," I said. "I don't know into what."

He accepted that. "What happens to the hall if your status changes significantly?"

"I don't know. The documentation doesn't address hall behavior for compound-primary candidates specifically. It addresses configuration variation by primary affinity type and by classification state, but the worked examples are for resolved cases."

"Because no one's written the section yet."

"Because no one's written the section yet," I agreed.

The mandatory interval didn't produce any formal activity. It was forty minutes of people standing in a cold courtyard exchanging information they found useful, which was, I'd come to think, the actual mechanism behind it. The system wasn't pulling us out for rest. It was pulling us out to observe how we functioned without a structured task.

I'd noted that in the session log around day three. It was either correct or an overread, and I didn't have enough data to distinguish yet.

Nadia came to stand near us toward the end of the interval, not intrusively, occupying a neutral position that left her in the conversation if anyone addressed her and out of it if they didn't. She understood geometric placement in social space and used it without drawing attention to the fact.

"Oren interviewed you," she said to me. Not a question.

"He asked about my reading pace."

"He asked me about the restriction flag." She said it without visible irritation. "He wanted to know whether restricted candidates received modified hall assignments or whether the delay was purely administrative."

"What did you tell him?"

"That hall assignments are configuration-based and that configuration requires a resolved attunement session, so yes, the delay is a consequence of the restriction, and no, the eventual assignment isn't modified, it's just later." She paused. "He then asked whether I thought a later assignment disadvantaged candidates competitively."

"Does he ever not ask the question he actually wants answered?"

"Not that I've seen." A short pause. "I told him I thought the question was interesting but not the right one. The right question is whether the delay produces different preparation, not different disadvantage. Preparation varies. Disadvantage implies a common metric I don't think exists yet."

Tobias made a quiet sound that was either agreement or appreciation.

The chime for session resumption rang somewhere in the annex corridor, a single sustained tone that carried well through stone. People collected themselves, and the courtyard traffic reversed through the archway.

Back at the cord by mid-afternoon.

The session log was open. The lamp was full from the morning's refill. The cord occupied its rings at the same middle tension, no observable change from the morning's null result or from any of the six prior testing sequences.

My panel was open to the entry I hadn't tested against the cord yet: COMPOUND-PRIMARY RESONANCE PROPAGATION: INTERACTION WITH TEMPORAL AFFINITY STRUCTURES NOTED. The inference log flag. The one timestamped at the same administrative cycle as the disc's first movement.

The cord shifted.

Not much. The fiber didn't unwind or collapse. The tension changed, pulling incrementally toward taut over three or four seconds, then held. The braided material was slightly more horizontal in its sag than it had been, which meant the rings were bearing a fraction more weight. It was small enough that if I hadn't been watching with my attention fixed on the cord's middle point, I might have read it as the natural settling of material.

I wrote it down immediately, as precisely as I could: cord registers response to COMPOUND-PRIMARY RESONANCE PROPAGATION entry. Direction: tension increase. Duration of transition: three to four seconds. Current state: held taut. Prior state: middle tension. Entry was opened, not navigated away from.

Then I moved away from the entry and watched.

The cord returned to middle tension, slowly, over a similar interval. Back to the same entry: tension increased again. Back to a neutral entry, and middle tension returned. Back to COMPOUND-PRIMARY RESONANCE PROPAGATION: tension increase.

Repeatable. The cord was reading that flag specifically, responding to whether the entry was actively open in my panel or not.

I wrote all of it down, dated and timed against the bell schedule, then sat back and looked at what I had.

The disc responded to AFFINITY: UNRESOLVED (ACTIVE), a primary classification state flag. The cord responded to COMPOUND-PRIMARY RESONANCE PROPAGATION: INTERACTION WITH TEMPORAL AFFINITY STRUCTURES NOTED, an inference log entry, a logged observation rather than a resolved status. Two different categories of entry, two different objects. The stone tile had produced null results to every entry I'd tested, which now had meaning it hadn't had before: the tile wasn't broken or irrelevant. It was waiting for a third category.

Three candidates remained in my status panel that I hadn't yet run against the tile. Two I'd already used and logged as null. The third was CLASSIFICATION REVIEW: BOARD CONVENING DATE CONFIRMED.

The review window board date. That entry would appear when the fourteen-day cycle produced a formal scheduling confirmation rather than just the administrative timer running. The panel currently showed the timer, not a confirmed date. When it did, the tile might respond.

Two days out.

I wrote: tile test pending board date confirmation entry. Predicted state change at fourteen-day window conclusion or prior if board scheduling populates early. Will observe at that interval.

Then I tested the cord against every remaining untested entry, recording each result. Most were null. PARTIAL ATTUNEMENT STATUS produced a small response, less pronounced than the resonance propagation flag, a brief increase in tension that released faster. The gradation was worth noting: the cord had at least two response levels, which meant it was reading something more granular than binary present-or-absent.

The lamp was at three-quarters by the time I wrote the final observation of the session. Three objects on pedestals, each one keyed to a different tier of my classification record: the disc to current primary status, the cord to logged inferences and their interaction properties, the tile to something that hadn't fully materialized yet. The hall wasn't abstract. It was a physical readout running parallel to the panel, tracking the same information in a format the panel didn't use.

The question was who had built it that way, and whether it was specific to me or standard architecture that every restricted candidate encountered. The documentation hadn't addressed it. The index didn't have a section for it. Whatever this was, I was either the first candidate to work through it systematically or the first to write it down.

The session log had been open the whole time. Every cord test, every tile observation, every disc check over the past three days, all of it was in the record with timestamps and a methodology note from day one.

I capped the ink, left the lamp to burn down on its own, and walked back through the annex corridor toward the orientation wing.

The corridor was quiet at that hour. Fass and Coll's door was dark under the gap. Tobias's light was still on, visible under his door, which meant he was running late into his session again. The orientation wing had the particular stillness of people who'd stopped moving for the night, not the stillness of an empty building but the stillness of eight people each contained in their own work.

My restricted theory index was open on the desk in my room where I'd left it, two pages into a section about compound-primary behavior during unresolved classification intervals. The section had unlocked three days ago and covered ground I'd mostly already mapped from cross-referenced footnotes. Near the end, though, there was a subsection I'd skimmed and marked to return to, about observer confirmation protocols for cases where hall behavior produced measurable physical register output outside a formal exercise context.

I'd flagged it as potentially relevant and pulled it back up now.

The subsection ran six paragraphs. The first three were procedural, laying out the general framework for when hall behavior counts as an observable record versus an informal log. The fourth paragraph introduced a specific condition: if physical register output occurs during an active session log interval and the log is subsequently reviewed by a Council-level observer within the same administrative cycle, the hall behavior enters the observer's formal notation pathway without requiring a separate submission.

I read that paragraph again.

The session log had been open. The cord test was in the record. The disc observation was in the record. The stone tile methodology was in the record, all of it timestamped, from an active session interval. If Hael reviewed the session log before the window closed, the hall's behavior would enter her notation pathway automatically. I hadn't needed to submit it separately. The log just needed to exist and be open during a Council-level review of the same session interval.

The fifth paragraph noted that this pathway was available only to restricted candidates undergoing compound-primary assessment and only during the classification review window. Standard candidates didn't have session logs that overlapped with physical register observations, because standard candidates ran their register work in formally configured halls under monitored exercise conditions.

The sixth paragraph said: the observer confirmation pathway for session-integrated physical register observations does not require candidate action after the log is created.

Outside the window, the annex courtyard was dark, the stone floor invisible under a clear, starless sky. Twelve days were behind me and two remained, unless the board pulled their convening date forward, which the documentation suggested was possible but rare.

The cord had given me its first response that afternoon. The tile was waiting on a flag that didn't exist yet. The session log was closed for the night, but tomorrow it would open again, and if Hael was reviewing at any point during that interval, the cord response was already in the record.

I hadn't staged any of this. The methodology had produced the results, and the results had fallen into a documentation pathway I'd read about in a subsection I'd flagged three days ago and only pulled back up tonight.

I dog-eared the page, set the index on the corner of the desk, and turned the lamp down to a low wick.