Read Access

The third attempt of the morning gave me sixty-eight seconds.

I logged it, stood from the writing shelf, and went back to the center of the room. The hall smelled of lamp oil and dry stone, with a faint staleness off the slot windows, which never moved much air. The pedestal objects were where I'd left them. Disc at eight degrees. Cord straight on its pin. Tile flat on the third pedestal, unchanged since the session two nights before.

Sixty-eight seconds was better than sixty-three. Fifty-eight had come before that, the first result of the morning. The progression was real, if uneven.

Fourth attempt: sixty-five seconds. That inconsistency was something the documentation hadn't prepared me for. The rate-separation framing was holding, both states pulling toward their own edges rather than converging at center, but the duration wasn't a clean upward line. Some attempts ran shorter than the one before. I'd been treating each result as a separate data point rather than a failure, because the aggregate was still moving in the right direction, and because the CANDIDATE OPERATIONAL NOTES section specifically said to document progressive duration increases, which implied they expected ragged progression.

Fifth attempt: sixty-one seconds. I wrote that down too.

The board review window had one working day left. What the log was building, taken across all five attempts this morning, was a documented pattern of engagement: starting from the thirty-second baseline, extending through fifty-eight, sixty-three, sixty-eight, sixty-five, sixty-one. The jagged edge of it would probably read as more convincing real training data than a clean exponential climb. Anyone who reviewed session logs regularly could tell manufactured numbers from someone working a variable and getting imperfect results.

Whether anyone was currently reviewing logs was the question Tobias had handed me yesterday morning, the one I hadn't been able to set down since. Someone had registered three observer accounts to the Arch-Tier Annex at initialization. Read access, based on the credential format. Every session log entry since day one potentially visible to someone with council-level authorization. The disc registration from day one sat in that record. The cord's first positive response from the afternoon session two weeks ago occupied a closed log. The tile entry from two nights ago sat in the open log. The compound-primary inference and its two unconfirmed properties, all logged. If any of those accounts were actively monitoring rather than sitting passively registered, they'd seen my entire working record from the start.

The question of whether to perform for that audience or work around it had occupied part of my attention all morning. The answer I'd landed on was that the distinction was probably false. The training I needed to do for the board review was the training I needed to do anyway. Documenting progressive duration increases wasn't strategic posturing; it was what the restricted subsection said was relevant, and it was the work I was already doing. If someone was watching, they'd see a candidate working a compound-primary training variable with methodical logging habits and improving results. That was accurate.

Sixth attempt. I located the partial resonance state in about twenty seconds, faster than the first attempts of the morning, which tracked with what the documentation said about rate-separation capacity being trainable. The partial resonance had a particular quality I'd been finding by feel since the night before last: a lateral pressure in the upper chest, no directional push, like two equal forces applied from opposite sides with nothing moving between them. When I introduced the inverse, the equilibration started immediately, the two states sliding toward center, and the task was to move faster than that slide, pulling each state back toward its own edge before the center could close.

The gap held at twenty seconds, then thirty. Forty.

At fifty-two, the quality of the states changed, a differentiation I hadn't catalogued yet. Less like two pressures maintaining separation and more like two distinct textures coexisting in the same space without either one bleeding into the other. Subtle enough that I nearly lost the separation trying to pay attention to it.

Seventy-one seconds. The equilibration finally closed and both states flattened. I stood still while the residual ran out, about four seconds, and then went to the shelf and wrote it down.

Sixty-eight, sixty-five, sixty-one, seventy-one. Best result so far. Also the most interesting attempt, because of whatever that textural distinction had been at the fifty-second mark. I added a note in plain language: "At approximately fifty seconds, concurrent states developed a qualitative distinction beyond positional separation, described as differential texture rather than differential pressure. Duration extended beyond previous maximum. Possible relationship between qualitative differentiation and extended maintenance duration. Requires replication."

The lamp had burned to the halfway mark. Outside, the light through the slot windows was mid-morning, warmer than when I'd started. I'd been in here a little over two hours.

Re-reading the CANDIDATE OPERATIONAL NOTES section for the fourth or fifth time, I heard the door open.

Sable was in the corridor, carrying a small tablet-style interface, a configuration panel by the layout. When she stepped into my hall she looked at the pedestals first, then at me.

"Progress check," she said. "I'll be quick."

She moved to the secondary access panel on the hall's north wall, the one I'd never received permissions to open. Her credentials cleared it in about three seconds, the same fluency I'd expect from staff-level access rather than a cohort credential. It was the same category of gap she'd shown at the courtyard gate and with the zero-point phrase, one more inconsistency I filed next to the others without an explanation for any of them yet. She scrolled through something, made a note, scrolled again.

"You've been working the null-state induction," she said, still looking at the panel.

"Since day one. The documentation recommended it."

"Which documentation?"

"The module access I received under limited status. Theory and system literacy."

She didn't push on whether there was other documentation. She looked at the panel for a few more seconds, then closed the configuration view and turned to face me. There was a professional neutrality to her, not cold exactly, just precise about what she made visible.

"Some hall configurations," she said, "generate observation flags when users hit certain threshold events. The specific thresholds vary by configuration. I'm not in a position to tell you which thresholds or which configurations."

She'd delivered that as carefully as anything I'd heard in the Annex.

"Have I generated any flags?" I asked.

"I can't discuss observation records." No hesitation, which meant she'd expected the question. "But flags are administrative documentation. They route to the relevant oversight accounts automatically."

She waited, and when I didn't fill the silence, she continued.

"Three of your cohort members have hit a first attunement plateau. Standard progression stall, typically resolves within a few days. Cross-discipline exposure helps, if you're in contact with the others."

"Who are the three?" I asked.

"I'm not able to give you names." She was already moving toward the door. "But it's cohort-wide information that the plateau is active, so I'm not telling you anything private by mentioning it. The voluntary session this afternoon might be useful timing for that."

"What voluntary session?"

She paused at the door. "Check your noticeboard."

Her footsteps continued down the corridor to the next hall. The door swung behind her, and the lamp flickered once from the draft, then steadied.

Administrative documentation. Routes to the relevant oversight accounts automatically. She'd said oversight accounts, not oversight personnel, which was a specific word choice in a field that cared about word choices. The observer credentials Tobias had described were registered accounts, and if observation flags routed to relevant oversight accounts automatically, then a flag generated by threshold activity in my hall could have already triggered an automatic notification to any of the three registered accounts.

The second condition of the three-condition protocol required a documented review of the session log with a timestamp showing I logged the inference after review. An automatic notification didn't match documented review, but it could create a plausible path toward one. If an observer account received an automated flag, and if someone with active access to that account then opened my session log to assess what had generated it, the timestamp would show the review.

I added a short line to my notes: "Sable: observation flags route automatically to oversight accounts. Possible automated notification path to observer accounts satisfying second condition. Status of condition two: upgraded from background variable to live possibility."

Then I put on my coat and went to look at the noticeboard.

The posting was from Fass, identifiable by the handwriting: compact, tilted slightly left, every capital letter a quarter-size larger than it needed to be. A working session in the courtyard for twin-synchronized attunement calibration, specifically an interference problem they'd been working on in their respective halls. Neutral resonance field required. Time listed as mid-afternoon, roughly two hours out. Participation optional but useful.

The last line said: "Null-type candidate specifically useful if available."

Wynn was standing beside me by the time I'd read it twice. He had the careful expression he'd been wearing since the rescheduling notice went up, aware of his own status asymmetry and not pretending otherwise.

"You going?" he asked.

"Probably. You?"

"Petra wants to." He looked at the posting for another moment. "We're still waiting on the Day Seventeen session. Doesn't mean we can't be around."

He said it as if checking whether I agreed, which meant either he'd gotten pushback about that elsewhere or was worried he might.

"Two weeks at the Annex without being in the same space as the rest of the cohort seems worse than whatever the social awkwardness is," I said.

He made a short sound, somewhere between agreement and relief, and went back down the corridor.

The courtyard in mid-afternoon had better light than in the morning. The low wall on the south side caught it at an angle that cut the worst of the wind off the well, and six people in a stone square raised the temperature noticeably. Osku arrived last, seven minutes after everyone else, and sat down on the wall with the expression of someone attending against their better judgment.

Fass opened it efficiently, no preamble beyond what he'd written: "The interference problem is a resonance-bleed at the harmonic boundary. Coll and I sync at a particular frequency and it creates a spillover into adjacent attunement states. We need a stable null field to test against."

"Null field," Renne said, "meaning Eli stands there and we see if the bleed goes somewhere or stops."

"Essentially," Fass said.

"I've done less flattering things," I said.

Fass didn't smile, but Coll did, briefly, the same smile Coll used when Fass said something accurate that was also technically absurd. They were different in ways that took time to map: Fass ran hotter, more direct, faster to state a position; Coll held things longer before speaking and tended to frame observations as questions.

The work itself was real. Standing roughly three meters from the twins, I held the partial resonance state while they synchronized. The resonance-bleed Fass had described showed up in the system's output notation, a faint oscillation in the harmonic readout rippling outward from the point of their synchronization. When I held the null state at the boundary of that ripple, the oscillation damped significantly, enough that they could isolate which part of the bleed was structural and which was the interference they were trying to correct.

For about forty minutes, the courtyard was a working session, not a social event. Fass and Coll ran the same synchronized sequence eight times, adjusting the harmonic boundary each run, and I held the null state through all of it, logging results in my head rather than on paper, noting that holding the partial resonance under mild external attunement pressure produced slightly different properties than holding it in isolation. More stable in some ways, the external resonance field giving the null state something to differentiate against. Less stable in others, the pull toward the ambient attunement register occasionally disrupting the rate-separation.

Nadia sat on the well's edge through most of this, arms loosely crossed, watching the readouts. She had a focused quality when she paid attention to something technical, different from her usual flatness, and she asked Coll two questions in a row about the harmonic boundary's behavior that shifted the calibration work in a direction that produced a noticeably cleaner result. Fass thanked her, direct and brief. Nadia blinked once, her chin lifting a half-inch as if the acknowledgment had landed differently than she'd expected.

The social temperature of the session changed around the ninety-minute mark, when Fass and Coll had gotten enough data to wind down the calibration and the session became less structured. People stayed because leaving felt like a statement, and the afternoon was pleasant enough that staying required no justification.

Renne, who had been taking notes in a small book with a cracked spine, looked up. "Osku, does your attunement output always anchor to the left?"

Osku had been sitting on the wall since he arrived. "Why?"

"Because the readout skews about twelve degrees consistently, which is either your natural anchor point or a compensated technique, and I can't tell which from here."

"Does it matter?" Osku said.

"Probably not to you," Renne said. "It matters to me because I've been trying to work out whether anchor skew is native or trained, and you're a clear example of one or the other."

"It's native," Osku said.

Renne made a note. Osku looked at the note.

"You're writing that down," he said.

"I write everything down."

"Why?"

"Because the documentation isn't comprehensive." She still didn't look up from the book. "Someone has to fill the gaps."

Osku stared at her, jaw set, gaze fixed. The stillness had a weight to it, the kind that meant calculation rather than withdrawal. "That's a specific way to think about your own training."

"You have a different way?"

"I train until I can do the thing," Osku said. "I don't need to understand why the thing works."

"That's going to run out of runway," Renne said, still writing.

"Hasn't yet."

"You've been here twelve days."

Osku opened his mouth and then closed it. Coll laughed, short and genuine. Fass's expression moved in the way that was probably the nearest he got to amusement. Wynn was grinning. Petra covered her mouth with two fingers. Even Tobias, who had been sitting at the far edge of the low wall since the start of the session, had something in the set of his eyes that looked like suppressed laughter.

Osku looked around at all of them, apparently decided that getting annoyed would make it worse, and said, "Fine."

"You're not wrong, by the way," Renne said. "Instinctive training works until the mechanics get complex enough that you need the framework. You're just going to hit that wall later than most people. Then all at once."

"That's not reassuring," Osku said.

"I didn't mean it to be."

The laughter that followed was real and slightly surprised, the kind that happens when a group discovers it can collectively find the same thing funny. It lasted maybe thirty seconds, and when it stopped the courtyard had a different quality than it had at the start: less like eight people occupying the same space for separate reasons and more like eight people who'd been in proximity long enough to develop a shorthand.

Wynn and Petra were in it. That was what I kept noticing: they were participating despite the deferred-attunement asymmetry, and the rest of the cohort let them participate without making it a point of comment either way. Whether that was tact or indifference I couldn't determine, but the group's social map was closer to complete than it had been two hours ago.

Tobias hadn't mentioned the observer accounts to anyone. I'd been tracking for it all session, watching for any version of what he'd handed me at the well yesterday morning, and he hadn't come near it. He'd been present, quiet at the edges, responsive when addressed directly, but the specific information he'd shared with me had stayed between us. Either he was holding it as private advantage, or he'd decided the cohort didn't need it yet, or he was waiting to see what I did with it before deciding his own position. Any of those explained the silence.

The session broke without announcement, the way it does when a group has mutually decided it's done without anyone saying so. Fass and Coll left first, already talking about the calibration results. Osku went without a word. Nadia gathered her coat in one practiced motion and was through the courtyard door before Wynn had finished standing up.

The status message appeared in the window overlay while I was still in the courtyard, during the thirty seconds between Nadia leaving and Petra finally getting Wynn to stop re-reading the session notes he'd made on a folded paper.

The window appeared for all of us simultaneously. Renne, who was tucking her cracked-spine notebook into her bag, stopped moving. Tobias looked at it, then at the others, then back at it. Wynn's expression went still, his jaw tightening as his eyes tracked across the window.

THIS COHORT HAS BEEN ASSIGNED A GALACTIC COUNCIL OBSERVATION DESIGNATION. THE SYSTEM IS MONITORING YOUR PROGRESS RECORDS IN ACCORDANCE WITH ADMISSION FRAMEWORK PROTOCOL 7. YOU MAY SUBMIT QUESTIONS REGARDING THIS DESIGNATION THROUGH YOUR DISTRICT COORDINATOR.

The sender identification was an alphanumeric string: seven characters, a hyphen, four characters, followed by a council suffix I didn't recognize but that had the same formatting structure as the credential type Tobias had described from his training interface configuration tab.

Nobody said anything for ten seconds. Ten full seconds of eight people standing in a stone courtyard, some in the act of leaving, all reading the same message in their own status window.

Renne closed her bag. "Well," she said, "that's new." She went back through the door to her hall.

Tobias was still looking at the message. His expression moved, brief and recalibrating, and then it was gone. He left without speaking, which was its own kind of communication about what he was thinking and whether he wanted anyone to see it.

The others filtered out in ones and twos over the next two minutes. Petra said something to Wynn in a low voice I didn't catch. Wynn nodded, looking at the middle distance with the particular unfocused quality of someone reading an internal window while also standing in a physical space. Osku, who had nearly reached the door before the message arrived, turned and stood in the doorway, pausing long enough to look back at the courtyard as if reconsidering, and then left.

I waited until the courtyard was empty before going back inside.

The message changed what had been operating theory into confirmed structure. Three observer accounts, registered at initialization, read access to Annex activity: Tobias had given me that yesterday, and I'd been reasoning from it since. But there was a difference between reasoning from private information and receiving a formal designation message from an alphanumeric council identifier that appeared simultaneously in every cohort member's status window. One was knowledge I held alone. The other was shared fact that everyone in the cohort now had to include in their picture of what this assessment was.

The question of whether Sable's observation flags had already triggered a documented session log review felt less theoretical than it had this morning. Whatever entity sent formal designation messages to entire cohorts didn't have passive registered accounts sitting dormant. Those accounts were active, and if they were active, they were reviewing the logs that generated the flags Sable had been too careful not to confirm.

Everything logged since day one sat behind whatever credential had just sent that message to eight windows at once: the disc's registration, the cord's first response from two weeks back, the tile entry from two nights ago, the compound-primary inference with its unconfirmed properties. Fourteen days of working record, and the account holding it had finally announced itself instead of staying invisible.

Back in the hall, the lamp needed oil. I filled it from the small reserve container Hael had left on the shelf during the first week. The tile sat on the third pedestal, flat and unchanged. The cord was straight on its pin.

At the center of the room, I located the partial resonance state.

Twenty seconds to onset, consistent with the morning session. The lateral pressure in the upper chest, the absence of directional push. When I introduced the inverse, I went faster than any previous attempt, not harder but faster, the active rate-separation applied from the first moment of contact between the states rather than waiting for the convergence to start.

The gap opened before the equilibration could begin its closing motion.

Twenty seconds. Thirty. The two states held their positions, each at its own edge, and the textural distinction I'd catalogued at seventy-one seconds during the morning session appeared earlier this time, around thirty-five seconds, clearer and more defined. Two separate qualities rather than two separate pressures, one residing in the chest and one at the back of the throat, distinct enough that I could track them independently without losing the separation between them.

Forty-five. Fifty. Sixty.

At sixty-seven the equilibration gained on the separation, the states beginning to slide toward center despite the rate-separation, the maintenance becoming active work rather than passive holding. I pushed through the first convergence attempt the same way I had during the seventy-one-second run, faster, pulling each state toward its edge rather than just holding position.

Seventy-three seconds before equilibration won and the states collapsed flat.

The residual lasted five seconds, slightly longer than the morning attempts, and I stood with it until it cleared. Then I went to the writing shelf and logged the result: time, duration, method consistent with previous entries, the qualitative distinction at thirty-five seconds described in the same language I'd used for the seventy-one-second attempt, the earlier onset of that distinction noted as a change from the prior run.

The session log now had: sixty-eight, sixty-five, sixty-one, seventy-one from the morning. Now seventy-three.

The board review window had one working day remaining, and the log was building what the restricted subsection had said to build: documented progressive duration increases demonstrating active engagement with the compound-primary configuration.

Whether anyone was reading it in real time, from an account with a credential format Tobias had recognized as council-affiliated, with access to everything logged since day one, was a question the system message had answered in general without answering in any way that would tell me what came next.

The disc was at eight degrees. The cord was straight. The tile was flat.

Seventy-three seconds. It was better.