246: Autumn

The paper response arrived September 18.

Two reviewers. One accepted with minor revisions. One requested major revision — the near-boundary resolution section, precisely as Wanqing had expected.

I knew she'd expected it because she'd said so in August. I knew it was the right call because she'd already written the response before the question arrived. The two-month gap between submission and response had felt long in the way waiting always felt long when the result mattered — not actively long, but the kind of waiting where the background awareness of pending outcome was always present. Then the result arrived on a Tuesday morning and Wanqing sent a single message: *Two reviewers. Major revision request on the near-boundary resolution. I have it.*

The message didn't need elaboration. "I have it" covered everything: the response existed, the revision was not a setback, the question was the one she'd prepared for. She'd sent me the relevant supplementary pages in August and I'd read them, and "I have it" was the sentence that connected those two points across the two-month gap.

She was at the bench when she told me.

"The reviewer's question was about the two-minute resolution claim when the system is operating within 3 minutes of the transition point," she said. "The near-boundary case."

"Pages 8 through 12 of the supplementary material."

"Yes. I revised the main text to reference the supplementary section more explicitly, added a clarification about the resolution degradation near the boundary — the two-minute resolution holds outside the 3-minute margin, and within the 3-minute margin the method requires additional probe iterations. Three to five iterations rather than the standard two." She turned a page. "The supplementary section was already written. The main text revision took three hours."

Three hours. She'd prepared the response before the question arrived.

"Resubmission," I said.

"October 1. One revision round." She turned to the problem set. "If the second round is clean, the paper publishes in January."

January. The third paper published five months faster than the first had been. A journal tier above the first. With five independent data points rather than two. The research trajectory Professor Liang had named was visible in the comparison between the papers — the first had been a good question; the second was a good question with a method; the third was a method tested against five independent domains with a class-level generalizability claim.

"Professor Liang."

"He's pleased," she said. Not elaborating. She didn't need to. He was pleased. That meant the work was what he'd hoped it would be when he'd read her thesis in December 2018 and called footnote 12 unusual. It was sufficient.

***

Mu Qingyao's September message.

*My brother is at full capacity. Dr. Liu cleared him on September 4. The protocol is complete.*

*Thank you* again would have been insufficient and she hadn't sent it. She'd sent: *The Tianhe Formation is in the competitive registration window. We submitted September 6.*

Doing the next thing. The brother's situation resolved; the formation's registration open. Both at the same time because both were part of the same life.

I thought about what it had taken to arrive at September 4: the three Chengdu specialists with their mismatched diagnoses in July 2019, the conversation where Mu Qingyao had named what she was actually looking for, the referral chain that had required knowing where to look, the treatment protocol at its own pace with the flat period at month five and the resolved curve by month nine. All of that had been happening in parallel with the constrained drill sessions, the Floor 15 record, the 0.7-second read time. September 4 was the date when both threads arrived simultaneously — the brother cleared, the formation in registration. Not coincidence. The same period of time, both things running on their respective timelines to the same month.

*Group A projection,* she sent. *The Tianhe competitive format has eight guilds in the group stage. Our formation is unseeded — first time any guild in the cluster has built toward the Pioneer's Path-equivalent class structure. The seeding committee doesn't have a precedent for us.*

Unseeded. No precedent. The same situation Black Dragon Guild had been in in CW I. The seeding committee would have to look at the training data and the practice match record and make a judgment without a template, the way any evaluator has to make a judgment about something that doesn't fit the established categories.

*How did you handle it,* she sent.

*We ran the first match,* I sent. *The data from the first match became the precedent.*

*Yes,* she said. *That's what we'll do.*

*Send me the first match data,* I sent.

*Of course.*

She said it the way she said things she'd already decided. There was no hesitation in the response. She'd been planning to send the match data from the beginning — she'd wanted to tell me the decision was made before asking for guidance she'd already given herself.

Wenqing, when I told him: *She's running the same development arc we ran in 2015. Different server, different class, different starting conditions. Same direction.* He paused. *I want to model the Tianhe Formation's CW-equivalent performance before the first match. To compare the model against the actual results. Not for her benefit — for mine. I've been watching her formation's training data for twenty-two months and I want to know how well I can predict what it does when it competes.*

He'd built models for every Black Dragon match since 2015. He was extending the model to a formation he'd never watched compete — as a test of the model itself.

*Tell her,* I sent.

*I already did,* he said. *She sent me the pre-match training data this morning.*

She was already in contact with him directly. She'd read Wenqing's name in enough of my messages to know how to reach him. Apparently she'd reached out without waiting to be introduced.

***

CW VI registration opened October 1.

We submitted October 3. The same two-day delay as all previous CW cycles.

Ningxia's seeding projection: Seed 1.

Iron Frost Ascent: Seed 5 (up from seed 6 in CW V — they'd improved their server ranking).

MoonShadow: Seed 2.

The bracket draw would not be known until after the group stage. But the seed positions suggested that if all three advanced to the quarterfinals, the bracket could draw Black Dragon against Iron Frost Ascent in the semifinals and MoonShadow in the final — the same path as CW V.

Chen Wei: *Same seeding structure as CW V. The coordination model has improved — the proximity density update in January reduced the 0.25-second average to 0.22 seconds. But QingxueTide's anticipatory window hasn't developed past 1.5 seconds. She's been working on something differently since December — I don't know what yet.*

QingxueTide working on something differently. She'd found that the ceiling was higher than she'd thought. Chen Wei had said she was deciding what to do with the space above it.

*Tell me when she finds it,* I sent.

*You'll see it before I tell you,* he said. *When she finds it, it will be visible in the session data.*

The session data that Wenqing tracked.

Bai Yueran's October message: *The third layer is complete. I'm not telling you anything about it until the match.*

Same as February — she'd told me the layer was complete and nothing else. The discipline of not sharing what she didn't have to share. She'd been working on the third layer since February. Eight months. Whatever it was, it was the result of eight months of deliberate design.

*Good,* I sent.

*You have whatever you have,* she said. *Same principle.*

The same principle applied to both guilds. The match was the reveal. Everything before the match was building.

***

The autumn bench.

The maple in its sixth autumn turn — the orange beginning at the leaf edges, the same timing as every year, the early-October quality that was both familiar and unrepeatable. The first turn had been October 2015, when the bench had been a bench I'd found because the campus had it and the shade was sufficient and there'd been a problem I needed to think through. The sixth turn was the same leaf process, the same tree, the same bench. Different problems.

Sixth autumn. The bench through six autumns.

Wanqing looked at the turning leaves.

"Six autumns," she said.

"Yes."

"The first one — October 2015." She turned a page. "The maple has done this six times since then."

"Seven times from the maple's perspective," I said. "We started watching it in October 2015. It had been doing this before we arrived."

She looked at the turning leaves. "Yes." A small pause. "That's a good way to say it."

The maple had been doing this before we arrived. The bench had been here before we sat at it. The patterns we'd been watching had been in the data before Wenqing started documenting them. This was one of the things that made documentation feel stable — it was catching something that existed independently of being caught. The record was of real events. The real events would have happened without the record. The record made them visible to whoever looked, which was why the record mattered. Not because it created what it described, but because it made the description available.

The autumn turn of a maple could not be archived in the way Wenqing archived formation data. But the bench sitting below it in six consecutive autumns was its own kind of record — a body knowing a place, returning to it, finding it the same and finding it different by the same amount of measure as a year of work.

"The paper will publish in January," she said. Not as a non sequitur — connecting to the same thought. The probing method had been valid before it was formalized. The crossover had existed in every delay-feedback system before she'd named it.

"Yes," I said.

She looked at the October campus.

"The research position extension," she said. "Professor Liang asked about 2021."

"What did you say."

"I said yes. The fourth paper is already forming. The probing method in the delay-feedback class opens a question about systems near the transition boundary that we've been treating as a limitation. Near-boundary systems are fragile — but why. The crossover is a phase transition. Phase transitions have a mechanism. We've described the transition but not the mechanism."

"The mechanism is the next paper."

"Yes. If I understand the mechanism, the probing method becomes more precise. Possibly better than two-minute resolution." She looked at the autumn maple. "That's a year of work. Maybe two."

A year or two. She said it without calculation — as an accurate estimate, not a commitment.

"The research position into 2022," I said.

"Probably," she said. "If the mechanism question is as deep as I think it is."

She turned to the problem set.

"CW VI group stage starts October 15," I said.

"Yes." She turned a page. "And the Tianhe Formation's group stage starts the same week."

"Yes."

"Two competitions in parallel," she said. "Wenqing will be monitoring both."

"Yes."

"Is that too much for him."

"No," I said. "He'll add the Tianhe data to Volume 3. He's been waiting to see what Mu Qingyao's formation does in a competitive format."

She looked at the autumn maple.

"So have I," she said.

The autumn bench. The sixth turn of the maple. The paper publishing in January. Two formations registering for their respective competitive seasons in the same week.

The work continued in more directions than could be modeled.

She looked at the October campus — the leaves just beginning their turn on the trees that lined the main path, the bench maple ahead of them by a week the way it always was.

"Good," she said. The same word she always used for the things that were right. Not complicated — accurate. The bench at the sixth autumn, both competitions starting in the same week, the paper heading toward January publication, the research trajectory named and continuing.

She turned to the problem set.