244: Submitted

Wanqing submitted the third paper on June 2, 2020.

Not June — she'd said June, and June 2 was the second day of June, which was close enough to say she'd been accurate. The accuracy wasn't the point. The submission was. She'd been building toward this submission since the January 2019 methodology seminar, through four datasets and one unexpected fifth, through two drafts and Professor Liang's review and a field that had not existed as a formal subfield eighteen months before she'd named it.

The submission itself was a single button in the journal's online system. She'd mentioned this the previous week when the final draft was in the last stage of formatting — "it ends with a file upload," she'd said, not as anti-climax but as a statement of what the system required. Everything that had accumulated since January 2019 reduced to a file upload. The file was the paper. The paper was the accumulation. The upload was the point at which the accumulation became available to the field.

The paper title: *Phase Transition Characteristics in Delayed-Feedback Optimization Systems: A Cross-Domain Validation and Search Method.*

Four domains: industrial process control (Guangdong), maritime routing (Singapore), power grid management (Shandong), and a fourth domain added in April — financial settlement systems, contributed by Professor Liang's contact in Shanghai who'd read the June 2019 crossover paper and had been running the probe experiment independently for eight months. The settlement system crossover was at 59 minutes.

Five data points: 55 (laboratory baseline), 56 (maritime), 58 (industrial), 59 (settlement), 61 (power grid).

The range held. The property held. The probing method, tested against five independent domains, located the crossover within 2-minute resolution in each.

The range across all five domains was 6 minutes — 55 to 61 — which meant any system in the 45-to-70-minute delay class needed a 20-minute search window and a specific probe sequence to locate its own transition point. The paper provided both. The first paper had said: the crossover exists. The second paper said: here is how to find it in your system. A practitioner who read both papers and ran the probes could answer the question that the first paper had opened.

This was what Professor Liang had meant by "research trajectory." Each paper answered the question the previous paper had raised.

"Five domains," I said at the bench on June 2.

"The settlement system was unexpected," she said. "Professor Liang's Shanghai contact had been running the probe since reading the first paper. He had the data and the methodology was already compatible. We added it in April."

"How does five domains change the paper."

She looked at the June bench — the summer in its early phase, the maple in full leaf, the bench in its warmest-of-year quality. "It changes the confidence level of the probing method's generalizability claim. With two or three domains, the claim is 'this method works across multiple independent domains.' With five independent domains across four different applied fields, the claim is 'this method generalizes across the delay-feedback system class.'"

"A class-level claim."

"A class-level claim with five independent data points," she said. "Professor Liang thinks it's sufficient for the generalizability conclusion."

"Is he right."

"Yes," she said. She said it the way she said things she'd thought through completely. Not hedging. Not overconfident. Exactly accurate.

She turned to the problem set.

The June bench had the quality of the year's longest days — the light reaching the bench at an angle that felt unhurried, the maple overhead giving the bench complete shade, the campus in its early-summer state that was different from the academic-year state. Fewer students on the paths, but the ones present had a different quality of movement: not the directed-purpose of the semester but the more deliberate pace of people working in the summer with their own schedules, their own targets. The research bench in summer was Wanqing's bench in the fullest sense — the period where the primary obligation was to the work itself, unscheduled by external demand. I had been watching the summer bench's particular character for five years now. It was the version of the bench that was most purely itself.

"The submission response," I said. "Timeline."

"Three to five months. The journal reviews at this tier are slower — the referees are fewer and busier." She turned a page. "October or November at the earliest. Which means the response will arrive during CW VI."

CW VI registration was in October. The paper's response and the tournament's beginning would overlap. Two separate things on different timelines, arriving in the same month.

"You'll manage," I said.

"Yes," she said. Not as reassurance — as a statement of fact. She would manage. She always did, and she'd stopped treating that as worth noting.

***

Mu Qingyao's May message.

*Dr. Liu's nine-month assessment. My brother's symptom curve has reversed and stabilized. The neurological inconsistencies are resolved. The fatigue is residual — Dr. Liu says two to three more months of the protocol and then a tapering phase.*

*He'll be at full capacity by September.*

Full capacity by September. From the three inconsistent Chengdu specialists in July 2019 to full capacity in September 2020: thirteen months. The protocol had worked. Dr. Liu's integrative approach had found what the specialty lenses had missed, and the treatment had done what it was designed to do at its own pace, on the timeline it required.

The referral chain: Father's surgery in this timeline, which had put him in Dr. Chen's network, which had put the referral to Dr. Liu within reach when the question arrived. A chain that had required four years and a trajectory I hadn't fully known I was on when it started. The protocol had been possible because of what had been built — and had been found because Mu Qingyao had been watching carefully enough to ask the precise question that the chain could answer.

*The formation,* she sent. *We ran our first full competitive simulation against another Tianhe server guild last month. Not a tournament match — a structured practice match. The results: our read time at 0.7 seconds, our dual-layer timing coordination operating consistently.* A pause. *We're considering CW registration.*

CW registration. Not the Tianlong server's CW — the Tianhe server had its own competitive format. Mu Qingyao's formation registering for Tianhe's equivalent competition.

*Your formation is ready,* I sent.

*Wenqing said the same thing,* she sent. *When I sent him the April drill data, he replied: "This formation is ready for competitive format. The development path has been unusually clean." I didn't expect him to say that.*

Wenqing commenting on her formation's readiness without being asked. The same way he'd offered to prepare the CW format analysis in October 2018 before being asked — he'd decided to help before the question arrived. He did this with formations whose development he'd been tracking long enough to form a judgment. He'd been tracking Mu Qingyao's formation for twenty months.

*He's right,* I sent. *The Tianhe competitive format — when does registration open.*

*August,* she said. *Group stage begins in October.*

*Register,* I sent.

*Yes,* she said. *I've already decided. I wanted to tell you.*

Not asking permission. Telling me. The same voice she'd had in October 2018 when she'd named herself — *SilverEdge, my name is Mu Qingyao.* Direct, without performance.

*The formation's name,* I sent. *For the competitive record.*

A pause.

*Tianhe Formation,* she sent. *The name of our server. That's what we are.*

Not a dramatic name. An accurate one. They were the Tianhe server's attempt to build the same thing from different materials. The name said that clearly.

*Good name,* I sent.

*Thank you,* she said. *For everything since October 2018.*

I looked at the message. What had she received since October 2018: the CW format analysis, the transitional class documentation, the constrained drill structure, the acceleration through shorter cycles framework, the Dr. Liu referral, and twenty months of Wenqing adding her formation's development to Volume 3 of an archive she'd never visited.

*The documentation should serve the people building with it,* I sent. *The class was built to serve the formation.*

*I know,* she said. *I've been living that sentence for two years.*

***

The June bench.

Summer — the same quality as every summer bench, but the fifth summer. The fifth time the maple had reached its full leaf and the bench had been in full shade in the midday and the morning light had come at the angle that only arrived at this latitude in this season. The bench in June had a specific quality distinct from any other month — the most complete, the most shaded, the most settled.

Wanqing at the bench with the research position notes and the summer schedule.

"Mu Qingyao's brother," she said. "Full capacity by September."

"Yes."

"The Tianhe Formation registering."

"Yes. Wenqing told her she was ready before she asked."

She turned a page. "She's been developing for two years from a standing start." She looked at the June campus — the summer quality, the leaves at their deepest green, the campus in its quieter summer mode. "What does the Tianhe Formation do in the Tianhe competitive format."

"I don't know," I said. "She'll find out."

"And you'll be watching."

"And sending the data to Wenqing," I said.

She looked at the summer maple. "The documentation propagates," she said. "The formation on Tianlong server, the formation on Tianhe server, and now the competitive format on both servers."

"Yes."

"How many directions is the documentation pointing in now."

I thought about it. The archive. The crossover paper. The workshop model. Feng Li's observation notes. Mu Qingyao's drill data. Chen Wei's pattern document. Bai Yueran's co-certification. The Suzhou workshop's formalization project. The referral path to Dr. Liu that was now part of Mu Qingyao's family's medical history.

"More than I could have modeled when it started," I said.

She looked at the June bench.

"Good," she said. "That's what propagation looks like."

The June bench. The sixth summer of sitting at this bench. The documentation propagating in more directions than could be modeled.

***

The second composition.

TwilightTide performed it at a conservatory event in late June.

She sent a recording link the morning after.

I listened to it at the bench.

It was different from the first composition in the way she'd described: the first was about arrival. The second was about continuation. The first had a clear arc — the anticipatory quality building toward the moment it arrived. The second had a different structure: it started mid-motion, as if the listener had walked into a room where something was already happening, and continued to develop without reaching a destination.

The destination wasn't the point. The continuation was.

Wanqing listened with me at the bench. When it ended:

"The match teaching the practice," she said.

"Yes," I said.

She looked at the summer campus.

"She'll write a third one," Wanqing said.

"Yes," I said. "She will."

The summer bench. The second composition. The work that continued teaching the work. The documentation propagating not just through analysis but through the compositions that described the practice — music as a form of archive, arriving at the same conclusions through a different medium.

I sent the recording to Wenqing that evening.

His reply: *The first composition documented the arrival. This one documents the continuation. They're a two-part record of what the 3 AM window has been doing since October 2017. Different medium from the archive — same function.*

Different medium, same function. TwilightTide had been documenting the Iron Hills sessions in music the way Wenqing had been documenting them in analysis.

I sat with that for a while after I sent the recording. The composition had started mid-motion, without orienting the listener first — you arrived into something already in progress and were carried through it without being told where it was going. That structure was not accidental. She'd built it to describe what the continuation felt like from inside: not a thing with a clear start and a clear end, but a thread that had been running before the present moment arrived at it and would continue running after. The listener's experience was the healer's experience. She'd put both in the same formal structure. That was not a small thing to have done, from inside a practice, about the practice. The first composition had required making an experience expressible. The second had required making a structure expressible — the structure of continuation itself, which had no arrival point to point to, no single moment of crystallization. Harder. She'd said so herself. Two forms of the same record, running in parallel, each capturing what the other couldn't.

The analysis record captured the numbers: the timing, the efficiency percentages, the minutes, the coordination intervals. The music captured the quality of the thing the numbers described — what it felt like from inside the practice to have the rhythm arrive at minute 34 rather than minute 44, what the continuation felt like after the arrival. Both were true. Both were incomplete without the other. The archive had the analysis. Now it had the link to what the analysis was about.

*Add a note in Volume 3,* I sent. *Both compositions by date.*

*Already there,* he said. *Volume 3, June 2020 entry. Composition 1: August 2020. Composition 2: June 2020.* He paused. *The archive includes the music now.*

The archive included the music. Volume 3 had started with formation data and session logs. It now included observation notes, training drill results from a different server, a recording link, and a note about a settlement system crossover at 59 minutes. The record had expanded to hold everything that was true about what we were building.

That was what the archive was for.