131: The Recruitment Window

The secondary recruitment window opened on Monday June 8 and Wanqing ran it the same way she ran everything — methodically, without announcement, in the margins of the time she already had.

She set up the application intake on Sunday evening in the guild's recruitment module. I'd offered to help with the text and she'd declined, which meant the postings would be better than if I'd written them. By Monday morning she'd put three separate listings in three different server-forum subsections: the Floor 5+ veterans thread, where players who had serious dungeon-clear experience would be looking; the Healer-class coordination thread, which was where she expected to find the support coverage the guild's roster needed; and the general recruitment listings, which was noise but was also reach.

The general listing was notable for what it didn't say. No mention of Severing Light's server ranking. No mention of the Floor 5 through 6 server-first records that were currently driving a significant portion of the forum's guild-discussion traffic. Just the charter terms, the council structure, the commitment expectations, and at the bottom, one line that Wanqing had written and I'd read twice when she showed me the draft: *Guild is currently preparing for CW I bracket entry. Applicants should understand that commitment expectations will increase during bracket preparation.*

"That's going to narrow the applicant pool considerably," I said.

"Yes," she said. "That's the point. I don't want a hundred members who joined without knowing what they were joining. I want seventeen members who read that line and kept reading."

By Thursday we had forty-one applications.

Wenqing processed the applications against the standard criteria: floor-clear history, formation discipline indicators from their public kill-cam footage, communication patterns in public guild and party channels, IRL-schedule stability signals from their logged session patterns. He ran the applications in the same sequence he ran everything — in order, without skipping, flagging items for council review when the evidence was ambiguous rather than when it was clearly yes or clearly no.

Eleven flagged for council review. Of the eleven, six passed the full council vetting. Of the six, four accepted the charter terms and completed the Iron Hills probationary session, which was the final gate.

Four new members by end of week. Guild at eighty-seven.

The three referrals from TwilightTide came through separately — they applied through the public window, individual applications, no different in format from any other. Wenqing processed them through the same criteria. He flagged them for council review in a separate file with a single note at the top: *Shared connection in applicant background. Connection is to a current guild member. Vetting otherwise clean. Recommend proceeding.*

He didn't name the connection. The three referrals hadn't mentioned a connection in their applications. Wenqing filed the note and moved on to the next item in the queue.

All three passed. Guild at ninety.

Ten short of the full charter target. Wanqing noted this in the council channel without comment, which was her version of noting that the next step was visible and someone should take it. I noted back that the Floor 7 attempt Saturday would put the guild in a position to attract the last ten from the server's Floor 5+ pool — players who'd been watching the server-first records accumulate and waiting for evidence that Severing Light was the kind of guild that knew what it was doing. The Floor 7 server-first would be that evidence.

The next recruitment cycle wasn't something that needed to be run. It would run itself.

***

TwilightTide's Floor 7 analysis came to the Wednesday council session in the same format as all her previous analyses: printed pages, the handwriting dense and annotated in the margins, the specific quality of a document that had gone through at least two drafts before it was printed. She set it on the table at the start of the session and let the council read before she spoke.

I watched the council read it. Old Wolf went straight to the second page — he always looked at the solution first and the problem second, which was the approach of someone who'd been solving tactical problems for long enough that the structure of the problem was less interesting than what you did about it. Wenqing read from the top and made small marks in his copy. Wanqing read it once through quickly, then went back to the first page and read it again slowly.

The Floor 7 boss was the Grave Sentinel — Lv 43, Physical-type, with a taunt-abuse mechanic that had been eliminating advance teams since the floor opened two weeks ago. The mechanic worked by forcing the highest-threat target into a locked aggro state for fifteen seconds, during which the Sentinel applied a stacking armor-reduction debuff at three-second intervals. After five stacks — fifteen seconds — the debuffed target was taking four hundred percent of the Sentinel's base damage per hit, which killed any tank-class character regardless of gear.

Every guild on the server had been solving it the same way: rotating tanks every twelve seconds, pulling aggro off the primary before the fifth stack applied.

"The rotation gap," TwilightTide said, "is 0.8 seconds. Tank A has to drop aggro, tank B has to establish aggro, and the Sentinel has a 0.8-second window where it selects a new highest-threat target from the full party. In that window, it frequently reselects tank A."

Wenqing: "Teams have been documenting the reselect problem on the public forum as a potential bug."

"It's not a bug," TwilightTide said, with a patience that suggested she'd already spent time being certain about this. "The aggro calculation resets on the taunt break, and tank A has residual aggro from the previous cycle. If tank B doesn't overcap tank A's residual by a sufficient margin, the reselect happens because tank A is still the highest-threat target when the new calculation runs."

Old Wolf: "So you overcap reliably."

"You can't. Not with the tank rotation approach — the overcap margin required is too large to maintain across repeated cycles without compromising the rest of the formation's output." She pointed to the second page of the analysis. "You change the approach. The mechanic targets the highest-threat target. If the highest-threat target has no armor to reduce, the stacking debuff becomes mathematically irrelevant."

The room was quiet for a moment.

Old Wolf said: "You want to bait the taunt with the Berserker."

"I want to bait the taunt with Bladeless, yes. His armor is zero by design — the Berserker class's zero-armor builds generate threat through damage output, not through defensive mechanics. The debuff applies to armor, not base defense. At zero armor, five stacks of armor reduction changes nothing. He tanks the fifteen-second taunt window at full Berserker output, the Sentinel's aggro table resets, and the tanks reassert standard formation for the next cycle."

"His HP."

"Zhu Yuhan and I have modeled the incoming damage at full debuff stack against his Lv 40 HP pool with two healers. The margin is eleven percent above his passive regeneration baseline with Warlord's Presence active." She indicated the annotation on the third page. "At the upper end of the variance in the Sentinel's attack speed, that margin closes. Which is why I pre-position at eight meters instead of eleven — I need the additional reaction time."

Old Wolf looked at me.

"The math holds," I said. I'd been running the numbers from the first page onward. "The variance is the real variable. The approach is correct."

The council approved the attempt for Saturday. Wenqing filed the analysis in the guild's dungeon records, attributed to TwilightTide's work.

***

On Friday afternoon Wanqing texted me from the Suzhou train. She'd gone down to talk to Xiaoyu. The message came in while I was in the campus library's periodical reading room, and I read it at the table with the afternoon light coming through the clerestory windows overhead.

*She has a plan,* Wanqing wrote. *She's been running it since February. She's not going to tell you what it is until she's ready to execute.*

*What's the shape of it.*

*She's been saving. Not from pocket money — contract translation work. Small contracts, academic documents, through a student referral network at her school. Her Japanese is better than she's been telling you. There's a sister exchange program with a school in Osaka and the teachers sometimes need informal translations for parent communications.*

I sat with that. Xiaoyu was fourteen. She'd been in seventh grade in February when she'd decided there was a 200,000-short problem and designed a plan for it herself.

*How much has she saved,* I typed.

*She says she'll tell you when it's relevant. I didn't push.* A pause in the typing indicator. *She's fourteen and she's been doing contract translation work in the margins of seventh grade. She has a plan she designed herself. Don't ask her about it until she's ready.*

*I wasn't going to.*

*I know. I'm telling you anyway.* Another pause. *She said to tell you she's watching the guild rankings. She's proud of the server-first.*

I sat in the library's periodical reading room with the clerestory light on the table and thought about that.

*She's proud of you too,* Wanqing added. *She just doesn't say it the same way the family does.*

*I know.*

*She's going to be all right.*

*I know,* I said.

The train notification cut in briefly — Wanqing's icon showed she'd switched to a call — and then her text continued: *One more thing. I asked her about the red-string bracelet. She said she made it herself. She said she found the pattern in a library book.*

*What pattern.*

*She didn't say. She showed me the bracelet knot and said she'd teach me if I wanted to learn.* A pause. *I said yes.*

I put the phone down and looked at the clerestory windows. The June afternoon light through them was the particular clear quality of early summer in Hangzhou before the humidity built up — clean, sharp-edged, the kind of light that made things look like themselves.

Xiaoyu at the staircase landing with her library book and her Japanese translation contracts and her plan she'd been running since February. She had the red-string bracelet on her left wrist and a document she'd covered when we came through the door. She'd been in seventh grade and had looked at a 200,000-short problem and decided she could be part of the solution. She hadn't told me. She'd told Wanqing that she'd tell me when it was relevant.

In the old timeline, Xiaoyu at fourteen had been working two part-time jobs and hiding one of them from my mother. The hiding had been about pride — she hadn't wanted to be seen as struggling. The plan she'd built for the two-hundred-short problem in this timeline had the same quality: designed by herself, held privately, executed quietly, not presented for approval.

She'd gotten that from somewhere. Probably from watching me.

I put the phone away and went back to the periodical reading room's desk and sat there for a few minutes without reading anything. The library's afternoon was the sound of paper and the distant click of the lending desk's keyboard and the particular quiet that reference sections had, the deep institutional quiet of rooms full of things that were being kept.

She was on it. She'd been on it since February.

The guild recruitment close came in at eleven PM. Final count for the week: eleven new members confirmed. Guild at ninety-four.

Six short of the full charter target.

I pulled up the Pioneer's Path cycle log. Forty-eight complete. Two more cycles and the Floor 10 Heritage access unlocked, and then we needed to clear through to Floor 7 to set up TwilightTide's Grave Sentinel analysis for Saturday's attempt, and in six weeks the CW I registration window would open and the full charter needed to be confirmed.

Six members. Six weeks. Tomorrow morning: Floor 7 attempt. Saturday after that: Beigong Yan's Floor 10 Heritage trial.

One thing at a time. All of them at once.

I closed the log and went to sleep.