180: Server Record
The kill notification had been public.
Every guild-first kill on a dungeon floor posted automatically to the Tianlong server's achievement board: floor number, guild name, clear time, date and time of kill. The April 13 Floor 20 kill had posted at ten PM with the time of 3h 41m 12s. The previous record — the guild's own, from November — had been 5h 23m 11s.
By noon on April 14, the achievement board thread had three hundred comments.
I read through the first hundred during the morning session break. The thread pattern was recognizable from the November record: the first hundred comments were calculation attempts. Server players reverse-engineering what had changed in the guild's formation to account for the time improvement. A Floor 20 clear required all bosses cleared in sequence, which meant the full five-Phase structure, which meant any improvement came from within a known combat framework. The structural constraints were the same as November. The question the thread was working through was what had changed inside the structure.
The time improvement was a 32% reduction in clear time. That was a large enough improvement that the initial thread assumption — formation efficiency, optimized positioning, practice accumulation — failed the math quickly. Formation efficiency gains didn't produce 32% improvements in runs that were already clean. Something categorical had changed.
The second hundred comments were the forums catching up — players arriving from game community boards who'd seen the achievement board notification and were now reading backward through the thread trying to find the answer the first hundred had arrived at.
The kill-cam for guild-first clears posted automatically. The Floor 20 kill-cam only showed the final thirty seconds: the Phase 3 kill sequence. The kill-cam showed Bladeless's position in the formation, the final strike, the Abyssal Sovereign's fall. It showed the kill-cam's timestamped header. It didn't show the combat log.
Wenqing had posted the combat log.
Not all of it — he'd posted a four-page extract that showed the resonance intervals, the adjacent output spikes, and the class designation badge in the corner of Bladeless's character window. The designation badge, which players could see in combat logs when they were shared publicly, now showed a string that hadn't existed on any server before April 13: *BLADE SOVEREIGN — HEAVEN-SEVERING CLASS.*
That was the classification string. No parenthetical. No TRANSITIONAL in parentheses, the way it had appeared since December. The full designation, in the permanent achievement record.
The third hundred comments were the game community identifying the class. Someone had done careful work — a twelve-paragraph post that cross-referenced the Pioneer's Path quest chain with the Heritage Fragment acquisition sequence and arrived at a plausible origin story for the class. They'd gotten the broad structure correct: the Heritage Blade quest chain, the four Fragment types, the Floor 20 transition chamber. They'd correctly identified the approximate level range required for the Pioneer's Path entry point.
They hadn't been able to find the Beigong Yan NPC. They hadn't found the Hall of Echoes corridor, which was the specific junction in the quest chain that required knowledge of the server-reset schedule that Wenqing had derived analytically in October 2015. Without the Hall of Echoes, the fragment sequence didn't connect. Without the connection, the class path terminated at the acquisition of three of the four fragments and didn't proceed.
The twelve paragraphs were impressive. They were also a good demonstration of how much careful work could get right without being able to reproduce what it had analyzed. Getting the structure right without the key junction was like having the map but not the door. You could describe the building correctly from the outside. The description was accurate. It was also incomplete in the specific way that a description without the door handle was incomplete. The person who wrote those twelve paragraphs would find the exterior of the building correctly and find no way in.
By the end of the week, the Tianlong server's achievement database had three permanent records under the guild's name:
— *First guild to clear all 20 floors of the Black Castle dungeon* (Nov 26, 2016) — *First player on any server to activate the Pioneer's Path Hidden Class* (Apr 13, 2017) — *Current fastest Floor 20 clear record: 3h 41m 12s*
The third record would change. Other guilds would study the combat log and begin replicating what they could. The resonance mechanic was documented now — the intervals, the distance relationship, the augmentation percentages. Any guild that built a formation infrastructure around the right class combination could theoretically approximate the effect. The first two records were permanent because no one else would be first to clear all 20 floors or first to activate the Pioneer's Path. Those were singular events. The time record was not.
I'd understood the distinction from the moment I started building the guild. The permanent records were not about time. They were about sequence — you could do something first or you couldn't, and once someone had done it first, that position was closed. Every other kind of record was repeatable, was subject to revision, was the kind of thing you held until someone held it better. The permanent ones were a different category of fact.
***
Wang Jian's response took six days.
On April 19, a post appeared on the server forum from a Tianxia Coalition analyst account — not Wang Jian directly, which was the way Wang Jian preferred to operate in the forum space. He treated public posts as official communications and sent his lieutenants to make them. The post was titled: *Bladeless's Class Change — Tactical Analysis and Competitive Implications.*
The analyst had done good work. Considerably better work than the twelve-paragraph community post. They'd identified the resonance mechanic from Wenqing's posted log, calculated the radius and augmentation parameters correctly using the distance table Wenqing had included, and derived the formation composition implications — which guild positions would be affected, what the practical change in aggregate output represented for sustained combat.
The section on competitive implications concluded: *The Heaven-Severing Blade Sovereign class changes the calculus of CW III. A guild with this class anchor and the formation infrastructure to use it represents a qualitatively different competitive threat than the same guild without it.*
They were right. It was a thorough and accurate analysis. The analyst had done what good analysts did: used the available information to build a correct model of the thing they were looking at. The available information was what Wenqing had posted. Wenqing had posted exactly what he intended to post.
The post hadn't been written for the general server community. The general community didn't need a six-section analytical breakdown with supporting calculations. The post had been written for Wang Jian, presented in the format he'd read it in — the format of a competitive intelligence briefing, with the structure of a formal recommendation, delivered through a public channel because that was the format Wang Jian used when he wanted his response to be visible without being direct.
He wanted us to know that Tianxia had read the combat log carefully. He wanted the server to know that Tianxia considered the class activation a competitive development worth analyzing. He was establishing, publicly, that he was prepared for whatever the April 13 activation represented.
I forwarded the post to Wenqing.
He replied in nine minutes: *They've correctly characterized the resonance mechanic. They don't have the Void Severance data because I didn't post it. Recommendation: hold that capability private until the competitive context where it will have the highest yield.*
The Void Severance tooltip had been visible in my character window since April 13. No one in the session had mentioned it in any channel Wenqing had posted. I hadn't used it during the session — I'd read it mid-fight and had filed it as something to test, not something to deploy without understanding it first. Wenqing had apparently noticed the same thing: he'd seen the skill in the character window data, had recognized it as something the public log didn't need to contain, and had edited the extract accordingly.
The council agreed unanimously. No discussion required. They understood the logic.
Void Severance was not mentioned in any guild communication for the next three months.
***
Bai Yueran sent a message on April 17.
Not in-game — she sent it through the guild's official contact channel, which she used for guild-to-guild communication. The message was six sentences.
*Congratulations on the Floor 20 record. We've been watching the achievement board since November. The class activation data is impressive, as is the analytical documentation your team published. Your guild's approach to information sharing — public record, honest format — is something we've tried to emulate in our own documentation.*
*I'm planning to be in Hangzhou in late May for an academic conference. If it's convenient, I'd welcome the opportunity to speak in person.*
Six sentences. The first four were professional courtesy. The last two were the signal.
Bai Yueran — MoonShadow Guild, top-ten server ranking, thirty-four members of her analytical layer. She ran the guild from Beijing. She was a third-year PhD candidate in computational mathematics at Peking University — that had been in her guild's public registration since the founding, the way our guild's founding charter had been in our public documentation. She came to Hangzhou for academic conferences approximately twice a year. Late May was not an unusual time for a conference in the computational mathematics field.
She had said it was convenient if it suited me. She had said she'd welcome the opportunity. Both of those phrasings were precise. She wasn't asking; she was making an opening available. There was a version of the message that would have been a direct request. She hadn't sent that version. She'd sent the version that gave me the choice without putting the weight of a request on the choice.
I read the message twice. Then I looked at the window of the dorm room for a moment — the April campus outside, the afternoon light, the paths between buildings where people moved at the mid-semester pace that was neither the early-semester ease nor the late-semester pressure but the middle state where things simply ran.
I sent to the official channel: *Congratulations received. Late May is workable. Send the conference schedule when it's confirmed.*
She replied four hours later: *Thank you. I'll send it when I have the dates.*
Four hours. She'd had the reply ready and had held it for four hours before sending. That was its own kind of information.
***
I told Wanqing about Bai Yueran's message at the Thursday bench.
She was on the problem set — the spring seminar's fourth unit, which she'd mentioned had moved into the applied sections after the heavy notation work of the third. She was reading rather than working through notation, which meant she had more attention available than the notation pages allowed.
She read the section of the message I recited from memory.
"Late May," she said.
"Academic conference. She comes to Hangzhou twice a year for them."
"And she sent a message after the Floor 20 record."
"Yes."
She turned a page. "She's been watching since November. The November kill was the first Floor 20 clear — the first record. The April kill was the class activation. She waited until the second one."
"Both times she acknowledged publicly — the November achievement was on the server board, the April one the same. She waited until the April one to contact directly."
"Because the April one changed the calculus." She looked at the problem set. "The Tianxia analyst said it. *A qualitatively different competitive threat.*" She turned another page. "What do you want to say to her in May."
"The same thing I'd say to any guild leader I was meeting with. The things that are relevant to what we're each trying to build. What our guilds have in common and where they differ."
"You've been in the same server bracket for two years."
"Yes. In different positions."
She looked at the maple — now in full spring green, the same tree at the same point it had been last April. The April quality of a campus maple that had completed its growth cycle and was now simply being itself. The tree held its shape, session to session, bench visit to bench visit, until October came and it became something different for three months and then came back.
"You knew her before. In the previous timeline."
It was not a question. She didn't ask the direct questions about the previous timeline. She tracked when information implied it and made the inference, and when she made the inference she stated it without asking for confirmation. She'd been doing this since the first time it came up, sometime in the second summer at the bench.
"Yes," I said.
She turned another page. "All right," she said.
And went back to the problem set.
That was the full extent of the conversation about Bai Yueran. She'd taken it in, noted it, filed it in whatever structure she used for the information I gave her that came from places she didn't ask about, and returned to the seminar work. The same way she'd handled every piece of information I'd given her that involved the previous timeline's shape — receiving it without questioning the source, incorporating it into the model. She didn't use the information frivolously. She used it when it was relevant to a decision that needed making. And then she put it away and went back to the problem set.
The bench was the bench. The maple was the maple. The thermos was the right temperature. The seminar problem set was the seminar problem set, turning toward its applied phase after the heavy notation work.
I watched the campus for a while after she went back to reading. The April afternoon had the settled quality of a day that had found its temperature and stayed there — warm without being warm enough to be noticed, the kind of day the campus ran well in. In a week the exam period would start and the campus would change its character, the settled pace compressing into the exam-period urgency that was its own kind of weather.
Before then, the bench.