179: Heaven-Severing

The character window finished refreshing at minute nine of Phase 1.

The new class tree was visible in full. I read it in the intervals between attack chains, during the boss's animation resets, in the two-second pauses that a practiced player learned to find and use — the moments between the end of one skill's animation and the beginning of the next, the quarter-seconds between Phase suppression pulses where nothing was happening to you and you could afford to look away from the center of the screen.

The tree was not what I'd expected from the transitional data.

The Blade Sovereign class had four primary nodes and eighteen secondary ones. The four primaries:

— *Heaven-Severing Edge*: the base attack capability, which had absorbed and transcended the Berserker's Crescent Moon Slash. The tooltip read: *A strike that severs not the opponent's armor but the distance between the blade's intent and the blade's effect. The edge lands where it was meant to land, at the moment it was meant to land, with the weight it was meant to carry.* The passive effect was described as "convergence of timing and intent" — in numerical terms, a 34% bonus to attack timing optimization, meaning the damage windows I'd been calculating manually since the first day of the game were now partially automated. The blade would find the window. It would land where it should land, when it should land there.

— *Sovereign's Reach*: the resonance, now in full expression. The radius expanded from 12 meters to 18 meters. The augmentation range increased from 4–22% in the transitional state to 8–31% at full designation. Duration extended from 2–6 seconds per interval to 4–12 seconds. TwilightTide at fourteen meters was inside the zone. The six members of the council's core formation were inside the zone. The tooltip: *The Sovereign's strikes are not individual actions. They are the formation's will expressed through the individual blade. Those who fight within the Sovereign's reach find that their own blades know the moment.*

— *Severing Presence*: a passive that replaced the old Warlord's Presence threshold mechanic. Where Warlord's Presence had been a self-buff triggered at HP thresholds — boosting my own damage output when I dropped below 40%, then 20% — Severing Presence was formation-wide. When the Blade Sovereign's HP fell below 60%, adjacent formation members received a resistance bonus. Not because the Sovereign was retreating, but because the Sovereign was holding. The tooltip: *The blade that refuses to yield becomes the axis around which the formation holds its shape.*

— *Void Severance*: an active skill with a 180-second cooldown. It wasn't in the Berserker tree at all — it was entirely new, the class's first original active skill in addition to the heritage strike tree. The tooltip read: *The blade severs the space between what is and what the enemy intended. Applies a disruption to one target's next three attack sequences, causing their damage windows to misalign with standard timing. Effect duration: 12 seconds.* Against a boss that had programmed Phase transitions, this meant the transition mechanics could be disrupted — the Phase change would still occur, but the boss's first three attacks in the new Phase would miss their optimized timing.

The boss's Phase transitions.

The skill could interfere with Phase transitions.

I kept attacking and finished reading the remaining fourteen secondary nodes in the two-minute window before Phase 2 initiated. The secondary nodes were augmentations and refinements of the four primaries — range extensions, cooldown reductions, condition triggers that would activate at character level milestones above Lv 80. I was at Lv 76. Four levels from the next tier of secondary unlocks. The secondary tree's upper half was visible but grayed out, waiting.

Phase 2 initiated.

Old Wolf's command channel: *Repositioning to void damage formation.*

Standard Phase 2 response. The guild moved into the four-adjacent cluster configuration they'd been running since January. I moved with them.

And then the resonance did something it hadn't done in the transitional state.

In the transitional state, the resonance intervals had been passive — they'd occurred when I attacked, and adjacent members had received the augmentation without perceiving it. No notification, no status icon, no subjective experience of the effect beyond Wenqing telling them afterward, in the combat log analysis, that their numbers had been higher than baseline. The formation had been hitting better without knowing why.

In the full designation, the resonance had a perceptible quality.

Not a system notification — there was still no popup, no status icon above anyone's health bar, no change in the game's visual presentation. But TwilightTide, at fourteen meters, sent a message on the healer channel at the first Phase 2 resonance interval:

*There's a rhythm now. I can feel the beat before the strike arrives. I know when to release the heal.*

She could feel it.

I read the message and kept hitting.

Zhu Yuhan, on the healer channel: *Same.*

Old Wolf, on the command channel: *Something changed. The formation is — tighter.*

It wasn't tighter in the numerical sense. The resonance wasn't changing the guild's combat positions or reducing the gap between formation members. The formation was tighter in the sense that Old Wolf described it: the guild's actions were oriented to a rhythm they could perceive and respond to, rather than receiving the augmentation silently and finding out later from a data table that something had been better. The full designation had added the perceptible quality. The formation could hear the blade now.

Not loudly. Not as something that demanded attention. But as a reference point — the way a metronome is present to a musician who's learned to play with one. Not the center of awareness, but the framework around which everything else organized itself.

Phase 2 cleared fourteen minutes faster than the previous week's record.

Phase 3: the soul-bind targeted Ironmark. Standard response — TwilightTide and Zhu Yuhan adjusted position and followed him through the bind sequence, holding the heal coverage while his movement was constrained. At 3%, the Sovereign fell.

Kill notification at ten PM.

Time: *3h 41m 12s.*

Previous record: *5h 23m 11s.*

One hour and forty-two minutes faster than the previous best. Not a marginal improvement — a structural one. The formation had been running at the same capacity ceiling since December. The class had changed what the ceiling was.

Old Wolf, on the command channel, in the post-kill silence: *Three hours forty-one.*

No one said anything else for a moment.

Then Wenqing, on the guild-wide channel: *Combat log analysis in progress. Stand by.*

***

Wenqing sent the post-session analysis at eleven PM.

Not his usual first-pass format — this was a different document, titled simply *Full Designation — First Session Record.*

He had formatted it like a specification document rather than an analysis report. Three pages.

The first page was the stat window — every capability from the class tree extracted and formatted as a clean table. Node name, effect type, numerical parameters, unlock conditions. It was a two-page table. He'd included a note at the bottom: *This is the first complete Heaven-Severing Blade Sovereign class record on any server. I've logged it with full documentation.*

The class was the first of its kind in the game's history. He'd known that since November. He'd put it in the header of the document anyway, because a record should say what it is.

The second page was the session analysis: the Floor 20 time, the Phase 2 efficiency increase, the Phase 3 performance. The aggregate formation output increase, attributed across the resonance effect and the two healers' response to the perceptible rhythm. The Void Severance cooldown had not been used — I'd read the tooltip in the middle of the session and had not deployed a new skill in an active fight without testing it first. That was a principle that had held since Lv 1: you tested skills before you relied on them.

The third page was a note to me directly:

*You've been running toward this class since October 2015. You told me about the Pioneer's Path in February 2016, which is when I started tracking it. In the eighteen months since then, I've logged approximately four hundred sessions of combat data, built seven sequential analytical models as the evidence changed, and watched you make decisions that the data supported before the data was complete.*

*The floor time tonight is 3:41:12. The previous record was 5:23:11. That's not an incremental improvement. That's a different game.*

*I'm updating the model for CW III.*

He'd been logging four hundred sessions. Seven models. The archive that had started with two sentences in October 2015 had, somewhere in the months I hadn't been counting, become the documentation of a two-year project — a complete record of what it looked like to build toward something no one else had built toward in the game's history.

I sent him one message: *Thank you.*

He replied: *The data is good. Keep running sessions.*

***

I logged out at eleven-thirty and went to the bench.

Wanqing was there. She'd been tracking the guild record tab. The April night was fully warm now — the spring that had arrived early had settled in with the confidence of a season that had decided it was staying. The maple was the same tree at the same bench and the thermos was the same thermos and it was April 13, 2017, and the class was complete.

"Three hours forty-one," she said.

"Yes."

She poured without asking. The steam in the April night.

"Tell me about it," she said.

I told her about the four primary nodes. The way TwilightTide had said she could feel the rhythm before the strike arrived — could feel the beat in the healer channel, the timing of when to release the heal cycle oriented to something outside herself. Old Wolf's observation: the formation is tighter. The Phase 2 clear time, fourteen minutes faster than the previous record. The final kill time, one hour and forty-two minutes below the previous floor.

She listened without interrupting. The kind of listening that wasn't collecting information point by point but letting the shape of what had happened arrive as a whole.

When I finished, she was quiet for a while.

The April bench. The full-leaved maple in the campus lights. A student on a distant path, walking home from a library or a seminar.

"He severed himself from the blade," she said.

She was finishing TwilightTide's sentence — the one that had trailed off in December, the one I'd thought about in January when she quoted it at the bench. The sentence TwilightTide hadn't been able to complete in December because the second half of it didn't exist yet.

She stopped, the way TwilightTide had stopped. Not because she didn't know the end of the sentence. Because the sentence was already complete. It had always been complete. The words just hadn't existed until tonight.

"The blade that knows the formation," she said.

"Yes."

She looked at the maple. "He walked east, out of the frame. Beigong Yan's training ground. The sequences. December 17. The transitional period. Tonight." She turned the thermos cup in her hands slowly, the way she turned things when she was thinking through the shape of something. "You've been carrying this for two years and it arrived as something that makes other people's timing better."

"The resonance augments the adjacent formation members."

"That's not what I said." She looked at me. The specific directness she used when she was correcting a formulation that was technically accurate but missing what mattered. "I said it arrived as something that makes other people's timing better. That's not the same as a buff with a percentage attached to it."

I considered that.

She was right that it was not the same thing.

Wenqing's analysis had described it in terms of status codes and augmentation percentages and coherence of timing. TwilightTide had described it as a rhythm she could feel before the strike arrived. Old Wolf had described it as the formation being tighter. None of those descriptions were the same as each other, and none of them were wrong, and they were all describing the same thing.

What Wanqing had said was a fourth description that contained something the others didn't.

"Yes," I said.

She poured again. The steam rose in the April night.

"April 13," she said. "One year."

"One year."

She held the cup. Not drinking it, just holding it — the familiar weight of the thermos cup that had been at this bench in how many seasons now. "Good," she said. "That's a good date for it."