143: The Bracket

The CW I bracket began at nine AM server time on Sunday September 6.

I'd slept four hours. Not insomnia — the bracket didn't make me anxious, it made me focused, which was a different physiological state that had a similar effect on sleep duration. I logged in at eight-fifty, ran a brief orientation check in the formation map, and logged out to eat breakfast before the eleven AM match.

The bracket format was double-elimination: every guild got at least two matches, with the loser's bracket running parallel to the winner's bracket until the championship stage. The Group B matches ran on two concurrent maps with a rolling schedule — six maps available per hour, the bracket progressing as matches completed. Severing Light's first match was Map 3, first session, eleven AM.

I had the guild's roster configuration established since Thursday. CW I battles were sixty-vs-sixty, drawn from each guild's registered roster, with substitutions allowed between rounds. The sixty I'd selected were the formation core — the council, the original forty-two from the pre-Black-Castle period, and the eighteen strongest recruits from the secondary window. TwilightTide and Zhu Yuhan were both in the sixty. Lin Yuxi had sent me a two-line message Thursday night, precise and characteristic: *I'm in the formation. Tell me the positioning and I'll hold it.*

I told her the positioning.

She was at the outer-support position. She'd hold it.

***

Round 1. Opponent: Ironclad Advance, Group B Seed 14. Sixty members, Floor 3 clear history, no CW I experience. Their formation had decent internal discipline for a guild their size — they'd been drilling the standard layered approach for the pre-bracket weeks and had produced a presentable result. Against a different opponent, that result would have been adequate.

The Map 3 environment was a highland terrain with two narrow corridor approaches and a central open zone that opened at the sixty-meter mark. The map design favored the formation that could control the corridors and hold pressure through the transition. The Scattered Fan was built for exactly that transition.

The match lasted eleven minutes and forty seconds.

The Scattered Fan Phase 2 — the variant we'd been running since the Redpeak Brotherhood match in March — was designed for opponents with layered tank formations, which Ironclad Advance ran because every guild their size defaulted to the same layered approach. It was the formation taught in the game's official combat tutorial, the one everyone knew, the one that made sense against the game's standard PvE content. It had a specific structural weakness at the first tank layer's gap points, and the Scattered Fan was designed to find those gaps and use them.

We'd drilled this pattern so many times that the execution was automatic. Old Wolf didn't need to give formation calls for the first four minutes because everyone was already in position. The sixty of us moved through the Phase 2 sequence without narration, the way a formation that's drilled something a hundred times moves: not thinking, just doing the right next thing.

TwilightTide ran the outer support position, maintaining range coverage while the DPS formation rotated in Phase 2 pattern. Zhu Yuhan held the center anchor. The gap-crossing sequence under the first tank layer took 4.1 seconds, same as it had in March, same as it had in every subsequent drill. Ironclad Advance's formation fractured at minute seven — the gap-cross had found the structural weakness and the fracture propagated through the formation the way structural fractures always propagate, faster than the formation could respond — and we spent four minutes eliminating the scattered members.

Final count: 60-0.

The bracket feed updated. First round win. Loser's bracket path for Ironclad Advance.

***

Round 2 came at two PM — Amber Crest Guild, Group B Seed 6. 60-member independent, Floor 5 clear history, two CW seasons in the previous year. They'd placed ninth in the Group B bracket last season, which meant they knew the format and had prepared specifically for this bracket, not just for general combat.

Their approach was different from Ironclad Advance: no layered tank formation, but a zoning defense that placed their healers at protected positions behind the main formation, relying on sustainable healing output rather than tank HP pools. The zoning approach was harder to exploit with the Scattered Fan because there was no tank layer to go under. Attacking a zoning formation directly meant taking sustained fire from the healing-supported main body, which was a slow attrition problem that favored the zoning side.

TwilightTide had flagged the zoning approach as the most likely Amber Crest configuration on Friday, based on their previous CW season records. Old Wolf had confirmed it on the pre-match scouting. Neither of them had said "I told you so" because neither of them operated in that register — the analysis was for using, not for crediting.

Old Wolf, on the formation channel before contact: *Zoning approach. Fan disperses to mid-range engagement. Wenqing's read is correct.*

The Scattered Fan adapted: instead of Phase 2 gap-crossing, we ran Phase 1 at extended range — the 3-to-2 formation compressing their zoning space without committing to the under-layer gap. The compression forced their healers out of their protected positions to maintain formation coverage, which moved them from protected positions to exposed ones. Once they were exposed, the mobile DPS arc had angles.

The match ran nineteen minutes. Final count: 60-47.

We'd taken thirteen casualties — all DPS, all from the forward arc that absorbed Amber Crest's sustained fire while the mobile arc worked around their flank. Extended-range engagement at their quality meant absorbing real damage for real time. Thirteen casualties was the cost of not having a cleaner counter-formation for their specific approach.

Old Wolf assessed the thirteen casualties post-match in the council channel with the tone he used for observed facts: expected losses from a zoning engagement without a dedicated counter-formation. Not mistakes. Costs. There was a difference. He listed the thirteen by name and role, noted which mechanic had caught each one, and said: "Next round." No further comment. The brevity was the analysis — if there were errors he would have named them. The absence of error-names meant the costs were accepted costs.

"Next round," he said.

***

Round 3 was Monday morning — the Group B coalition sub-unit. Seed 8 in the bracket but significantly stronger than the seed suggested: they'd drawn weak opponents in Rounds 1 and 2 and had advanced without meaningful opposition. Wenqing had pulled their public match records from both rounds. The sub-unit ran the coalition's standard formation structure but had depth that the weaker Seed 14 and Seed 11 opponents hadn't been able to test.

They'd prepared for Scattered Fan variants. Wang Jian had instructed every sub-unit to study our public match records from the March exhibition series. Wenqing had flagged this in August — the coalition's internal communication channels had shown increased traffic around the Phase 2 timing analysis threads on the server forum.

"They've prepared for Phase 2," I said, at the Monday morning formation review.

Old Wolf: "Then we don't run Phase 2."

"We run Phase 3."

Phase 3 was a variant we hadn't deployed in any public match. It was the logical extension of Phase 2 — instead of going under the first tank layer, Phase 3 went *through* the formation entirely, with the forward arc exploiting the specific gap that formed when a prepared formation tightened its first layer in response to watching for Phase 2. The tightening of the first layer was the Phase 2 counter. The tightening created a gap at the formation's center that Phase 2 didn't use. Phase 3 used it.

We'd drilled Phase 3 against practice formations in the Iron Hills since April. Six months of drilling a technique that we'd held in reserve. No one outside the guild had seen it in a competitive context.

"They'll adjust when they see it," TwilightTide said. Her voice was even — not alarmed, just calculating. She always sounded most certain when she was talking about timing.

"They'll need three minutes to adjust," I said. "The Phase 3 opening window is the first ninety seconds of contact. We land the through-cross in the first ninety seconds and they won't have adjusted before we're already inside."

"Ninety seconds is optimistic," Old Wolf said.

"One-twenty to one-fifty," I said. "More realistically. They have to see it's not Phase 2, identify that it's a different approach, communicate the counter-adjustment down the formation chain, and implement a counter that none of them have drilled for. At their communication speed, that's a hundred-twenty to a hundred-fifty seconds."

"Still inside the window."

"Yes."

The Round 3 match started at ten AM Monday.

The coalition sub-unit deployed their Phase 2 defense configuration immediately — first tank layer compressed, flanks tightened, the counter-formation they'd been drilling for weeks. Exactly what we'd expected. They'd arrived prepared for the fight they'd studied.

The Phase 3 through-cross executed at sixty-four seconds after contact. Our forward arc moved through the compression gap at the formation's center — the gap that the compressed first layer had created, that no one in the sub-unit had been watching because no one was supposed to use it. The arc was inside the sub-unit's formation before their communication channel registered that the approach wasn't Phase 2.

Inside the formation, their depth advantage disappeared. Formation depth is a front-pressure advantage. When the pressure is already inside, depth becomes a liability — more bodies in a compressed space with confused positioning. Severing Light's formation precision was built for coherence under chaos. The sub-unit's formation depth was built for sustained external pressure. Inside the formation, we had the better attribute by a significant margin.

The match ended at seventeen minutes. Final count: 60-31.

The sub-unit's healer line had held longer than I expected — they had their healers in overlapping coverage positions, a refinement of the zoning approach that Amber Crest had used, applied with more depth. TwilightTide had spotted the overlap arrangement at minute four and repositioned accordingly, shifting her coverage to peel the overlap apart. The healer line collapsed at minute nine.

Wenqing, in the post-match analysis channel: *Wang Jian observed the Round 3 match in the spectator channel. He watched the Phase 3 execution. He's adjusting his championship bracket sub-unit's formation plan accordingly.*

Of course he was. Wang Jian was always watching, always adjusting. That was the correct response to new information and he was a correct operator.

We advanced to the championship bracket as Group B's top qualifier.

Wenqing's updated probability: *Top-five confirmed probability: 84%. Top-three: 51%.*

The bracket draw for the championship stage was Monday evening. Our championship bracket opponents would be Group A's representatives.

The merged coalition unit had advanced from Group A.

I looked at the championship bracket structure and thought about Phase 4, which was the variant we'd drilled six times in the Iron Hills in July and August and had not yet used in competition. Phase 4 was designed for a specific formation condition that required one of the previous three phases to have been already observed by the opponent. The merged coalition unit had now seen Phase 2 — from the sub-unit's records — and Phase 3 — from today's spectator channel. They'd seen two phases. Phase 4 was the third thing they hadn't seen.

Wang Jian had made his adjustments. We had one more adjustment waiting.

I sent TwilightTide a private message: *Phase 4 tomorrow if they run open spacing. You'll recognize the geometry.*

She replied in eleven minutes: *Already noted. I'll be at the outer ring position from the start. If the spacing changes mid-match I'll signal.*

We'd drilled Phase 4 six times. She'd been at the outer ring position for all six. She knew the geometry. The signal she'd used in the drills was three short protocol pulses on the healer channel — an audio pattern distinctive enough from the combat cadence that it would register without requiring her to break her healing sequence. We'd established that in the first drill and she'd never changed it. When something was right the first time, she kept it.

*Sleep,* I sent back.

*After the match analysis,* she replied.

There was nothing to add to that. I sent the championship bracket summary to the council channel and closed the laptop.

Tuesday. The championship bracket. Phase 4 in reserve, if they gave us the opening.

They would give us the opening. Wang Jian's open-spacing counter was the correct counter to phases two and three. It was also the condition Phase 4 required.

I went to sleep.