Round two ran Thursday.
His match was the third of the session, midmorning. Opponent: Priya Lal, C-rank, close combat — the fighter Rowan had categorized as tier two, excellent foundation work through Instructor Vane's advanced cohort. The analysis had flagged her as technically precise, less effective under sustained pressure, with a preference for controlled spacing that she used to set up specific entry angles.
He arrived early enough to watch her warm-up from the preparation corridor.
Priya Lal moved like someone who'd been training since before the Awakening — the quality of martial background that carried over through class development rather than starting from scratch with the class's native movement patterns. Her channel architecture was visible in the warm-up: clean, high-density for C-rank, with the specific signature of someone who'd been running an integrity-first approach to development. Not unlike what Castellan had built in him over the past months.
The comparison was useful.
He spent forty minutes reviewing the Cruz preparation framework, then changed into his tournament kit and went through the channel warm-up sequence.
The shoulder was taped. The bruising wasn't affecting range of motion — it was affecting the way the impact feedback registered, which was a different problem. If he took a hit to the same site, the pain response would spike faster than the damage warranted. He noted this and adjusted the approach: fight more positionally, reduce the exchanges where the right shoulder carried load.
The match against Priya Lal lasted seventeen minutes.
Longer than either of them had probably wanted. She was good — the technical precision was real, not a projection from the footage, and her spacing control made the fight a series of problems to solve rather than a sequence of exchanges to win. Every time he found an angle, she'd already adjusted the spacing to close it. Every time she created an entry opportunity, he read the follow-through and redirected.
Seventeen minutes of problem-solving.
He won it in the end through attrition. Her foundation work was excellent but her channel efficiency under sustained load was the gap Rowan's analysis had flagged — she was burning more architecture than he was to maintain the same spacing control, and by minute twelve she was managing rather than controlling. He pressed that gap and she ran out of margin.
She shook his hand after the examiner's call. Her grip was firm and she didn't say anything. He respected that.
"Clean win," Castellan said, finding him in the corridor after. "The shoulder held."
"Yes." He rolled it. Stiff but functional. "The spacing management she was running — the channel cost was higher than I expected for that technique."
"Vane's program teaches spacing control as a foundation skill. It's technically correct and expensive." She made the note. "You exploited the cost differential. That's the right approach in a long match." She paused. "You'll see that approach again in a later bracket if Cruz advances."
He looked at her.
"Not spacing control. The same principle — a technique that's expensive at the channel level." She looked at her tablet. "The fracture absorption. It's not a cheap technique. If the fight goes long enough, he's paying for every absorbed impact."
He'd considered this. "The charge time between absorptions was about three seconds in the first-round fight. If I can maintain contact without giving him the charge time—"
"You remove the absorption window." She paused. "But maintaining contact with someone who can fracture your impact at the channel level is dangerous if you're the one applying the contact."
"I'm not applying direct force in the contact window." He'd been working through this. "If I control the position rather than the power — if I make the contact about geometry rather than impact—"
"Positioning rather than striking." She was thinking. "You use the contact window to control his geometry. Force him to respond to position rather than impact. He absorbs strikes, not positions."
"Yes."
She looked at him. "That's an interesting framework." A pause. "I need to think about whether it's correct."
"So do I." He looked at the corridor. "I have about three days."
---
Cruz's round two match was the seventh of the session.
D-rank certification against a C-rank opponent — the bracket seeding had put him against Yuen Min, one of the tier-two fighters Rowan's analysis had flagged. Multi-element class, high-output offensive, the documented issue with sustained combat.
It didn't matter that Yuen Min had a sustained-combat issue, because the match lasted ninety seconds.
Yuen Min opened with an elemental volley — the high-output opening that his record showed as his standard opening move. Standard anti-ranged response would be to create a barrier or use mobility to close the range. Cruz walked through the volley.
Not blocked — walked through. The elemental impacts landed and the force fractured at the channel level before reaching structural tissue. Cruz's channel architecture absorbing the impact cost and returning the force vectors into his own channel loop rather than dissipating them.
He was charging.
The fracture absorption wasn't just defense. It was fuel. Every impact Cruz absorbed fed into a channel compression that increased his output pressure for the next exchange.
Yuen Min, seeing his volley fail to slow the advance, switched to a different approach — barrier construction instead of direct strike. A correct tactical adjustment that arrived about four seconds too late. Cruz hit the barrier with the accumulated charge from three absorbed volleys and the barrier didn't hold.
Ninety seconds.
The observation deck went quiet for a moment after the examiner's call. Not the confused quiet of an audience that didn't understand what they'd seen — the specific quiet of two hundred people processing the same understanding at the same time.
He was in the secondary preparation room watching the display feed. He played back the relevant seconds twice, watching the moment after each absorbed impact where Cruz's posture shifted infinitesimally — the charge accumulation expressing itself in body mechanics so subtle that a real-time viewer would miss it entirely.
The fight wasn't about the exchange. It was about the accumulation.
Every impact Cruz absorbed increased the pressure available for subsequent output. He was a battery that charged from being hit. The more force directed at him, the more he had to return.
He thought about the geometry framework he'd discussed with Castellan.
If he controlled position rather than applied impact, he denied Cruz the input needed to charge. But position control required contact, and contact without impact meant wrestling, grounding work, weight control. The channel architecture for that kind of sustained positioning pressure was different from striking architecture — it required a sustained low-level channel load rather than bursts.
It was possible. It was also something he'd never trained for specifically.
He sent Rowan a message.
*Cruz's round two. He's accumulating from absorbed impacts. The fracture isn't just defense — it's an energy source.*
ROWAN: *I saw.* A pause. *Illen's paper didn't describe the accumulation aspect. The paper described the fracture technique as defensive. Cruz is using it as both.*
*Either the paper was incomplete or Cruz has developed the technique beyond what the paper described.*
ROWAN: *If Maren Voss has been working with Illen's full research set — including unpublished notes — there may be material beyond what's in the public archive.* A pause. *Kael. I've been trying to reach Professor Illen through the Association's faculty directory. He's not responding to the standard contact form.* Another pause. *He's listed as on research leave. I found a research station address in the northeast of the Ravenscrest district.*
He stopped.
Northeast.
*What elevation is the research station.*
ROWAN: *It's in the elevated terrain north of the city. The canyon research district — the Association maintains several monitoring stations on the ridge system there.* A pause. *The elevation is significant. Two hundred meters above the city proper, approximately.*
He put the tablet down.
Two hundred meters. Northeast of Ravenscrest. The bearing Iris Voss had traced — northeast, elevated.
*Is Illen at the research station.*
ROWAN: *The registration shows him as on-site.* A pause. *Kael. If the anchor marking's directional orientation is pointing at Illen—*
*I don't know that it is.* He picked the tablet back up. *The bearing is northeast and elevated. Illen is northeast and elevated. That's two data points. It's not a conclusion.*
ROWAN: *It's enough to be interesting.*
It was enough to be interesting. He looked at the display feed, where the next match was being called.
He couldn't go to the research station until after the tournament. He had three rounds left. Then he would figure out what the anchor marking was pointing at and whether Illen was any part of it.
*Note the location. I'll follow up after the tournament.*
ROWAN: *Noted.* A pause. *Round three runs Saturday. Your projected matchup, based on the bracket, is Cruz.*
*Yes.*
ROWAN: *How's the preparation.*
*Incomplete.* He looked at the corridor. *Which is accurate for a fight against someone running a technique no one's seen in combat until two days ago.*
ROWAN: *For what it's worth — he's never fought someone at the C-rank architecture threshold who uses geometry rather than impact.* A pause. *If your framework is correct, he hasn't needed to adapt to it yet.*
*He hasn't. That doesn't mean he can't.*
ROWAN: *No.* A long pause. *Kael. He's good. I want to be clear about that. He's genuinely good — not just an unusual technique. His spatial reading in the Yuen Min match was precise. He knew where Yuen Min was going to build the barrier before the decision was made.*
*How.*
ROWAN: *I'm not sure. The fracture technique may include a proprioceptive component I haven't fully characterized — the force vectors he absorbs might give him environmental information the standard class set doesn't have access to.* A pause. *He might be reading you through your own contact points.*
He sat with this.
If Cruz's force absorption gave him environmental read data — if every impact he absorbed told him something about the attacker's position, weight, and next movement — then the geometry framework was also compromised. Control the position rather than apply impact. But if position control also fed Cruz information—
He needed to think.
He closed the display feed and sat in the quiet preparation room and thought.
The fight was Saturday. He had thirty-six hours.
---
Friday evening.
He'd been at Castellan's facility for three hours — not a standard session, a working session. The two of them working through the geometry framework against a training partner Castellan had brought in: a B-rank hunter named Fell who specialized in grappling-pressure techniques and who, at Castellan's instruction, had been told to absorb every contact without fighting back.
Fell had been useful as a dummy. The geometry framework worked when Kael was controlling position against a passive partner. The channel load for sustained positioning was manageable, the approach was geometrically clean.
The problem was the information component Rowan had flagged.
Even against a passive Fell, Kael could feel the feedback from every positioning contact — the pressure points where his own channel architecture was interfacing with Fell's. Position control was contact. Contact was information. If Cruz was reading through contact points—
"You're hesitating at the contact phase," Fell said. He was permitted to speak; Castellan had been running it as a diagnostic session as much as a training one. "Not physically. Something in the channel layer."
"I know."
"What's the problem."
He looked at Fell. Not a fighter with unusual techniques. A grappling specialist whose whole methodology was about reading contact.
"How do you handle a partner who's reading your contact points," he said.
Fell looked at him. "Is this a general question or a specific one."
"Both."
Fell thought about it. "You give them noise," he said finally. "You create contact that carries false information. Pressure that doesn't lead where the pressure suggests it's going." He paused. "If someone reads contact data, the counter is contaminating the data source." He looked at Kael. "Can you deliver impact that's directionally misleading at the channel level."
He could. The channel architecture at the current threshold could produce impact vectors that diverged from their surface direction — a strike that felt like it was coming from the right while the actual force traveled from the left. S-rank technique variants. He'd been using the simpler D-rank versions in standard training.
"Yes," he said.
"Then your contact in the geometry framework doesn't just control position — it delivers false read data. Every time he absorbs your positioning contact, he gets information that's wrong." Fell looked at Castellan. "If the technique is reading contact information, contaminate the contact."
Castellan had been listening. "That changes the approach. It's not just geometric control — it's information warfare at the contact interface."
"Yes." He thought about Cruz watching the Yuen Min match's positioning approach. The precise reading he'd shown. "He's good at processing contact data. The question is how fast he can adjust when the data is unreliable."
"We don't know," Fell said. "That's the part you'd have to find out in the match."
No previous footage. No record of Cruz fighting someone using contaminated contact data. He'd be reading Cruz's adjustment speed in real time.
That was what the unreferenced dungeon had been for. Read what's there. Respond to what's there.
He ran the framework against Fell for another forty minutes with the contaminated contact approach.
It worked. Not cleanly — Fell's grappling experience meant he was difficult to deceive even with unreliable data, and there were moments where Kael's contamination pattern was itself readable. But it worked often enough to create openings that the passive positioning approach wouldn't have created.
"Enough," Castellan said. "We've found the framework. The rest is execution." She looked at him. "Eat. Sleep. Don't analyze it anymore tonight."
"I'll analyze it when I wake up."
"You'll stop by 2200." She held his gaze. "The channel architecture needs recovery time. You've been running it at near-peak load for four days. If you go into Saturday's match on depleted architecture—"
"I know." He flexed his hands. "I'll stop by 2200."
"Good." She turned to Fell. "Thank you."
He found Elara outside the facility, which meant she'd been waiting. Not impatient waiting — she was looking at the canal from the building's front step, turned partly away, in the attitude of someone who was comfortable with the waiting.
He came out and she turned and looked at him.
"Cruz," she said.
"Saturday."
She looked at him for a moment. "Are you going to tell me to stop worrying."
"No." He started walking. "You're not worried. You're assessing."
She fell into step beside him. "And."
"And you should be. He's very good." He looked at the canal. "The technique I've been working around has more layers than the paper described."
"But you have a framework."
"I have a framework that's never been tested under real conditions."
She was quiet for a moment. They walked the canal path at the pace of the evening — not hurrying, the city doing its nighttime settling around them. Three blocks from her building, two from his.
"The first time you ran a dungeon in the original—" She caught herself. "In the other version. Did you have a framework for it."
"I had training and instinct."
"This is training and instinct with an additional layer of analysis." She looked at him sideways. "That's better than the original conditions."
He looked at her. "When did you start making the optimistic argument."
She held his gaze for a moment. "I've always been making it. You just don't usually let me finish." She looked ahead. "You're going to have a hard fight on Saturday. You're also going to figure out what to do in it."
He looked at the canal path ahead.
"Yes," he said.
She left him at her building's door — not coming in this time, the session had run too late and she had an early evaluation in the morning. He walked back to his apartment in the canal district quiet and went through the 2200 stop point exactly, circuit work cooling down rather than running up, and was in bed by 2300.
He thought about the fight for about twenty minutes. Then his body stopped caring about his analysis and he went under.
