Kael sat at his desk with four names written on a piece of paper and a timeline that was already wrong.
The shoulder had stopped running hot sometime during the night. Four days of recovery, Castellan had said, and she'd been right — the integration cluster had settled back to baseline, the catalyst's passive load folding into the architecture's rhythm like a second heartbeat he was still learning to ignore. He could feel it now without focusing: a steady draw on the channel density, not painful, just present. The Void integration threshold was five weeks out, maybe four. The catalyst didn't care whether he was ready.
He looked at the paper.
Four names. Four resources. Four things that, in the original timeline, had turned ordinary people into the specific weapons that had killed him.
Dorian's hidden class quest. The Shadow Throne — a dungeon that had appeared in month twenty-four of the original Awakening timeline, in the industrial quarter's abandoned foundry district. Dorian had found it through Foundation contacts who were tracking anomalous spatial signatures. He'd gone in solo, spent three days inside, and come out with a class evolution that pushed his Shadow Assassin architecture past the C-rank ceiling in ways the standard development track couldn't replicate. The quest had given him the Shadow Sovereign subclass — the thing that made him untouchable in enclosed spaces, the thing that let him move through walls and floors like they were water.
Month twenty-four. Four months from now.
Elara's elemental grimoire. The Ashcroft Archives — a private collection maintained by a retired A-rank hunter in the western district. In the original timeline, Elara had gained access through a mentorship program in month twenty-two. The grimoire had contained three elemental convergence techniques that no other resource in the region could replicate. Without them, her elemental class development had a hard ceiling at B-rank. With them, she'd reached A-rank by year four.
Month twenty-two. Two months from now.
Marcus's divine artifact. The Temple of the Mending God — not a dungeon, exactly. A sanctified zone that appeared in month twenty-six, in the ruins of an old cathedral in the eastern outskirts. Marcus had been the first healer to enter, guided by Church contacts who'd detected the sanctification event. The artifact inside had bonded with his healing architecture and given him the Divine Healer class evolution. Without it, his healing class was strong but conventional. With it, he could reverse biological processes — including the reversal that had accelerated Kael's death.
Month twenty-six. Six months out.
Sera's sponsorship deal. The Vanguard Guild — one of the three major guilds that would form by month twenty-three. Their recruitment drive had targeted tanks with defensive class specializations. Sera's Iron Fortress architecture had caught their attention. The sponsorship gave her access to guild-exclusive training facilities, equipment, and — critically — the debt restructuring program that Dorian later used as leverage. Without the guild sponsorship, Sera's family stayed in debt. Without the debt, Dorian couldn't blackmail her.
Month twenty-three. Three months.
He'd written the timeline on the paper too, and that was the part that was already wrong.
These dates were from a world where Kael Ashford hadn't come back. Where the Greystoke Vault's catalyst had gone to Dorian. Where a divergent dungeon in the outer district hadn't appeared because nobody had been making the changes that generated timeline pressure. Where the channel architecture system was running on a track that no longer existed.
The dates might hold. Some of them. The spatial signatures that generated the Shadow Throne dungeon were probably independent of Kael's interference — deep geological channel formations that operated on their own schedule. The Ashcroft Archives were a physical collection maintained by a living person, so access was about timing and connection, not system events. The Temple of the Mending God was a sanctification event tied to the Church's influence cycle, which Kael hadn't directly affected.
Or they might not hold. Any of them. The divergent dungeon in the outer district had proven that Kael's changes could generate entirely new spatial formations. If the system was responsive to large-scale behavioral changes — and Rowan's analysis said it was — then the original timeline's dungeon schedule was a guideline, not a guarantee.
He was planning a campaign against targets that might not be where he expected them, on a timeline that might not match, in a world that was changing faster than his knowledge could track.
He picked up the paper and folded it in half.
Good enough. He'd work with what he had.
---
Rowan's first message came at nine.
ROWAN: *I've been running the divergence model against the four resource targets you gave me yesterday. Short version: two are probably stable, two are not.*
*Which two.*
ROWAN: *The Ashcroft Archives and the Temple of the Mending God are both anchored to pre-existing physical infrastructure. The archives are a private collection — the divergence model doesn't show significant probability of that collection moving or being destroyed before month twenty-two. The temple's sanctification event is tied to the Church's influence cycle, which has been remarkably resistant to timeline pressure. Both of those should be accessible on roughly the original schedule.*
*And the other two.*
ROWAN: *Dorian's class quest dungeon and Sera's guild sponsorship are both system-responsive. The Shadow Throne's spatial signature depends on foundry district channel density, which has been fluctuating since the outer district divergent dungeon appeared. My model puts the Shadow Throne's appearance window anywhere from month twenty-two to month twenty-six — that's a four-month spread instead of a single-month prediction.* A pause. *The Vanguard Guild's formation timeline is even less stable. Guild formation depends on hunter population density and Association regulatory decisions, both of which have been affected by the things you've changed. The guild might form earlier, later, or not at all in its original configuration.*
Kael read the analysis twice. The two targets that mattered most — the ones that directly empowered the people who'd killed him — were the ones he couldn't pin down.
*Dorian's quest. If the appearance window starts at month twenty-two, that's two months. Can you narrow it.*
ROWAN: *Not with the current monitoring resolution. I'd need to place observation equipment in the foundry district to track the channel density fluctuations in real time. Which means getting access to the foundry district, which means filing with the Association's spatial monitoring division, which means creating a record.* A pause. *Or I could do it without filing. But that's a different kind of risk.*
*Do it without filing.*
ROWAN: *Understood.* Another pause. *There's something else. Dorian filed a new dungeon registry inquiry yesterday afternoon. Not the Greystoke — he's moved past that. Something else.*
*What dungeon.*
ROWAN: *That's the thing. It's not in my records. Not in the Association's public registry. Not in any historical data I can access.* A pause that went longer than Rowan's usual analytical rhythm. *It's a spatial formation inquiry for a location in the canyon district. West side, above the ridgeline. The inquiry references a classification code I haven't seen before — it's formatted like a standard dungeon registry code but the prefix is non-standard. The Association's system accepted it, which means it's a valid code in their internal taxonomy. But it's not one that's been used for any publicly registered dungeon in the city.*
Kael's hand stopped moving.
The canyon district. West side. Above the ridgeline — near the access road where Illen's research station sat.
*When was the inquiry filed.*
ROWAN: *Yesterday at 4:47 PM. Nine hours after you came out of the Greystoke Vault.* A pause. *Kael. He's not looking backward at the vault. He's looking at something new. Something I can't identify from public records. And it's close to Professor Illen's station.*
---
He went to Illen's on Wednesday, as scheduled.
The canyon trail was quiet — mid-morning, no other traffic on the access road. The red trail marker at the junction point, then the station's low profile against the ridgeline, the same view he'd been walking toward every Wednesday for two months.
Illen was waiting with tea and the next section of the fourth regressor's record.
"Month thirty-three," Illen said, setting the printed pages on the table. "This is the section I mentioned I'd been hesitant to share."
Kael picked up the pages. Three entries, handwritten transcriptions of the original journal. The fourth regressor's handwriting had gotten smaller by month thirty-three — cramped, as if the margins were closing in.
The first entry was administrative — check-in protocol, monitoring status, nothing useful.
The second entry stopped him.
*Month 33, Entry 47: The stolen advantages. I've been cataloguing the results of the interception campaign — the resources I've taken from the people who, in my timeline, used those resources to gain the power they used to destroy me. The grimoire. The quest reward. The sponsorship. The artifact. I intercepted four of the five targets I'd identified.*
*Results: mixed.*
*The targets did not become weaker. They became different.*
*D. lost the class evolution quest and found a different evolution path — one I hadn't mapped, one I couldn't predict, one that created a capability set I had no counter-strategy for because the capability set didn't exist in my timeline. The grimoire target lost access to the convergence techniques and developed a brute-force methodology that was less elegant but more dangerous at close range. The artifact target — without the divine artifact, the healing architecture didn't develop the reversal function. Instead, it developed a corrosion function. Different mechanism. Same result: things I'd planned to survive became lethal in ways I hadn't prepared for.*
*I'm calling this "stolen advantage syndrome." You don't weaken them by taking their tools. You redirect them. The ambition remains. The drive remains. The talent remains. The tools change. And the new tools are the ones you don't know about.*
*The interception campaign cost me seven months of effort and significant resource expenditure. The net result was replacing known threats with unknown ones.*
*Month 33 is when I realized this approach was wrong.*
Kael read the entry again. Then a third time.
He set the pages down.
"The fourth regressor stopped intercepting after month thirty-three," Illen said. "The journal shows a strategic pivot to direct confrontation. Less resource denial, more capability development."
"What happened."
"The direct confrontation approach worked better in some cases. In others, it accelerated the conflict timeline beyond what the fourth regressor could manage." Illen looked at his tea. "The fourth regressor lasted until year five. The interception campaign consumed months thirty through thirty-three. The direct confrontation approach consumed months thirty-three through year four. Neither approach was sufficient on its own."
Kael sat with that. Let it press down on him. A predecessor who had made the same list. Identified the same targets. Run the same campaign. Found that stealing the board pieces didn't end the game — it changed the rules.
"You're sharing this now because I'm about to start my own interception campaign," Kael said.
"You're already planning it. The vault was the first move."
"The vault was different. The catalyst was for me, not denial of Dorian."
"Was it." Illen looked at him steadily. "The catalyst accelerates your development. But you chose the Greystoke specifically because Dorian had filed for it. The acceleration was the objective. Denying Dorian was the method."
Kael didn't answer. Because Illen was right.
"The fourth regressor's warning is specific," Illen said. "Stolen advantage syndrome. The resources are real, but the threat isn't the resource — it's the person. Take the resource and you haven't addressed the person. You've just made them harder to predict."
"And if I don't take the resources?"
"Then they have what they had in your timeline. Known quantities."
"Known quantities that killed me."
Illen was quiet for a moment.
"Yes," he said. "That's the problem the fourth regressor couldn't solve either."
---
Castellan cleared him at two in the afternoon.
"The shoulder cluster is stable. The catalyst load has normalized — the architecture adapted faster than I projected." She ran the final assessment sequence. "You're cleared for active operations. Modified protocol: no engagement above eighty-two percent architecture load for the next two weeks. After that, standard thresholds resume."
"Eighty-two percent."
"The catalyst is adding a passive two-percent draw that wasn't there before. Your previous safe ceiling was eighty-four. I'm adjusting to maintain the same margin."
He nodded. Two percent. The catalyst's presence, measurable and ongoing.
"The Void integration timeline," he said.
"Five weeks. Maybe four." She closed the assessment notes. "When it crosses — I want to be there. The threshold event will generate a channel signature that I need to document in real time."
"You'll be there."
She looked at him. "The tight timeline you mentioned. Is this the next phase."
"Yes."
"Then the modified protocol isn't optional. Eighty-two percent ceiling for two weeks. Not eighty-three. Not 'just this once.'" She held the look. "The integration work holds if you let it. It breaks if you don't."
"Eighty-two percent. Two weeks."
"Good." She picked up her equipment. "Rest tonight. Start fresh tomorrow."
She left. He sat in the assessment room and felt the architecture running at its new baseline — the catalyst's draw woven into the channel density, the shoulder cluster stable, the re-coordinated rhythm from the outer district modifier settled underneath everything like a foundation.
He was ready. Four targets, four timelines, one warning from a dead regressor who'd tried the same approach and failed.
His phone buzzed.
ROWAN: *Update on the canyon district inquiry. I pulled the registry filing through a secondary access point. The classification code Dorian used — I found one match. It's not a dungeon code. It's a spatial anomaly monitoring code. The kind the Association uses for locations where dungeon formation is probable but hasn't occurred yet.*
*He's not looking for a dungeon that exists,* Kael sent.
ROWAN: *He's looking for a dungeon that's about to appear. In the canyon district. West ridgeline.*
Kael stood.
The canyon district. West ridgeline. Where Illen's research station had been tracking regression instances for six years. Where the fourth regressor's stone sat on a shelf, a relay point that the anchor marking had oriented toward for two months.
Dorian wasn't looking backward at the Greystoke Vault. He wasn't mourning a loss.
He was looking at something new. Something close to the one place Kael couldn't afford to have him looking.
And the classification code meant the Association's spatial monitoring division agreed with him — something was forming there.
Something that hadn't existed in the original timeline.
