Chapter 11: Cross-Document Recognition
The tenth activation unlocked something new.
I sat in Document Review Room 3 with three Tanner case files spread across the table — the original complaint, the discovery response, and a new motion Tanner had filed yesterday. The Ledger pressed against my sternum as I touched all three documents simultaneously, and the synthesis that followed was different from anything I'd experienced before.
Not just pattern recognition. Cross-document pattern recognition.
[CASE FILE OMNISCIENCE: Cross-document synthesis unlocked at activation 10. Multiple document integration: ACTIVE. Pattern correlation across files: ENABLED.]
The three files weren't separate documents anymore. They were a single picture, assembled from fragments that had been scattered across weeks of litigation. Tanner's strategy emerged in three dimensions: his complaint establishing the framework, his discovery response narrowing the target, his motion revealing the endgame.
And there, in the overlap between the discovery response and the motion, a timeline fracture.
The fracture was subtle.
A deposition date in Tanner's original complaint: March 14th. A document production date in his discovery response: March 21st. A reference in yesterday's motion to evidence that "emerged from the deposition" — evidence that, according to Tanner's own timeline, was produced after the deposition occurred.
The sequence only made sense if Tanner had known the document's content before it was officially produced. Someone had told him.
[TIMELINE FRACTURE DETECTED: Tanner filing sequence inconsistent. Evidence reference predates production. Confidence: HIGH.]
[WARNING: Two facts in synthesis may reflect user interpretation bias. External verification recommended.]
I paused. The corruption flags again.
This time, I checked. I pulled the original documents, traced the dates against the filing records, verified the timeline against the court docket. One flag cleared — the deposition date was accurate, confirmed by three sources. The second flag was more complicated.
The document production date required inference. The official production log said March 21st, but the actual production had occurred in batches over several days. The specific document Tanner referenced could have been in an earlier batch, which would explain the timeline without requiring a leak.
I traced the batch records. The document was in the March 21st batch — confirmed. The second flag should have cleared.
But there was a third-order inference I'd made without realizing it: I'd assumed the March 21st batch was the first time anyone at Pearson Hardman had seen the document. That assumption was built into my synthesis, shaping how I interpreted the fracture.
I checked the batch records again. The assumption looked solid. Multiple sources confirmed the production timeline.
"Close enough," I decided. "The fracture is real."
I accepted the inference and moved on.
Harvey's office. 4:30 PM.
"Timeline fracture," I said, setting the analysis on his desk. "Tanner referenced evidence in his motion that, according to his own filing dates, he shouldn't have known about until after the deposition."
Harvey picked up the analysis. His eyes moved across the first page with the speed of someone who was looking for a specific kind of information.
"The March 14th deposition."
"The document he references was produced March 21st. He cites it as supporting evidence for a position he established in the deposition — seven days before he should have seen it."
Harvey set down the analysis. His expression was neutral, but something shifted behind his eyes — the kind of calculation that happened faster than conscious thought.
"Who was in that deposition room?"
The question landed with specific weight. Harvey wasn't asking for information he didn't already have — he was directing my research, pointing me toward something he'd seen that I hadn't.
This was different from receiving my analysis. This was using me as a tool.
"I have the attendee list," I said. "Seven names. Counsel, parties, court reporter, two observers from the client's legal department."
"Get me backgrounds on all seven. Focus on the observers."
"Timeline?"
"Yesterday."
I nodded and left.
The human moment came at 9:00 PM.
I sat at my desk with a private system log open — a text file I'd been maintaining since day one, tracking Ledger activations without explanation. I added a single line: "Activation 10. Cross-document unlock. Pattern correlation enabled."
The entry was small. The significance was not.
Ten activations. Five weeks. The Ledger had evolved from document synthesis to cross-document integration, and the evolution had happened without fanfare, without notification, without any indication that a threshold had been crossed until I crossed it.
The quietness of the act was disproportionate to what it meant. But that was how the system worked — no celebration, no acknowledgment, just capability expanding in increments that I could only track after the fact.
I closed the log and went back to the attendee backgrounds. Harvey wanted them yesterday, which meant he wanted them now.
The seven names were routine except for one.
Six of the deposition attendees had clean backgrounds — counsel on both sides, a court reporter with fifteen years of experience, two client representatives with appropriate clearances. Nothing suspicious. Nothing that explained how Tanner had known about a document before production.
The seventh name was an observer: Daniel Welles, senior paralegal at the client's legal department.
Welles's background showed three years at the client company, standard paralegal credentials, no obvious connection to Tanner or his firm. But his employment history had a gap — eighteen months between his previous position and his current role, listed as "independent consulting."
I dug deeper. The consulting period overlapped with a case Tanner had litigated three years ago. Welles had worked as a document specialist for the opposing party in that case — the party that had lost, badly, after Tanner exploited a timeline fracture almost identical to the one I'd just identified.
The pattern was visible if you knew where to look.
I compiled the background report and sent it to Harvey at 10:47 PM. Three minutes later, my phone buzzed.
"My office. 7 AM."
Day thirty-four. Harvey's office. 7:00 AM.
Harvey was already at his desk when I arrived, the background report open in front of him. His expression was the same neutral he always wore, but his posture carried a specific tension that I'd learned to read as preparation.
"Welles," he said without preamble. "The consulting gap."
"Eighteen months. He worked for Tanner's opposing party in a 2009 matter — the Kessler case. Tanner won that one using a similar timeline discrepancy."
"You think Welles is Tanner's source."
"I think Welles has a pattern. Independent consulting that overlaps with Tanner's cases. Documents that appear in filings before they should."
Harvey set down the report. His eyes moved across my face with the assessment I'd seen before — the evaluation of a tool's usefulness.
"The deposition attendee list," he said. "Seven names."
"Seven names."
"I recognized one of them."
He didn't elaborate. He didn't need to. The recognition had been visible on his face when he'd asked the original question — a half-second shift in expression that meant something I didn't fully understand.
"Welles isn't just Tanner's source," Harvey said finally. "He's a plant. Tanner has been running this play for years — placing document specialists in opposing legal departments, waiting for the right case to activate them."
The revelation landed with specific weight. Not a leak. A long-term operation. Tanner hadn't stumbled onto information he shouldn't have — he'd built a network designed to deliver it.
"What do you want me to do?"
Harvey smiled. It was a small expression, barely visible, but it changed his face completely.
"Now we take him apart."
Get Early Access to New Chapters
Thank you for reading. For those who want to skip the wait, my Patreon is currently 21 chapters ahead of the public sites.
Schedule: 7 new chapters released every 10 days.
Benefit: Gain a significant lead of 7 to 21 chapters depending on your tier.
Support the project and start reading the next arc now: Patreon.com/IsekaiStories