Chapter 22 : What Ethan Missed
Seventeen weeks of notes filled fourteen spreadsheet tabs.
I sat at my desk in the associate bullpen at 6:47 AM on Monday, Week 11, watching the cursor blink over the first entry. Day 1. Trial binder. Three factual connections the senior partner missed. Two corruption flags I accepted as clean.
The retrospective audit was the only response to what I'd learned in the recalibration. If the Ledger had been weaponizing my wishful thinking since Day 1, then every synthesis I'd run contained two errors I hadn't caught. Every deliverable I'd sent upstream carried facts I wanted to believe rather than facts that existed.
Seventeen weeks. Roughly forty Omniscience activations. Eighty corrupted facts, minimum.
"Start with the ones that mattered," I told myself.
The spreadsheet organized by case, by date, by whether the output had reached a partner's desk. Most of my early work had been background research — useful but disposable, the kind of associate labor that built the foundation without supporting the roof. If those contained errors, the errors had been buried under subsequent corrections.
The merger matter. The Tanner complaint. The Folcroft footnotes. The Rees regulatory strategy.
I opened the Dressler-Hawkins file first. Day 3. The liability thread connecting three documents. Gregory's "good catch" afterward. Two corrupted facts accepted.
[RETROSPECTIVE AUDIT: Synthesis 2 — Dressler-Hawkins. Original output: liability thread + two flagged points. External verification status: UNVERIFIED.]
I pulled the original documents from the archive system and ran the manual cross-reference I should have run eleven weeks ago. The liability thread held — the connection between Exhibit D, Exhibit F, and the financial summary footnote was real, documented, traceable. Gregory had missed it; I had caught it; the catch was legitimate.
The two flagged points were the seller's motivations and opposing counsel's strategy. I had interpreted the seller as motivated by timeline pressure; the documents supported timeline awareness but not the specific pressure I'd inferred. I had interpreted opposing counsel as aiming for quick settlement; the subsequent behavior suggested they'd been preparing for extended negotiation.
Neither error had cost the client. Gregory's work had covered the gaps my synthesis missed. The matter closed successfully eight days after my memo.
"Mark it," I told myself. "Non-fatal. Self-correcting ecosystem."
I moved to the next file.
The Tanner matter was the first where my errors had reached Harvey's attention.
I opened the complaint analysis from Day 11. The synthesis had mapped Tanner's strategy: discovery overload, settlement pressure, volume as cover for precision. The two flagged points had been the volume assumption and the target identification.
The volume assumption I'd caught and corrected in the rebuild after Tanner's discovery response revealed the mismatch. That correction had cost four hours and produced the "wrong predictions you catch" conversation with Harvey. Non-fatal. Professionally useful, even.
The target identification was different.
I scrolled through my notes to the deposition. Tanner had gone after billing records, not timeline fractures. I'd assumed the timeline was his primary target because the show's Tanner had used timeline arguments in season 1. I'd mapped my meta-knowledge onto a person who existed in reality, not on a screen, and reality had deviated from my expectations.
Harvey had improvised. Harvey had saved the deposition. Harvey had given me a harder case afterward, which I'd interpreted as trust.
But the underlying error was mine. The target identification was one of my two corrupted facts, and I'd accepted it because I wanted the show's playbook to be accurate. I wanted my foreknowledge to be useful.
The Ledger had noticed. The Ledger had flagged it. I hadn't listened.
[RETROSPECTIVE AUDIT: Tanner target identification — corrupted fact. Cost: Deposition surprise. Mitigation: Harvey improvisation.]
Eleven weeks of case files took four hours to audit.
The pattern became visible around Hour Two: the corrupted facts clustered around meta-knowledge applications. When I'd used the show's plot points to predict real-world behavior, the synthesis had given me what I expected — and the expectations were almost always slightly wrong.
Tanner as volume bully — wrong.
Hardman's timeline as matching season 2 exactly — uncertain.
Partner voting blocs as stable — unverified.
The Ledger wasn't just flagging wishful thinking. It was flagging the specific wishful thinking that came from believing I knew how this story would unfold. Every time I'd reached for meta-knowledge as a shortcut, the synthesis had handed me a corruption flag I'd ignored.
"The show is a map," I'd told myself on Day 1. "Use it."
The map was wrong. Not entirely — the terrain existed, the landmarks were real — but the distances were different, the paths were different, the people were different enough to matter.
I closed the Tanner file and opened the one I'd been avoiding.
The Claire Benning obligation chain. Day 14. The Social Debt draft that had given me early discovery access.
The Benning draft had been clean. The obligation chain was real, traceable, documented. Claire had needed pro bono help; I had provided it; she had owed me a professional favor; she had honored it. The debt had registered accurately.
But the draft had borrowed from another chain. Marcus Crane's counterparty relationship.
I found the entry in my notes: "Crane's exposure to Benning's supervising partner creates secondary leverage. If Crane's current deal involves the same counterparty structure, the borrowed debt extends further than the immediate transaction."
Two corrupted facts. I hadn't identified them at the time. I hadn't verified.
I pulled the Crane matter from the archive.
The counterparty relationship was narrower than my synthesis had suggested. Crane's deal involved a different subsidiary structure — related but not identical. The leverage I'd borrowed in the Social Debt draft was against an obligation chain that was thinner than I'd believed.
"Downstream consequence," I realized.
If Crane's deal changed — which it might, because deals at this firm changed constantly — the leverage I'd borrowed could collapse retroactively. The Social Debt draft had been built on a foundation I hadn't verified, and the foundation was weaker than my synthesis had claimed.
Crane didn't know I existed. He'd never heard my name, never seen my face, never registered my presence in the firm's ecosystem. But if his deal shifted, the obligation chain would reorganize, and someone I'd never intended to touch would feel the ripple.
I added a new entry to my private file:
"Crane chain problem. Narrow but intact. Monitor deal status through legitimate firm channels. Prepare contingency."
Louis Litt passed my desk at 11:34 AM.
I was deep in the retrospective files, screen showing three case documents I hadn't been assigned to. Louis slowed. His eyes moved across my screen the way they'd moved across my document organization on Day 4 — cataloging, filing, building his picture.
He kept walking.
Three minutes later, he passed again. This time he stopped.
"Calder."
I looked up. His expression was professionally neutral — the same neutral he'd worn during the associate review, the same neutral he wore when asking questions that weren't really questions.
"Research for an active matter?" he asked. "Or general preparation?"
"Active matter," I said. "Cross-reference check."
This was technically true. The Crane chain problem was active in the sense that it was unresolved. The cross-reference was legitimate in the sense that I was verifying whether my earlier synthesis had been accurate.
Louis held my answer for a beat. His expression didn't change, but something behind his eyes shifted — not suspicion exactly, more like confirmation of something he'd already suspected.
"The Crane files are in the third-floor archive," he said. "If you need them."
"I'll note that."
He walked away.
I watched him go and understood what had just happened. Louis had caught me doing cross-case research. He'd accepted my explanation because it was plausible. But the explanation was now part of his file on me — another data point in the pattern he was building, another notation in whatever documentation he was assembling.
The Exposure Debt warmed in my chest. Not critical. But climbing.
[EXPOSURE DEBT: Louis Litt — cross-case research witnessed. Explanation accepted. Skepticism: HIGH.]
The retrospective audit closed at 1:15 PM.
Six cases. Eleven weeks. Twelve identified corruptions, eight of which had self-corrected through subsequent work or partner intervention. Four that hadn't.
The Tanner target identification — mitigated by Harvey's improvisation.
The Crane counterparty relationship — active, unresolved.
Two others in minor matters — errors buried under client exits and case closures.
I saved the audit file to my encrypted folder and closed my laptop.
The list of things I'd gotten wrong because I wanted them to be true sat in my awareness like a photograph I hadn't known existed. This was not a catalog of failures. This was a map of everything I cared about, rendered in negative space by the corruption protocol's targeting.
The show's accuracy. The meta-knowledge edge. The belief that I knew how this story would end.
Those were my wishful thoughts. Those were what the Ledger had been weaponizing since Day 1.
I packed my bag and walked toward the elevator. The Crane chain was narrow but intact tonight. The specific quality of intact things under stress was that they broke when you stopped watching them, not while you were looking.
I needed to start watching.
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