Chapter 104: Six Weeks
Week 1.
Sarah Chen looked at the mathematical framework that Soojin spread across six whiteboards in the Nexus R&D lab and said, “This is the most elegant threat I’ve ever seen.”
It was a Monday morning. The R&D lab — the sixth floor of the Nexus building, where Sarah’s team built the AI models that powered the platform — had been cleared of non-essential personnel. The official explanation was “infrastructure upgrade.” The real explanation was that two of the most brilliant analytical minds in Korea were about to build a mathematical shield against a nation-state intelligence service, and they needed space to think.
“The entropy analysis layer is extraordinary,” Sarah continued, tracing the equations on the second whiteboard. “You’ve essentially formalized the intuition that something is ‘too good to be true’ into a mathematical operator. The operator measures the gap between what a person should know and what they appear to know, and the gap is the signal.”
“And our job is to close the gap,” Soojin said.
“Not close it. Explain it. If we can show that a legitimate analytical tool produces predictions that approach the quality of Daniel’s actual decisions, the gap becomes ‘he’s using a really good tool’ instead of ‘he knows the future.'”
They worked for fourteen hours. Sarah brought the AI expertise — the understanding of how machine learning models could be designed to produce predictions that were genuinely good. Soojin brought the mathematical architecture — the understanding of exactly which patterns the detection framework looked for and how to design outputs that evaded detection.
The result was AMI 2.0: a genuine analytical platform that combined Nexus’s existing cultural signal analysis with Soojin’s temporal correlation methodology. The platform was real — it actually worked, producing market predictions with approximately 82% accuracy, which was significantly better than any existing commercial tool. It was also specifically designed to produce outputs that, when mapped against Daniel’s actual decision history, reduced the entropy gap from 40% to 8%.
“Eight percent is noise,” Soojin said, running the final calibration at 11 PM. “The detection framework cannot distinguish 8% from normal analytical variance. You’re invisible.”
“From your framework,” Sarah said. “What about a modified version? If the MSS adapts the methodology—”
“They can’t adapt what they don’t understand. The calibration parameters are the key, and I’ve designed AMI 2.0’s outputs to exploit specific limitations in those parameters. To counter AMI 2.0, they’d need to rebuild the entire calibration model from scratch — which would take an additional six months.”
“Buying us a year total.”
“Buying us time. Time is the only currency that matters in intelligence work.”
Week 2.
Wang Lei called from Shenzhen at 3 AM Seoul time, which meant it was 2 AM in Shenzhen, which meant Wang Lei was operating on the schedule of a man who no longer distinguished between day and night because the operational clock overrode the biological one.
“I’ve identified the Seventh Bureau team,” he said. “Four analysts, one supervisor. The supervisor is someone I trained — Colonel Zhao Mengxi. We worked together for six years in the Seventh Bureau before I… left.”
“You trained the person who’s hunting us.”
“I trained the person whose team has been assigned to evaluate a mathematical framework that may or may not be applied to identify anomalous decision-makers in the Asian technology sector.” Wang Lei’s precision was deliberate — the intelligence officer’s habit of framing threats in their most clinical form. “Zhao is methodical. He will follow protocol. The evaluation will proceed in three phases: validation, application, and reporting.”
“Where are they in the process?”
“Validation is complete. They’ve confirmed the mathematical framework is sound. Application begins this week — they’ll start building decision history datasets for test subjects.”
“Test subjects meaning us.”
“Test subjects meaning a list of high-profile decision-makers whose public records are accessible. You’ll be on the list. I’ll be on the list. Jimin will not — her decisions are classified, and the MSS would need to activate a Korean asset to access them.”
“Which brings us to Jimin’s sweep.”
“Which brings us to Jimin’s sweep.”
Week 3.
Jimin sat in her office at the Ministry of Foreign Affairs — the small, institutional space with the frosted glass door and the filing cabinets that contained thirty years of diplomatic precedent (the real kind, from her first life, committed to memory and deployed through reports that attributed the analysis to “established methodological frameworks” rather than “I was there”).
Wang Lei’s behavioral assessment framework sat on her desk, printed on paper that she would shred after memorization, because Jimin did not leave sensitive materials in accessible locations. The framework was elegant in its simplicity: a checklist of behavioral indicators that distinguished normal government employees from those who had been recruited by foreign intelligence services.
The indicators were subtle — not the dramatic tells of spy movies (dead drops, coded messages, furtive glances) but the quiet, human tells of a person leading a double life: changes in financial behavior, shifts in social patterns, the specific stress markers that appeared when someone was carrying a secret they hadn’t chosen and couldn’t put down.
Jimin recognized the markers. She recognized them because she carried her own secret — a different kind, a different weight, but the same fundamental architecture of concealment. The same careful management of what you said and what you didn’t. The same fatigue.
She spent a week observing. Not surveillance — observation. The diplomat’s tool: watching people in meetings, in hallways, in the cafeteria. Noting who was relaxed and who was performing relaxation. Noting who checked their phone at specific times and who avoided making eye contact with specific colleagues.
On Thursday, she found him.
Kim Taewoo. Deputy Director of the Northeast Asia Division. Fifty-two years old. Twenty-three years in the foreign service. A career diplomat with an impeccable record, a wife who taught at Ewha Womans University, and two children in university. The kind of man who blended into the Ministry’s corridors like wallpaper — competent, reliable, invisible.
The indicators were there. Subtle but present. A recent change in his commute pattern — he now took a different route home, one that passed through an area with limited CCTV coverage. A new phone — not a Ministry-issued phone but a personal one, used in specific locations at specific times. And the most telling indicator of all: a shift in his financial behavior. Small deposits, irregular timing, to a secondary account that Jimin identified through the Ministry’s financial disclosure records (which all senior officials were required to file and which no one ever actually read).
The deposits were modest — 500,000 won per month, approximately. Not enough to suggest wealth. Enough to suggest supplementary income from a source that wasn’t declared.
Jimin didn’t confront him. Confrontation was the amateur’s tool — it produced drama but not intelligence. Instead, she adjusted. She began routing her most sensitive assessments through a different authorization chain, one that bypassed Kim Taewoo’s division entirely. She created a parallel reporting structure that was invisible from the outside but that ensured no assessment she produced would pass through compromised hands.
And she sent Wang Lei a message on the secure channel: Found one. Deputy Director level. Northeast Asia Division. Active but low-level — likely a source of convenience, not a committed agent. Contained through routing changes. No confrontation.
Wang Lei’s response: Good. Containment is better than confrontation. A source you know about is a source you can manage. A source you’ve exposed is a source they’ll replace with someone you don’t know.
Week 4.
Sarah’s AMI 2.0 paper was submitted to the Journal of Strategic Management. Co-authored by Sarah Chen (Nexus Technologies), Han Soojin (KAIST), and three academic collaborators from Seoul National University who had been brought in to provide peer review and institutional credibility.
The paper was titled: “AMI 2.0: Advanced Temporal-Correlation Analysis for Strategic Decision Optimization in Cross-Border Technology Markets.”
It was, by academic standards, groundbreaking. The methodology was novel, the results were robust, and the implications for strategic management were significant enough that the journal fast-tracked the review process — a decision that Soojin had anticipated and Sarah had facilitated through a call to the editor-in-chief, who happened to be a former colleague of Sarah’s Stanford advisor.
“Is this ethical?” Daniel asked, reading the preprint in his office.
“The paper is real,” Sarah said. “The methodology works. The results are genuine. The fact that we published it on an accelerated timeline for strategic reasons doesn’t invalidate the science.”
“It feels like we’re using academic publishing as a weapon.”
“We’re using academic publishing as a shield. There’s a difference. Weapons attack. Shields protect.” She paused. “And Daniel? The paper is going to change the field. AMI 2.0 is a genuine contribution to strategic analysis. Five years from now, business schools will teach it. The fact that its creation was motivated by self-preservation doesn’t diminish its value.”
“You’re defending the integrity of a paper that was designed to hide the fact that I’m a time traveler.”
“I’m defending the integrity of good science that happens to also solve a personal problem. The two aren’t mutually exclusive.” She adjusted her Hello World hoodie. “Good science often starts with personal motivation. Penicillin was discovered because Fleming was messy. GPS was developed because the military needed targeting accuracy. AMI 2.0 was developed because three people needed to not be detected by a Chinese intelligence service.” She shrugged. “The origin story is unusual. The science is sound.”
Week 5.
Wang Lei’s documentation strategy was complete. Zhonghua Digital’s R&D archive now contained a comprehensive record of the analytical frameworks, market studies, and technology assessments that retroactively explained every anomalous decision Wang Lei had made since 2002. The documents were genuine — the analyses were real, the data was accurate, the conclusions were sound. They just happened to have been created in January 2020 and backdated through the archive system to appear as though they’d been produced at the time of each decision.
“Isn’t this falsification?” Daniel asked on the secure line.
“The analyses describe the reasoning process that actually informed each decision,” Wang Lei said. “The reasoning happened in my mind. The documents merely externalize it. Whether the externalization occurred at the time of the decision or seventeen years later is a question of archival timing, not intellectual honesty.”
“That’s the most sophisticated justification for document fraud I’ve ever heard.”
“Thank you. I spent twelve years developing the skill.” Wang Lei’s voice carried the faintest trace of dry humor — the intelligence officer’s wit, deployed sparingly and always in service of a point. “The alternative is allowing the MSS to identify me as a temporal anomaly, which would result in my involuntary return to Beijing for ‘consultation’ that would never end. I prefer archival creativity to institutional captivity.”
Week 6.
Soojin ran the final test on a Friday evening in February 2020. The Nexus R&D lab was empty — Sarah had sent the team home, because the test required processing power that would be noticeable on the network and explanations that neither of them wanted to provide.
The test was simple: apply the original temporal pattern analysis framework — the same framework the MSS possessed — to the updated decision histories of all three regressors. If the countermeasures worked, the scores would drop below the 0.7 detection threshold. If they didn’t, six weeks of work had been insufficient.
Daniel was there. He sat in a chair in the corner, watching the progress bar move across Soojin’s screen with the specific anxiety of a man watching a medical scan — the terrible patience of waiting for a number that would determine everything.
The results loaded at 8:47 PM.
Cho Daniel: *0.61*. Below threshold. Undetectable.
Wang Lei: *0.58*. Below threshold. Undetectable.
Seo Jimin: *0.64*. Below threshold. Undetectable.
Soojin exhaled. The exhalation lasted five seconds — the release of six weeks of tension, compressed into a single breath.
“You’re invisible,” she said.
“All three of us?”
“All three. The countermeasures work. AMI 2.0 explains your decisions. Wang Lei’s documentation explains his. Jimin’s routing changes prevent her data from being accessed in the first place.” She looked at Daniel. “When the MSS applies my framework to your public records, they’ll find scores within normal range. The conclusion will be ‘talented decision-makers using sophisticated analytical tools.’ Not ‘time travelers.'”
Daniel closed his eyes. The relief was physical — a loosening of muscles he hadn’t known were clenched, a settling of weight he hadn’t known he was holding. For six weeks, the possibility of exposure — of being identified, extracted, studied, weaponized — had hung over them like a blade. And now the blade was gone. Not destroyed — blades don’t disappear. But deflected. Turned aside. Aimed at something that wasn’t them.
“Thank you, Soojin,” he said.
“Thank the mathematics. I just followed where it led.”
“The mathematics didn’t build AMI 2.0 in two weeks on four hours of sleep a night. You did that.”
“I had Sarah. Sarah is… Sarah is what every mathematician wishes they had — a mind that translates theoretical frameworks into operational systems with a fluency that borders on supernatural.” She paused. “Is she also—”
“No. Sarah is just brilliant. The old-fashioned kind.”
“The old-fashioned kind.” Soojin almost smiled. “In a room full of time travelers, the most impressive person is the one who got there through normal means.”
Daniel drove home through the February night. Seoul was cold, dark, the winter refusing to release its grip even as February edged toward March. But the drive felt different from the drives of recent weeks — lighter, faster, the car moving through streets that were not empty but alive, carrying the specific energy of a city that was always building, always growing, always becoming.
He texted the group: Scores: 0.61, 0.58, 0.64. All below threshold. The shield holds.
Wang Lei: Confirmed. Phase 3 complete. Recommend maintenance monitoring — quarterly re-scans to ensure the countermeasures remain effective.
Jimin: Agreed. And thank Professor Han. She built the weapon and the shield. That’s a rare duality.
Soojin (she’d been added to the group that evening, by unanimous agreement): The weapon existed in the mathematics. The shield existed in the people. I provided the equations. You provided the trust. Both were necessary. Neither was sufficient alone.
Daniel: My mother would say this calls for galbi.
Soojin: Your mother has been sending me galbi since last week. She found out about me through what I can only describe as an intelligence network that makes the MSS look amateurish.
Wang Lei: Kim Soonyoung’s galbi distribution network is the most effective intelligence operation in East Asia. I’ve been saying this for years.
Jimin: She sent me kimchi yesterday. With a note that said “eat this and stop looking tired.” How does she know I look tired? We’ve never met.
Daniel: She knows everything. She always has.
He put the phone down. Pulled into the driveway. The jade tree was dark against the darker sky — bare branches in February, waiting for spring, growing in the slow way that trees grow, one ring at a time, invisible and unstoppable.
The shield held. The secret was safe. The future was unknown.
And in a house in Songdo, a man who had once known everything was learning to live with knowing nothing — and discovering, to his surprise, that the nothing was enough.
More than enough.
It was everything.