Three days before the budget committee meeting, Chen Haoran had the full dataset.
He had gotten into the Records Management Division's server in six hours. Government infrastructure from two budget cycles ago, running on software with a documented vulnerability that had not been patched because the IT department was understaffed and the procurement cycle for a new server contract was stuck in administrative review. He had left no trace. He had only read — he had extracted nothing through the server itself, had not touched the files in any way the system would register as access.
He had made notes.
The fabricated data points were in the combined report exactly where the system said they would be. Pages 4, 7, and 12. Efficiency metrics from cherry-picked nine-week windows in Q2 of two years ago — periods that covered the Bureau's transition between two processing systems, when throughput was temporarily depressed by training demands. Compared to the full quarterly dataset, the metrics looked normal. Compared to the cherry-picked windows in the opponents' report, they looked like a failing operation.
He built a response document. Full dataset comparison, properly sourced through publicly available quarterly reports that the Bureau had filed with the city administrative office — all public, all citable, no indication of how he had known which periods to look at. The document was twelve pages. The significant information was in the first three.
He wrote a cover note explaining what the document was and why Director Cao might find it useful for the upcoming budget discussion.
Then he sat with the problem of how to get it to her.
The Favor Token created one opportunity for direct professional contact. He had not used it. He needed to use it in a way that made the contact about the statistical anomaly — the thing that was actually interesting, the thread that would matter for the main quest — and not just about the budget report. The budget report was useful, but it was a door. He needed to walk through it, not stand at it.
The system generated a side task that morning:
```
[SIDE TASK: Director Cao has a scheduled working lunch at the Pavilion House restaurant at 12:30 PM tomorrow. She eats alone. The reservation is under a secondary name she uses for meals where she does not want to be approached by petitioners. Table 7. REWARD: [NO LP] — [FAVOR TOKEN: CAO MINZHI x1 ACTIVATE CONDITION MET]]
[NOTE: This is not a transaction. This is a circumstance. You choose what to do with it.]
```
He read this carefully.
No LP reward. The system was telling him the meeting opportunity was present, not instructing him to take it. *You choose what to do with it.*
He thought about what he would say. He had one meeting, probably fifteen minutes before she decided whether to continue or terminate it. Fifteen minutes to build a reason for her to give him more time.
The budget document was the entry. The statistical anomaly was the point.
He was case eight.
---
The Pavilion House restaurant had good natural light and a lunch menu that was expensive enough to keep it quiet on weekdays. Table 7 was near the window but not on the window wall — a deliberate choice by someone who wanted ambient light without being visible from the street.
Director Cao Minzhi was already seated when he arrived, twelve minutes after her reservation time. She had a tablet and a report in front of her and a cup of tea she hadn't touched. The restaurant's B-rank awakened maître d' would have identified her power level at the door, noted the secondary name, and ensured she wasn't disturbed.
He approached anyway.
She looked up. The expression: efficient, not hostile, already assessing. "I'm using this table for a working lunch," she said. "I'd appreciate—"
"I have the full dataset for the budget efficiency metrics," he said. "The ones in the combined report from Directors Chen and Hu. You have seven days."
She looked at him at him.
"Sit down," she said.
---
He spent eleven minutes at her table.
He gave her the document. She read the first three pages with the specific attention of someone who read fast and retained everything. She checked two of the citation references on her tablet. She looked at page 7's comparison table and her expression changed slightly — the shift of someone finding what they suspected was there.
"These periods are cherry-picked," she said.
"Yes. Q2, two years ago. The system transition window. The full quarterly data normalizes completely."
"How do you have this?"
"The quarterly reports are filed with the city administrative office," he said. "Public record. The comparison analysis is mine."
She looked at him across the table. She had the particular expression of a B-rank administrator with data sensitivity — an ability that read informational irregularities instinctively. She was reading him right now. He could feel it — a slight pressure, like air before a storm.
He was not irregular. Everything he had told her was true.
The pressure eased.
"You walked up to my private lunch table," she said, "with a document that solves a problem that I haven't told anyone is a problem."
"You haven't needed to," he said. "The combined report is filed with the committee secretariat. The filing is accessible to parties with an interest in the committee's work. I have a professional interest in Bureau resource allocation." He paused. "That's accurate."
"What professional interest?"
He looked at her steadily. "I'm a zero-assessment from fourteen months ago," he said. "I've since reached F-rank cultivation. I have a professional interest in how the Bureau handles non-standard cases."
Her expression did not show what he hoped to see. It showed something more useful: specific, intense attention.
"Non-standard development," she said.
"Yes."
"What method?"
"I can't answer that yet." He met her gaze. "What I can tell you is that I believe I'm your eighth case."
The silence at Table 7 lasted about four seconds.
"The eighth," she said.
"Seven previously documented cases of zero-assessment subjects developing ability through non-standard means," he said. "Four years of tracking. I don't know how you're identifying them, but you are." He paused. "I'd like to know what you've found."
She looked at him.
She picked up the report and looked at the front page again. "This is real work," she said. "Not generated."
"Yes."
"You have a computer science background."
"Yes."
"And F-rank cultivation reached from a zero-assessment in fourteen months." She set down the report. She was working through something, adding it up. "Most F-rank cultivators from zero-assessments take two to three years to stabilize. Fourteen months is fast."
"Yes."
"How?"
"I told you — I can't answer that yet." He held her gaze. "I'm asking for a second meeting. When you have more time than a working lunch. I'm offering the following: the full analysis behind this document, which goes deeper than what's on those twelve pages. And whatever you'd like to know about case eight."
She looked at him.
"You walked up to my private table," she said again. "You had information you shouldn't have had access to through the process you described. You are either very well connected or—" She stopped.
He waited.
"Or the system that tracks these cases flagged you in advance," she said.
It was not a specific accusation. It was a hypothesis, spoken aloud because she had data sensitivity and it was pointing her somewhere.
He did not confirm or deny. He placed a business card on the table — his actual contact information, nothing fabricated. "The committee meeting is in seven days," he said. "That analysis should reach the relevant committee members before then."
He stood. "I look forward to hearing from you, Director."
He left.
---
She called him the next day.
Not through the number on the card — she had found his registered contact through the Bureau's own records, which told him she had looked him up officially. That was better than a direct call. Official interest, recorded.
"Monday morning," she said. "Eight-thirty. My office."
"I'll be there."
The call lasted forty seconds.
He spent the weekend preparing. The analysis behind the budget document. The zero-assessment anomaly — what he knew, which was limited, and what questions he wanted to ask, which were very specific. His own case, presented accurately without revealing the Luck System. He had thought carefully about how to describe his development without lying and without exposing what he needed to protect.
The answer he arrived at: *I cannot explain the mechanism. I can describe the effects and timeline accurately. A good investigator will find that more interesting than an explanation I can't give.*
He prepared three hypotheses about the zero-assessment anomaly, each one more specific than the previous, building from the public case data and what the system's basic profile on Cao had told him. The third hypothesis was the one he found most interesting: that the zero-assessments were not a failure to awaken but a different category of awakening, one that required an external catalyst the standard awakening event had not provided. He had written it as a framework rather than a claim. A good investigator would improve the framework.
He had two days. They were enough.
He thought about this on the walk back.
He had not used the Favor Token. The system had said the token *creates one opportunity for direct professional contact* — and he had walked up to her table without it, and she had said sit down, and they had spent eleven minutes together. The token was still unused.
He checked:
```
[FAVOR TOKEN: CAO MINZHI x1 — STATUS: UNUSED. STILL ACTIVE.]
[NOTE: YOU DID NOT NEED IT. THE MEETING WAS REAL. THE TOKEN IS HELD FOR A SECOND CONTACT IF NEEDED.]
```
He read this on the subway platform. The token had been available as insurance, and the meeting had not needed insurance. He filed this as evidence of what preparation looked like when it worked: he had understood enough about Cao Minzhi and about the budget situation that when the circumstance arose, he had been worth sitting down.
That was the whole mechanism. Not luck. Not manipulation. Just having done the work and being ready when the door opened.
He still had one Favor Token.
He would know when to use it.
The system was quiet that weekend. No new side tasks. He checked the main quest log:
```
[MAIN QUEST 1: CAO MINZHI]
[STATUS: UNLOCKED — PRELIMINARY CONTACT ESTABLISHED]
[PROGRESS: 1/3 MEETINGS]
[NOTE: SHE IS INTERESTED. THE ANOMALY THREAD IS THE CORRECT ENTRY POINT. DO NOT RUSH THE REMAINING STEPS.]
[CURRENT LP: 250]
[RECOMMENDED: COMPLETE ADDITIONAL SIDE TASKS BEFORE NEXT MEETING. BUILD LP RESERVES.]
```
He noted *do not rush* and filed it. He was not planning to rush. He had learned from month one that the system's tasks rewarded patience and preparation, not speed.
Monday morning. Bureau building. Eighth floor. He checked in with the reception desk using his actual name, his actual registration, and was directed to the elevator.
He rode up.
The doors opened on a corridor that looked exactly like the floors below it — fluorescent light, that particular government-building smell, the slight hum of data servers somewhere behind the walls.
He found office 804.
He knocked.
"Come in."
Director Cao Minzhi looked different behind her own desk than she had at Table 7 — the table had been her, choosing where to work privately. The office was her job, the B-rank administrator who ran twelve thousand assessments a day and did not remember the faces. She was in a charcoal blazer. The desk had three monitors. She had been working since before he arrived.
"Sit," she said. She did not look up from one of the monitors. "I'm finishing something."
He sat in the chair across from her desk and looked at the office. Efficient. No personal items except one — a small framed statistical chart on the wall beside the bookshelf. Seven data points with red markers, plotted against a timeline. The anomaly chart.
She had it on her wall.
She finished whatever she was finishing and turned to face him.
"Case eight," she said.
"Yes."
"Start from the beginning."
