After Lin Wanqing closed the door, Han Luyi activated Sight again.
The data panel in the upper right corner of his vision lit up:
[Experience: 12/200]
He was still a long way from leveling up.
He switched off Sight, went downstairs for braised chicken and rice, and thought things over as he ate.
At nine that evening, he showered and lay down in bed.
Today was surreal. My head's a mess. Sleep first. Let the background processes run a round of GC.
He set his alarm for seven in the morning.
The next day.
Sunlight woke Han Luyi ten minutes before his alarm.
He opened his eyes and activated Sight.
[Energy: 100/100]
After washing up, he made a cup of instant coffee and sat down at his computer.
He opened a freelance developer platform.
He used to pick up the occasional side gig there for extra cash. Back then, though, he screened projects purely through skill and experience, which wasn't particularly efficient.
Things were different now.
Sight activated.
An extra data panel appeared beside every outsourcing request on the platform.
First listing: data API development for a social app.
Han Luyi's mouth twitched. A sixty-seven percent chance of nonpayment. He had taken jobs like that before: once the work passed acceptance, the client vanished, and the final payment was forever "on its way."
Second listing: an AI course recommendation algorithm for an education company.
Isn't this basically playing with the whole map revealed?
He picked a few jobs with high credibility scores and low risks of nonpayment, then got to work.
Sight's code-scanning function revealed its terrifying efficiency.
The first job was optimizing an e-commerce backend. He opened the code, scanned it for three seconds, and every performance bottleneck and potential bug appeared as a highlighted block. A code architecture that would take someone else a day to understand made complete sense to him at a glance.
Forty minutes. Done. After he submitted the fix report, the client took quite a while to reply: "That fast? How many people are on your team?"
The second job, an AI recommendation algorithm, was a little more complex, but this was his old specialty. He finished it in an hour.
The third job was a database migration script, the simplest kind. Half an hour.
Three jobs. One morning.
The platform displayed an account balance notification: +8,200 yuan.
Back at Dingsheng, his monthly take-home pay had been a little over fifty thousand yuan, or just over two thousand per working day.
In half a day, he had earned almost four days' pay.
Han Luyi leaned back in his chair and stared at the number.
Five years of grinding at Dingsheng had never felt like this.
His energy had fallen to 51. Code scanning wasn't cheap; the three jobs had consumed almost half of it. But the money was already in his account, and he could spend the remaining time with Sight switched off, recovering slowly.
He stretched, closed the freelance platform, and opened his browser.
The alert Sight had picked up yesterday kept lingering in his mind: Flash Tech, severe anomalies in core operational data, substantial experience available for fixing them.
He searched for Flash Tech, opened its official website, and activated Sight.
It took just one second.
A glaring burst of red exploded across his vision.
A red overlay covered the entire page.
An enormous warning appeared in the center.
[Critical anomaly!!!] [Discrepancy between the company's publicly reported DAU and system-estimated actual figures: >400%] [Probability of data falsification: 99.7%] [Associated risk: anomalous access to user privacy data. Active scanning required for details.]
Fourfold.
Its DAU figures were inflated fourfold. A billion yuan in funding rested on numbers like these.
He picked up his phone, searched for "Flash" in the app store, and downloaded and installed it.
First, he scanned Flash's public materials: its website, app store page, and public funding announcements.
Sight's data overlay marked every set of published figures with its "discrepancy from actual value."
Registered users: officially claimed to be 80 million.
Monthly active users: officially claimed to be 60 million.
Daily active users: officially claimed to be 35 million.
A red label hung beside every set of figures.
Han Luyi's eyes swept back and forth over the three lines twice.
Wait.
On the surface, 80 million registered users and 35 million daily active users meant a daily activity rate of 43%. Perfectly reasonable.
But Sight put the actual registered user count at just 23 million.
Thirty-five million daily active users, twenty-three million actual registered users. Daily active users exceeded the total number of real users by more than ten million.
Even if every real user logged in every day, they couldn't reach thirty-five million. Where had the extra people come from?
His energy was dropping fast. 50, 46, 42...
With that question in mind, Han Luyi picked up the pace. He grabbed his phone, opened the Flash app he had just downloaded, and casually scrolled through a few videos.
At the same time, Sight scanned the underlying data flows.
After fifteen seconds of scanning, Sight produced a string of red warnings.
Contacts, text message history, GPS routes. Flash was stealing everything from twenty-three million users.
Sight probed deeper. The outlines of the backend architecture peeled away like layers of an onion, revealing a deeply buried module.
A chill ran down Han Luyi's spine.
But Sight wasn't finished. Another red warning appeared, so glaring this time that he narrowed his eyes.
The final piece of the puzzle fell into place.
The private data Flash stole was raw material. The behavioral patterns of twenty-three million real users were fed into the simulation engine to mass-produce twenty-eight million convincing "active users." These virtual accounts watched videos, liked posts, and left comments. Their behavior copied real people, making the deception almost impossible to spot from the outside.
Meanwhile, that same data was flowing overseas through multiple layers of jump servers. Contacts, text messages, GPS routes—Han Luyi hardly dared imagine what gray-market operators could do with them. Identity impersonation, precision scams, personalized price gouging through big data: every one was a business dealing in real money.
Six-point-eight million real people plus twenty-eight million ghosts made up the impressive figure of thirty-five million daily active users.
A perfect chain of fraud: steal data → imitate real people → inflate DAU → scam investors. And the data itself was sold a second time.
The billion yuan in funding had been obtained through fraud, while they also pocketed dirty money selling private data to gray-market operators. They were cashing in on both ends.
The revelations hit too hard. Han Luyi switched off Sight and took a long while to recover.
He opened a new text file and typed out everything he had just seen from memory: data discrepancies, privacy theft, overseas transfers, the behavior simulation engine, and the chain of fraud, listing each item separately.
Once he had recorded it all, he stared at the file in silence for a while.
He couldn't explain where this information had come from.
He needed a clean chain of evidence, one that could withstand scrutiny at every step.
Han Luyi started gathering evidence.
First came APK reverse engineering. After decompiling Flash's installation package, he could see its database interfaces. User IDs used an auto-incrementing primary key, with the highest value stopping around twenty-three million. The official claim of eighty million registered users was more than three times that number.
Next, he set up a proxy to capture packets. He connected his phone to the packet-capture tool on his laptop, opened Flash, and browsed casually for a few minutes. Network requests appeared one after another, and he isolated several deeply concealed encrypted requests directed to IP addresses in Singapore.
Finally, he wrote a crawler. A script collected behavioral data for "active users" on Flash's public pages. After half an hour, the results were in: large numbers of accounts had highly similar intervals between likes, browsing paths, and active periods, displaying clear signs of machine generation.
He saved all three pieces of evidence. This was material he could actually put on the table.
Han Luyi opened the "About us" page on Flash's website again.
Co-founder and CTO: Zhou Mingzhe.
He had seen that face before. At Dingsheng's annual party last year, the man had sat to Wang Zhiyuan's right at his table. Chen Bowen had even gone over to toast him.
VPs inviting outsiders to annual parties was perfectly normal, so he hadn't thought anything of it then. But now that face had appeared in Flash's CTO listing, a small question mark formed in Han Luyi's mind.
Two paths lay before Han Luyi.
The first: pretend he had seen nothing and delete everything he had just gathered. No risk whatsoever.
The second: expose it all.
Which do I choose?
Flash had a billion yuan in funding behind it, backing from leading VCs, and a dedicated legal team. What did he have? Evidence whose source he couldn't explain.
But if I do nothing—
Twenty-three million people's private information was flowing into the abyss every day, turning into romance scammers who fussed over your well-being, "customer service refund agents" who could recite your ID number, and "Officer Chen," claiming to be with the Guangdong police.
At that thought, he picked up his phone, opened WeChat, and found a group chat named [Dorm 404].
It was already dark outside. The glow of his phone lit his face.
A newly unemployed programmer taking on a company valued at ten billion yuan.
It sounded like a joke.
But...
Han Luyi typed a line: "Come out for dinner tonight. My treat. There's something serious we need to talk about."
Before you continue