Chapter 161: On Paper and on the Ground

The time in the lower-right corner of the screen flipped to 3:47 a.m. Lin Chen stopped typing, the knuckles of his hands stiff and pale. In the document, the section titled “Multi-Source Heterogeneous Data Cleaning Algorithm” had already been revised for the seventh time. He broke down the exception-handling logic for ConnectionResetError and Timeout from the factory’s stress-test logs into a state-machine flowchart, then redrew it in Visio. The lines had to be perfectly horizontal and vertical; the arrows could not cross. Academic grant applications did not recognize improvised shortcuts—they only recognized processes that could be reproduced. He rubbed his dry, burning eyes, and under the desk his left foot curled unconsciously. A numbness crept upward from his ankle. He got up to pour himself water, but only a shallow layer of lukewarm water remained in the thermos. He mixed in a little cold water and drank it in one gulp. A faint spasm tightened in his stomach.

He spread open his draft notebook and, following the table of contents of Computer Networks, replaced the wording line by line. The dirty-data filtering inside the try-except block became “rule-engine-based edge-side data preprocessing.” Queue buffering and memory monitoring became “dynamic resource scheduling strategies under high-concurrency environments.” The itemized billing logic on Old Zhao’s side was transformed into “a lightweight operations and maintenance model based on valid data throughput.” Every term had to land inside the keyword library of the Science and Technology Department’s application guidelines. This was not packaging. It was translation. Translating experience dug out of the dirt into language that could be set on a review committee’s table.

At seven in the morning, a long line had already formed at the dormitory washroom. Lin Chen stood at the end of it, holding an enamel mug. He had changed his left foot into an old pair of wide-toed cloth shoes, the laces tied very loosely, but with every step the swollen top of his foot still rubbed against the upper. He lowered his head and looked at his wavering reflection in the water tank. The dark circles under his eyes were heavy, but his gaze was steady.

Back at his desk, he arranged the materials he had sorted the night before into a kraft paper file envelope in order: project log summary, data-flow architecture diagram, protocol mapping table, detailed cost accounting sheet. Last was a handwritten outline of the application, copied over in neat black pen with no cross-outs. He tucked the envelope under his arm, locked the door, and went downstairs.

Autumn had already touched the campus in September. The edges of the parasol tree leaves had begun to yellow, and when the wind blew they came down with a rustling sound onto the asphalt road. Lin Chen walked very slowly, deliberately avoiding the crowded main road and taking the side path along the outside of the lab building. He did not dare put weight on his left foot, so his center of gravity rested entirely on his right leg. His gait was somewhat limping, but he moved steadily. Department Building, Room 304, was on the fourth floor. There was no elevator. Holding the stair rail, he made his way up one step at a time. Every ten steps, he stopped to catch his breath twice and steady his breathing. Sweat slid down from his temples, but he did not wipe it away.

At exactly two o’clock, he knocked on the door of Room 304. Professor Zhou was reading an official document with a red-lettered header. Hearing the knock, he looked up and pushed his glasses higher. “Come in.”

Lin Chen set the file envelope on the desk and pulled out a chair to sit down. Professor Zhou did not bother with pleasantries; he pulled out the materials directly. He flipped through them slowly, his gaze lingering longest on the Visio diagram and the cost accounting sheet. The office held only the sound of paper sliding against paper and the whir of the old ceiling fan.

“The logic works,” Professor Zhou said at last, closing the materials and tapping the desk twice with his fingers. “But an application is not a technical manual. What the Science and Technology Department looks at is industrial impact, technology maturity, and budget rationality. Here—” he pointed at the cost sheet, “you’ve listed eight thousand for hardware procurement, but you haven’t clearly stated the sensor models or the communication protocol standards. The reviewers will ask: why not use an off-the-shelf industrial gateway? What exactly is the advantage of your script?”

Lin Chen straightened in his seat. “Off-the-shelf gateways don’t support the legacy protocols of non-standard PLCs. The factory’s equipment is from before 2005. It runs on Modbus RTU and custom serial commands. My script performs a layer of protocol conversion, cleans the dirty data at the edge, and only then uploads it to the cloud. That cuts out the middleman’s data-cleaning service fee and reduces maintenance costs by about thirty thousand yuan per production line each year.”

Professor Zhou looked at him, a note of scrutiny entering his eyes. “And the data? What are you using to prove that?”

“The factory’s stress-test logs. Three thousand concurrent connections, disconnect rate of three per thousand. Memory usage stays stable under 420 megabytes.” Lin Chen took out a USB drive from his bag and handed it over. “The raw logs and the post-cleaning comparison samples are all in there. You can run them yourself.”

Professor Zhou took the drive and plugged it into the computer. The screen lit up, the terminal window filling with densely scrolling logs. He watched for several minutes without speaking. The office was left with nothing but the tapping of the keyboard.

“The formatting is too messy,” Professor Zhou finally said. “The application attachments require PDFs and structured tables. What you have here are terminal screenshots and CSV files. The review committee does not have time to look through raw code. You need to turn the stress-test results into charts: a concurrency-versus-response-time curve, a memory leak comparison, an exception-capture success rate. And—” he paused, “the application also needs a clear division of labor within the team. You can’t handle data collection, cleaning, uploading, and later maintenance alone. At minimum, you need two undergraduates assigned to data labeling and on-site debugging.”

Lin Chen nodded. “Understood. I’ll redo the charts tonight. I’ll draft the team roles and find two juniors from the computer science department.”

“Cut the budget by another fifteen percent.” Professor Zhou pushed the materials back toward him. “The department’s funding pool is only so large. If the hardware cost is too high, it will be rejected outright. Go ask the supplier for a channel price, or replace part of the modules with open-source solutions. Send me the revised version by four o’clock Friday afternoon. Late submissions will not be accepted.”

“Okay.” Lin Chen gathered up the materials and stood. The moment his left foot touched the floor, a sharp stab of pain made him sway slightly. He steadied his balance without making a sound.

“Lin Chen,” Professor Zhou called after him.

He turned back.

“When it comes to implementing technology, the first step is making people understand it, and the second step is making them willing to pay for it.” Professor Zhou looked at him. “Your code is solid, but the application is written for people who don’t understand code. Take what comes out of the dirt and pave it into a road people can walk on. Go on.”

By the time Lin Chen walked out of the department building, the sun was already slanting westward. His shadow stretched long behind him. He made his slow way back along the side path, still limping, but with his back held straight. Professor Zhou’s words ran through his mind again. It wasn’t criticism. It was calibration. He needed to translate technical language into policy language, package individual ability into team structure, and compress a wildcat approach into a standardized process. That was harder than writing code, but it had to be done.

When he returned to the dormitory, his roommates were still out. He spread the file envelope open and turned the computer back on. New folder: Application_v2_Charts_and_Budget. He opened Excel and began entering the stress-test data, generating line charts. Axes, legends, data labels—he adjusted them one by one. The glow of the screen reflected cold white on his face. He poured hot water over a bowl of instant noodles, but did not eat them, only watched the steam rise.

His phone vibrated. A WeChat message from Old Zhao: The gateway in Workshop No. 3 is down. The logs showConnectionRefused. The reconnect mechanism in your script didn’t trigger. Bring your laptop over tomorrow at nine in the morning.

Lin Chen stared at the screen.

Gateway failure. Reconnect mechanism failed. Which meant that before nine the next morning, he had to revise the application’s charts and budget while also troubleshooting network jitter inside the factory’s isolated internal network environment. Time had been split cleanly in two.

He replied: Received. I’ll send over a troubleshooting plan before ten tonight.

He closed WeChat and reopened the code editor. The cursor blinked on the screen. He typed: def reconnect_handler():. Under the desk, his left foot tapped lightly against the floor. The pain had already gone numb, fading into background noise. He knew that the application and the on-site debugging had to proceed in parallel. He could not stop, and he could not let either side fall into disorder.

Outside the window, the sky went completely dark. The streetlights came on, their halos passing through the glass and cutting a bright rectangle across the desk. He drew a deep breath and set his fingers on the keyboard. The tapping began again—dense, steady, without pause.