The Anti-AI Trend: Hardcore Programmer Writes a C Directory Viewer from Scratch

Overview

In the current landscape where Generative AI (AIGC) deeply permeates the software development workflow, a "pro-AI" hardcore programming wave is quietly emerging within the developer community. Recently, tech blogger Hayyoth released a video showcasing a C language directory tree viewer created entirely by manual handwriting with no AI-generated code, quickly sparking profound reflections within the industry on "code sovereignty" and "bottom-level capabilities." The work reconstructs the original logic of POSIX system calls with an extremely minimal technology stack, becoming a landmark event currently hotly discussed in technical circles.

Hands-on Code Implementation

Hayyoth explicitly stated in the video that the entire process followed the "No AI" principle, meaning no code was generated with the help of ChatGPT, Copilot, or any large language models. The developer built the core functionality similar to the Linux tree command from scratch, relying solely on a reference book and personal memory.

The program's core logic depends on standard POSIX interfaces:

• Using opendir() to open a directory stream;
• Traversing directory entries via readdir();
• Utilizing the d_type field to determine if an item is a subdirectory (DT_DIR) and performing recursive processing.

The entire development process took approximately 8 and a half minutes. Although the code volume was extremely small, it fully implemented directory traversal and hierarchical printing functionality. This "open the book, read the knowledge, and code whatever you want" workflow deliberately avoided the convenience brought by modern AI programming assistants, instead emphasizing the programmer's direct control over underlying system mechanisms.

"No AI. Just open a book, read it, and code whatever you want." —— Hayyoth

Industry Trends and Deep Reflections

This move is not an isolated display of technical prowess but rather reflects a counter-trend in the current software development field. With the widespread adoption of LLM-assisted programming, some senior engineers are concerned that developers are gradually losing their understanding of memory management, system calls, and algorithmic complexity.

1. Return to foundational capabilities: Manually writing system-level code requires developers to have a clear understanding of stack frames, pointer operations, and operating system kernel interfaces, which serves as a correction to "black-boxed" programming methods.
2. Credibility and security: AI-generated code may contain hallucinations or security vulnerabilities, whereas manually written low-level tools are often more interpretable and controllable.
3. Educational significance: For beginners, bypassing AI dependency helps establish a solid foundation in computer science, avoiding becoming "glue programmers" who only know how to call APIs.

Although the video currently has few views, the values it conveys have garnered widespread resonance in the technical community. In today's AI-saturated world, retaining the ability to "hand-write core code" is seen as a key defense line for maintaining engineers' professional dignity and technological independence.