Mourning the Erasure of Detail

Note: This is less of a formal article and more of a personal journal entry. I usually keep these reflections private, but I'm sharing this publicly in the hope that others experiencing the same displacement might feel less isolated.

I have been reflecting deeply on my place within the current trajectory of the software industry. Before the rise of "vibecoding" and the pivot toward valuing developers primarily as architects of high-level abstractions, I took great pride in identifying as a coder rather than an engineer.

The Philosophy of the Machine

To me, the distinction was vital. Being a "coder" meant a devotion to the granular: performance tuning, mastering the hidden quirks of a language, and possessing a transparent understanding of the execution flow. I preferred this over the "Java-style" juggling of endless layers of abstraction.

While I recognize this view may have been partially based on a misconception, I feel the ground shifting. The strengths I've cultivated are becoming irrelevant, and the industry is migrating toward a cognitive model that my mind simply cannot adopt.

My Path to the Metal

My obsession began with a visual spark: the history sequence in Tron: Legacy. As a ten-year-old, I was consumed by a need to understand those lines of code. This sparked a grueling journey into programming. I wasn't driven by utility or the desire to "build a product"; I simply wanted to understand the machine.

My evolution looked something like this:

1. The Linux Era: Years of immersion to grasp the OS fundamentals.
2. Specialization Attempts: I dabbled in:
   • Networking
   • Cryptography
   • Rust
3. The Return to Base: I always gravitated back to the hardware.

I found my bliss in the "small" things: writing toy operating systems, manually counting CPU cycles, and hand-crafting machine code. To me, the computer is not a tool, but an extension of my own being—much like a fisherman's relationship with their rod.

The Cognitive Burden

I have spent eight years in this pursuit, and I am still overwhelmed by the depth of it. My relationship with hardware is so intimate that it causes genuine anxiety when I cannot map the logic.

I recently encountered an annotated version of The Story of Mel and felt a surge of disbelief: Who actually needs an explanation for hexadecimal?

The Bottom-Up Requirement

I am not a generalist with a niche; I am a low-level specialist by necessity. I can build a website or research physics, but I cannot function using a "top-down" learning approach. I cannot operate with gaps in my knowledge.

My Learning Process:

This need for total comprehension extends to every facet of my life:

• Mathematics: I spend hours refining $\text{Axioms} \implies \text{Theorems}$ until they feel intuitive. (My girlfriend and I even spend quality time proving theorems together).
• Culinary Arts: I cannot experiment with a recipe without first understanding the underlying chemical reactions.
• Software: If a tool doesn't "click" at a fundamental level, it is unusable to me.

The Great Displacement

For a long time, the low-level world was my sanctuary. Even those who didn't share my obsession respected the necessity of the work. There was a shared understanding that:

• Small compiler optimizations yield massive compound gains.
• Writing assembly for JSON parsers is how you make code fast go brrr.
• Software shouldn't be "Electron slop."

I built my reputation on this expertise—the ability to glance at a bug and intuitively know the failure point.

The LLM Impact

The industry consensus is that Large Language Models (LLMs) render "worrying about details" obsolete. The goal is now to let the AI handle the minutiae while humans focus on "the big picture."

For me, this is catastrophic.

// The LLM can now do this automatically:
// 1. Identify a hot loop
// 2. Find a vectorization trick from the web
// 3. Implement it perfectly

If an AI can explain pointer provenance and babysit a developer through the process, the value of a human expert vanishes. I cannot scale my mind to "expanded scope"; I feel physically ill when a project's architecture exceeds what I can hold in my active memory.

The Aftermath

The last time I used an LLM for a personal project, I realized the AI knew more about the project than any human I could share it with. Being told that my specific way of thinking is no longer required—and being redirected to a chatbot—is more than insulting; it is an erasure of my identity.

The result of the last year: Planned Future $\rightarrow$ Considering applying for disability.

LLM-driven development has optimized away every element that made programming bearable for me. The market has embraced this as "the future," leaving me with no viable alternative. Outside of Big Tech (which welcomes the AI shift), few companies have the resources or the desire to care about the "details."

Even my hobbies are fading. Retrocomputing and performance optimization are being invaded by "LLM fans" who seek the status of being an expert without the experience of the struggle. The joy is being sucked out of the machine.