We Are the Last People Who Know How It Works

---

Filed: 2026-06-26 | Word Count: $\approx 770$ | Read Time: 4 Min Author: Cyrus Lopez

"You only ever know the things you can lose to."

The Era of Friction

Back in the 1990s, interacting with a computer wasn't a seamless experience; it was a negotiation. To simply launch a game, you first had to achieve a fundamental understanding of the hardware's inner workings.

It often started with a boot disk—a floppy drive's sole purpose was to force the machine into a specific configuration required by a single piece of software. You didn't learn this because you wanted to be an engineer; you learned it because you wanted to play, and the machine refused to cooperate until you paid the toll of knowledge.

The Sensory Experience of Computing

The process was visceral and audible:

• The Modem: It didn't just connect; it sang. That piercing shriek was actually two machines arguing over a shared language. After a while, you could hear the exact moment the argument failed and the connection was about to drop.
• The Sound Card: You had to manually assign the correct interrupt. If you guessed the IRQ wrong, you were met with total silence.

The machine was defined by its smoothness sharp edges. You discovered where those edges were by cutting yourself on them. You came to understand the computer the same way you understand anything that resists you: through struggle.

The Modern Paradox

This historical context makes our current trajectory—the rise of AI-driven services—deeply strange. We are told this is the "ultimate convenience." We are building the Last Convenience: the AI assistant designed to sand down every remaining bit of friction.

The AI Experience:

1. It reshapes itself to fit your phrasing.
2. It apologizes when you are dissatisfied.
3. It iterates until you are happy.

But there is a philosophical cost here. A machine that cannot challenge you is a machine you cannot truly know.

Comparing the Epochs

Competence vs. Information

There is a common narrative among the "graybeards" that we are witnessing a collapse of competence. They worry that because no one makes their own kernel anymore, a critical failure deep in the stack will occur and there will be no one left with the skills to descend into the depths and repair it.

# The old way: Manual kernel compilation
make menuconfig
make -j$(nproc)
sudo make modules_install install

However, the information itself isn't disappearing. In fact, it's more accessible than ever. The AI will recite the technical specifications of every legacy system flawlessly and eternally.

If this were merely a question of competence, we would be in the most secure era of computing history. But it isn't about competence; it's about intimacy. It's the unglamorous, gritty experience of fighting a machine, losing, trying again, and finally feeling the system yield.

The Knowledge Equation

We can think of the relationship between knowledge ($K$) and friction ($F$) over time ($t$) as: $K = \int_{t_0}^{t_1} (F \cdot \text{Struggle}) \, dt$

The Generational Divide

We are leaning on these new tools more heavily than any tool in human history, yet we understand them far less than I understood a beige box in 1995 that required me to manually rearrange its bits.

The younger generation won't perceive this as a loss. Why would they?

• Tools that do everything.
• Tools that ask for nothing.
• Interfaces as intuitive as a light switch.

The Final Silence

We are the final group of people who knew a machine the only way a machine was ever truly known: by it being difficult. Consequently, we are the only ones who will realize that something essential has vanished.

The other night, I searched for a recording of a dial-up modem connecting. I listened to the dial tones, the hiss, the static, and that sudden, heavy hush when the two machines finally reached an agreement.

The irony is that the machine playing that recording back to me is a machine I will never truly know.