pixari.dev

Senior Engineering Manager focused on DX, AI-Driven Development, and Data-Driven Leadership. Building innovative digital products and inclusive engineering culture.

When the hard part was the point

I still remember the weight of the book.

It was 2003. I was building a text search engine in Perl. I was trying to write a recursive function to traverse a directory tree without blowing up the server’s memory.

I didn’t have Copilot. I didn’t have ChatGPT. I didn’t even have StackOverflow open, it didn’t exist. I just had a heavy, physical Perl manual with a cracked spine, an open space with many loud colleagues and the hum of the computer fans.

I spent days on that function. I remember the frustration. I remember the panic of staring at a cursor that wouldn’t move. But mostly, I remember the texture of the moment the logic finally clicked. It was a physical sensation, a headache dissolving into pure clarity.

For the last 20 years, I have defined my professional worth by my ability to endure that friction. I was a watchmaker, and I took an immense amount of pride in the fact that the gears were incredibly small, the manual was hard to read, and my hands were the only ones steady enough to place them.

And now, I am grieving.

The Robbery

Last weekend, I sat down to work on a personal side project. This wasn’t for work; this was for me. I hit a roadblock with a particularly nasty piece of logic involving data synchronization.

A few years ago, this would have been the best part of my Saturday night. It would have been a ritual: a fresh pot of tea, a blank notebook, and three hours of deep work until I cracked the code.

This time, almost out of muscle memory, I pasted the error into a prompt window.

I didn’t even get to take a sip of my tea.

Four seconds. That’s how long it took. The code appeared. It handled the edge cases. It was cleaner than what I would have written. I pasted it in. It worked perfectly.

I didn’t feel efficient. I felt robbed.

I had robbed myself of the flow state. I had robbed myself of the “Aha!” moment. It was like taking a helicopter to the summit of Everest. Yes, the view is the same. But the person standing at the top isn’t the person who climbed the mountain. They haven’t been changed by the ascent.

The Identity Crisis

We talk about AI velocity. We talk about “10x engineers.” But we aren’t talking about the silence that comes after the code is written.

For many of us, engineering wasn’t just a trade; it was an identity built on suffering. We were the magicians who knew the secret spells. We were the ones willing to read the documentation at 2 AM.

When the “hard parts” become trivial, when the struggle is removed, it forces us to ask a terrifying question:

If I am not the one struggling, who am I?

If my value isn’t in my ability to grind through the logic, and my value isn’t in my encyclopedic knowledge of the Perl manual, then what have I been doing for the last 2 decades? Was I just a slow, biological text-generator waiting to be optimized?

The Wisdom of the Scar

But perhaps I am asking the wrong question. Because there is a fundamental difference between an LLM and a Senior Engineer, and it isn’t intelligence. It’s trauma.

An AI has never been woken up at 3 AM by a critical production alert. An AI has never felt the cold sweat of realizing a migration script just dropped the wrong table in production. An AI has never had to sit in a post-mortem meeting and explain to a CEO why the site was down for four hours.

That terror? That is where wisdom comes from.

The “friction” we are mourning wasn’t just annoying; it was educational. Every time we struggled, we were building a map of the minefield in our heads.

The AI is an eternal optimist. It assumes the happy path will work. It assumes the API will respond. It assumes the data is clean. It has infinite knowledge, but zero scar tissue.

And in our line of work, the scars are the only things that tell you where the ice is too thin to walk.

Where the Friction Lives Now

So, how do we navigate this? If the “writing” is gone, where do we put our obsession?

I’ve realized that we don’t have to abandon our standards; we have to elevate them. We need to take that restless energy that used to go into syntax and pour it into verification and architecture.

  1. Become the Editor, Not the Writer: The AI is an eager, hallucinations-prone engineer. Your new struggle is not creating the code, but having the taste to look at 50 lines of generated logic and spot the subtle architectural flaw that will haunt you in six months. The “friction” is now in the review.
  2. Protect Your “Gym” Time: I have made a new rule for myself. Every day I turn the AI “off” for ~50% of my coding time. I force myself to write code by hand. Not because it’s efficient, it isn’t, but because my brain needs the gym. We lift weights not to move iron, but to keep our muscles strong. We must code manually to keep our intuition sharp.
  3. From Photorealism to Impressionism: Photography didn’t kill painting. It forced painters to stop trying to be photorealistic. When the camera arrived, painters realized: “The machine can capture the light better than I can. So I must capture the feeling.”

We are at that same juncture. The era of the “Photorealistic Coder”, the one who takes pride in memorizing syntax, is over. The era of the Impressionist Engineer is beginning.

Our value is no longer in how we build the wall. Our value is in knowing where to put the window so the light hits the room just right.

The “Good” Struggle

If you are reading this and feeling a twinge of sadness, I want you to know it’s okay.

It’s okay to miss the blinking cursor. It’s okay to miss the frustration of the physical manual. It’s okay to miss the era where the barrier to entry was high, because that height made us feel safe.

We can accept the new tools. We can use the helicopter when the destination is all that matters. But every once in a while, for the sake of our souls, we should still climb the mountain on foot.

I suspect I’m not the only one feeling this phantom limb.