The Ladder Problem

The Pitch

The pitch is always the same: AI will handle the tedious parts so developers can focus on the interesting problems. Junior devs won't need to write boilerplate. They won't need to struggle through their first CRUD app. They won't need to debug a null pointer exception at 2am because they forgot how reference types work. The AI handles the grunt work, and humans handle the architecture. It sounds efficient. It sounds humane, even. It's also how you build an industry that can't train its own replacements.

What Junior Work Actually Does

The tedious tasks aren't obstacles on the path to expertise. They are the path. Writing boilerplate teaches you what boilerplate is abstracting. Debugging null pointers teaches you how memory works. Building your tenth CRUD app teaches you which parts of the pattern are load-bearing and which are convention. Struggling through a bad API teaches you what a good API feels like: not as a principle, but as a physical memory in your hands. I spent my first year at Logpoint documenting six security products I didn't understand. SIEM, SOAR, UEBA, Director, Integrations, Security Research. Nobody assigned me all of them; I picked them up because someone had to. That year of writing about systems I was still learning was my staircase. It was tedious. It was also the reason I could later build an AI system that evaluates documentation quality: I had internalized what good docs feel like, not as a checklist, but as a bodily sense. The senior engineer who "just knows" that a particular architecture will cause problems at scale? That knowledge was built rung by rung, from the accumulated residue of things that went wrong in smaller systems. Remove the rungs and the intuition never forms.

The Ladder Metaphor

Imagine a company that installs elevators in a building and then removes the stairs. The elevator works fine. Everyone gets to the top floor. Efficient. But now nobody develops the cardiovascular capacity that stairs provided. And when the elevator breaks (which it will), nobody can climb. Worse: the people who designed the elevator were people who once took the stairs. They understood what the elevator needed to replicate. The next generation of elevator designers won't have that understanding. They'll optimize for speed and smoothness, not for the structural requirements they've never experienced. This is not a theoretical problem. It's happening now.

The Compounding Effect

The first generation of AI-assisted developers still has the foundation. They learned to code before the tools arrived. They can evaluate AI output against their own experience. They use LLMs as leverage on top of existing intuition. The second generation learns with AI from day one. Their mental model of programming is "describe what you want and something happens." They may be productive. Perhaps even more productive than the first generation was at the same stage. But productive and competent are different things. Productivity measures output. Competence measures understanding. For a while, you can have one without the other. But eventually the gap shows up as brittle systems, cargo-culted patterns, and debugging by trial and error because nobody understands the layer below.

What We're Not Talking About

The conversation about AI replacing developers focuses almost entirely on jobs. Will there be fewer developer positions? Will salaries change? Will entire roles disappear? Those are real questions. But there's a deeper one that almost nobody is asking: what happens to the development of developers? Jobs are about economics. Development is about epistemology. We're automating the process by which people learn to think like engineers. And treating that as a feature.

The Uncomfortable Middle Ground

I'm not arguing against AI tools. I use them constantly. They make me faster and more capable, precisely because I have the foundation to use them well. The question isn't whether to use the tools. It's whether we're designing pathways that build the foundation the tools require. Right now, the industry incentive is to skip the foundation entirely. It's slower, it's harder, it doesn't look good on sprint velocity charts. The ladder doesn't just get you to the top. It builds the muscles you need to stay there. Remove it, and nobody knows how they got where they are. Or how to get back when they fall.

We're building an industry of people on upper floors who have never seen a staircase.