Of Tools and Timber

Kant meant this as tragic realism about human nature. But there's another reading: if you want straight timber, use tools that compensate for the grain.

The CNC Machine for Ideas

A master woodworker can cut perfect joinery by hand. It takes years of practice, deep material knowledge, and absolute focus. The result is beautiful—craft in its purest form.

A CNC machine can cut the same joint in minutes. You program the design, set the parameters, verify the output. The joint holds just as well. Often better.

The question isn't which method produces better joints. The question is: what are you optimizing for?

If you're optimizing for craft—the meditative practice of hand work, the satisfaction of doing it the old way—hand-cut joinery wins. But if you're optimizing for structural integrity, speed, and the ability to build something larger than a single artisan could produce in a lifetime, the CNC machine is the obvious choice.

This is how I think about working with AI.

How I Work

I think in dialogue with reasoning models. My role is to:

• Supply structural intuition—what the thing should be
• Define constraints—what counts as true, coherent, or useful
• Judge outputs—accept what survives interrogation, reject what doesn't

The AI's role is to:

• Render half-formed intuitions into articulate prose
• Generate conceptual variations at speed
• Synthesize patterns I describe but can't yet name

We iterate until the output reflects my intent. I'm responsible for what survives, not what gets generated. If the AI produces something real and useful, I use it. If it generates plausible-sounding nonsense, I kill it.

This is the CNC approach to thinking: I program the constraints and verify the output. The machine handles articulation and synthesis.

On Writing

I think structurally—in frameworks, logical dependencies, causal relationships. The AI translates that structure into readable prose. I then edit for precision, coherence, and fidelity to intent.

This is specialization. I focus on what I'm uniquely positioned to do: pattern recognition, domain grounding, knowing what questions matter. The AI handles what it's good at: fluent generation, conceptual recombination, rendering structure into language.

The result is faster, clearer, and more internally consistent than what I could produce alone. Not because the AI is smarter—because it's a better tool for this specific task.

Why This Matters

Some people find this troubling. They see AI-augmented writing as fundamentally inauthentic—like using a CNC machine betrays the spirit of woodworking.

They're not entirely wrong. There is real craft in doing things by hand. There's a meditative quality to the slow work of shaping language sentence by sentence. But if you're building something that needs to stand, you want joints that hold. The method is secondary.

The output is what matters. Judge the lumber, not the mill.

The Real Question

The question isn't whether to use AI. The question is: what are you optimizing for?

If you value the craft of unassisted writing—the experience of wrestling with language alone—then by all means, write by hand. That's a legitimate optimization.

But if you're optimizing for speed, clarity, and the ability to externalize complex ideas faster than traditional methods allow, then AI-augmented thinking is the obvious tool.

I'm not interested in craft for craft's sake. I'm interested in building ideas that hold weight. The CNC machine lets me do that faster and better than I could alone.

In 2025, avoiding AI because it "doesn't feel right" is like avoiding power tools in 1925. You can do it. But you're limiting yourself unnecessarily.

The tool exists. The question is whether you'll learn to use it well.