About a quarter in I figured out, if there was a point to the article he should have gotten to it already, if it's taking this long maybe he just wants to write about watches. So I skimmed and I was mostly right, the way it's presented I think you can probably draw comparisons to a few other things, not just writing software, but it's more of an optional exercise for the reader.

There was even a post on HN yesterday making a similar comparison of LLMs and the impact of quartz in watchmaking.

I think the comparison is warranted, and human-crafted code will perhaps become a brand differentiator in the future too.

Ditto. I kept waiting for the AI comparison. My interpretation was less agentic coding than the commodification of LLMs, forcing Anthropic and OpenAI into a pivot to focus on brand. Anthropic's spat with the DoD could be viewed through that lens: losing money on a deal to better position the brand.