Today's Read: "The bottleneck was never the code" (Rémi Louf)

Enjoyed reading Rémi Louf's "The bottleneck was never the code" today. Lots of alignment with the "things have changed but also not really" feeling I've had for the past year or so. A few highlights I really resonated with:

...
Software is what’s left over after a group of humans finishes negotiating with each other about what the system should do. The code matters, but it is the residue of the harder work, not the work itself.
For fifty years the residue was expensive enough to keep our attention on it. Typing speed, language design, framework choice, IDE plugins, code review tooling were all about reducing the cost of the residue. With coding agents the cost has fallen far enough that we can see what’s underneath: people trying to agree.
Negotiating, agreeing, communicating the shared picture of what we are building has become the work. And it’s just as hard as it was.
For 50 years the residue was expensive enough to keep our attention on it. With agents the residue got cheap — and the negotiation underneath got visible.
...

I've been expressing this to people as "coding was never the job". My philosophy/approach to/of engineering has always been practical, applied; the game has always been about the outcomes created, not just the playing of it.

Yes, the craft matters and there's a certain inherent goodness to just the thing itself. I love playing catch and could toss a baseball back and forth with my son pretty much endlessly. It's also true that all catch, no baseball game is less interesting and good. I want to play, lose sometimes, win more often, and get to the championship.

To me, every process construct, technology choice, architectural decision, test case, microservice, database schema are about creating something greater than just a sum of parts.

The new moat is organizational, not technical.
The companies that win the next decade will not necessarily have the best models or the best agent infrastructure. It will be the companies whose fifty people, then two hundred, then two thousand, can stay aligned on a shrinking set of decisions while shipping more output per head. They will be the ones that already knew, before agents arrived, that their hardest problem was coherence.

"AI an amplifier" has held true to me for the past few years. Good to great, bad to worse. Seen this happen at the individual and organizational levels.

Agents are overestimated as a way to make individuals write code faster, and underestimated as a way to make organizations externalize what they know.

A robust writing culture has always been valuable as way to encode context, history, intent, etc. It matters as an alignment tool, to help onboarding, explain decisions, describe a vision, and more.

I'd argue that without the creation of a strong writing and context-creation culture, the speedups from AI tools will be capped, degraded, or wiped out over time as 'organizational context rot' kicks in.