One issue is that tooling and internals have been optimized for individual people's tastes currently. Heterogeneous environments make the models spikier. As we shift to building more homogenized systems optimized around agent accessibility, I think we'll see significant improvements

Elegantly, agents finally give us an objective measure of what "good" code is. It's code that maximizes the likelihood that future agents will be able to successfully solve problems in this codebase. If code is "bad" it makes future problems harder.

> Elegantly, agents finally give us an objective measure of what "good" code is. It's code that maximizes the likelihood that future agents will be able to successfully solve problems in this codebase. If code is "bad" it makes future problems harder.

An analogous argument was made in the 90's to advocate for the rising desire for IDEs and OOP languages. "Bad" code came to be seen as 1000+ lines in one file because you could simply conjure up the documentation out-of-context, and so separation of concerns slipped all the way from "one function one purpose" to something not far from "one function one file."

I don't say this as pure refusal, but to beg the question of what we lose when we make these values-changes. At this time, we do not know. We are meekly accepting a new mental prosthesis with insufficient foresight of the consequences.