Comment by layer8

The first culture is motivated by solving problems as well, namely how to make the program maintainable and easy to understand. Having the code be an ideal representation of what the program does, having it well-typed, modular and high-cohesion/loosely-coupled and all that stuff, is all in service of that purpose. A program that is hard to understand, hard to reason about, hard to extend and change, is a problem, a problem that is disliked by that culture because it continuously affects them in their programming work when extending or changing the program.

With agentic coding, one can argue that this problem doesn’t affect the software developer any more. But one can also argue that it’s just as much a problem for the AI as it is for the human programmer.

It’s not that the first culture isn’t enjoying solving problems, it’s that it’s enjoying solving a different type of problems than the second culture does, which for example doesn’t seem to care the program should be easy to reliably reason about, or maybe doesn’t realize that the shape of the code is important for that.