AI didn't solve coding—it exposed what software engineering really is
"Coding is solved" misses the point
Every few weeks, someone declares that coding is solved. If they mean turning a well-specified problem into runnable code, they're increasingly right. But was writing runnable code ever the whole problem? Language models have made implementation dramatically cheaper, shifting the bottleneck to understanding business problems, navigating organizational context, and making sound decisions within existing systems. This article argues that software engineering has always been more than coding, and AI is revealing the layers above implementation that truly matter.
While the programming problem is similar, the environment makes it harder.
- globular-toast
Even "add a button" should be questioned. See the XY problem.
- 9rx
> If they mean turning a well-specified problem into runnable code, they’re increasingly right.
Has coding ever meant anything else? This is just semantics, but I don't think a product person talking to customers to find out what they need has ever considered their work to be coding, even if the fruits of their discovery eventually ends up as code.
> Code was always the medium
Exactly. If you discovered that what the customer really needed was a wooden cabinet you'd build that instead. You wouldn't give them code and expect it to hold their belongings. Code has never entered the picture until you have something that is well-specified.
- hackable_sand
I think it's easier to explain the anomalous "chess is solved".