This is a much better way to put an argument I’ve been saying for a while.
Making code easier to generate, will 10x the number of projects built, which will 10x the need for competent coders to find bugs, maintain the environment, test under other conditions, fill in the gaps for the poor context window or lack of creativity in the LLM.
The demand for competent coders will not fall. It may actually increase, and being even more proficient will be even more valuable.
Yea but what if the ai LLM etc, become the debuggers as well?
Yes! Jevon’s paradox applied to the labor market. We have many reasons to be optimistic about the future of software.
The better analogy is the cobbler. It used to be they made high quality shoes that they would maintain over the life of the shoe. Now shoes are mass produced and disposable, instead of getting you shoes repaired you toss them and get a new pair. Only a minority of shoes, the highest quality, are handmade or maintained.
Software will be disposable and maintenance won't even be possible.
That’s an interesting thought. Although I don’t think it’s entirely true. It will be true for a new class of software between the real products and the user just manually operating things. This is the realm of what I think of as “tiny apps.” Mostly because I have made dozens of them for myself. They are single purpose apps that do some part of my workflow or process, and when I change my workflow, they just get tossed in the bin. But it doesn’t remove the need for an operating system or my editing software, or actually literally any other software I use. It hasn’t eliminated anything, it’s just created a new class where there wasn’t one before
