RE: https://social.luca.run/@luca/117455226271149354 This is an interesting take but I have to point out that it’s not a new phenomenon. From the dawn of computing we had single use software. Basically all first programs were single use as described in the OP: hardcoded variables, no rights management, too specific. That’s mostly because for a while computers were too resource constrained for reusability. Programs had to cut every piece of configurability to be as small as possible. It was easier to write a new program because CPU time was expensive and scarce and people time was cheap and abundant. Reusable software started to appear a bit later when computers started to become more capable and numerous. It became somewhat viable to write reusable software that would accommodate at least a class of a problem and sell it to multiple computer users. Later at the dawn of personal computers we again see explosion of single-use software. You had to type in your program every time you wanted it to run because you had too little disk space (or no at all). So programs again had to cut every unnecessary part to be usable. A popular distribution model at the time was a program print out. You could find programs in magazines. But you still had to type it to run it. As PCs became more capable that sort of faded away but even so many people still keep producing single-use software. That script you wrote to rename a bunch of files. That piece of JS to animate something on your home page. Your company’s CI pipeline. All have hardcoded variables, too specific, and no general purpose rights management. And all are hard to maintain. AI might’ve expanded the size of single use software but it hardly invented it.