The term "Software Crisis" is older than me, and basically boils down to the fact that building software is more difficult and time-consuming than you'd think. There has been a number of proffered solutions for it over the years, LLM-based code generators being only the latest (if perhaps the loudest and most fulsome.) A few of them have helped a little, but in the end the crisis remains because, fundamentally, the problem lies as much, if not more, in deciding the details of what
1/n
,,, you want the software to do as in making the software do it.
I'm starting to wonder, though, if we're now also facing another, deeper Software Crisis in that not merely is software systems difficult and time-consuming to make, but that -- more and more often -- once made, new software systems do not provide any noticeable, or at least not significant, gains.
More and more software systems are written that either underperforms or are simply replacing older systems
2/n
... which provided roughly the same gains, but is being replaced for technical or political reasons. While the new system might be more modern and even slightly better, they're not going to transform anyone's business in the same way that replacing your old car with a new one might give you a smoother rider and better mileage, but you'll still spend the same amount of time commuting.
Now, there's nothing wrong with making replacements for systems that needs to be replaced, but
3/n
Skjeggtroll
Bob Young