Flutterby™! : Still no silver bullet

This is such a good read: Ian Cooper: Is AI a Silver Bullet?

Like previous attempts to find a silver bullet such as 4GLs, it does not seem likely that LLMs as a software authoring tool will be succeed. In particular, like 4GLs they suffer from the problem that because software spends most of its life in maintenance, the cost of change makes most improvements to the cost to author irrelevant; in fact, some rapid authoring techniques make maintenance harder and increase the lifetime cost of software.

And, yeah, it's nominally about LLMs and AI for code generation, but really it's about complexity management. As I try to debug Mithril TypeScript today, with the wacky stuff that Safari does to code maps (and, yay, TypeScript's error messages), and think about some of the performance issues that we're seeing because we're blindly accepting Google's Firebase client libraries without really understanding what that abstraction layer imposes on us, I've been thinking a lot about the abstraction layers that have had lasting value.

SQL. OpenGL. I think there's some reasonable value in the scene graph of your choice, but... I'm not sure that we've seen a good replacement for CGI for web serving yet, it sure seems like the ability for PHP and Perl scripts to cohabit on a server has been lost with the "fuck it, let's put everything in its own virtual space" direction that everybody's gone (with all of the wasted RAM and CPU cycles of all of those services waiting around for a few hits a minute).

And we really don't have a better mechanism for keeping artifacts around than the shell, despite 4+ decades of attempting GUIs. Hopefully work project will get there, but...