bert hubert 🇺🇦🇪🇺🇺🇦Mar 13
Tempted to write a post that software development lost the plot a long time ago, and that the recent LLM developments are merely the icing on that cake. Software these days is not the painstaking work by people like @bagder or @hyc or @vitaut who write the best code they possibly can. Over the past decade, "the software world" has been developing in a very different way than that.
Show threadRobrab
@bert_hubert @bagder @hyc @vitaut
In a similar vein. https://techtrenches.dev/p/the-great-software-quality-collapse
Software Quality Collapse: When 32GB RAM Leaks Become Normal

Apple Calculator leaks 32GB RAM. VS Code leaks 96GB. CrowdStrike crashes 8.5M computers. How we normalized catastrophe—and the $364B spent avoiding the fix.

From the Trenches
Mar 13 at 1:04pmWeb
Show threadJérôme PetazzoniMar 13
@MortonRobD @bert_hubert @bagder @hyc @vitaut I'm a bit uncomfortable with the fact that this article sounds like the output of an LLM though (complete with weird diagrams and other tell tale sentence constructions, and some technical arguments that a knowledgeable engineer probably wouldn't have made 😅)
Show threadRobrabMar 13
@jpetazzo @bert_hubert @bagder @hyc @vitaut
Interesting. It looked to me like it was originally done as a presentation - bullet points etc. Can you identify some of those technical arguments please? Genuine interest.
Show threadJérôme PetazzoniMar 13

@MortonRobD @bert_hubert @bagder @hyc @vitaut

Ah, if it was initially a talk, that might explain the weird pacing.

Technically, the part that made me wince was this:

“Today’s real chain: React → Electron → Chromium → Docker → Kubernetes → VM → managed DB → API gateways.

Each layer adds “only 20–30%.” Compound a handful and you’re at 2–6× overhead for the same behavior.

That's how a Calculator ends up leaking 32GB. Not because someone wanted it to—but because nobody noticed the cumulative cost until users started complaining.”

That "real chain" looks like a weird salad of unrelated tech, totally out of place in the context of the memory leaks mentioned in the rest of the article.

I personally think that the problem isn't as much the compounding of layers in itself (we already had embarrassing memory leaks in the 90s when the stack was much simpler), but the lack of investment in quality and customer support; and that comes (imho!) from enshittification and the desire to extract as much value as possible for as little cost as possible.