> Why does dev-owned testing look so compelling in theory, yet fall short in almost every real-world implementation

Seems like a weird assertion. Plenty of startups do “dev owned testing” ie not hiring dedicated QA. Lots of big-tech does too. Indeed I’d venture it’s by far the most common approach on a market valuation-weighted basis.

It's very common and most software is extremely buggy, so that adds up.

I think most consumer software just would not stand the scrutiny of enterprise or professional applications. It just does not work a lot of the time, which is truly insane to think about.

I mean, imagine if your compiler just 10% of the time did not work and output a wrong program. But when I log into website XYZ, that's about the rate of failure.