I think the author is missing a key point in that software is vastly more complex than it was in the 90’s. The few times I’ve tried to contribute to open source, I spent more time trying to convince projects that my pull requests fixed the problem without causing other issues. And this is largely because my fixes tend to fix integration problems I had with the software. These weren’t small changes that could be unit tested, and setting up test repro steps were non trivial. I don’t blame maintainers for being slow to accept these kinds of fixes, when the bugs are likely to effect a small minority of users. 30 years ago you could look at a patch and at least reason about it’s efficacy, now you have to have some idea about supporting ecosystems, and not just libraries you’re already using.