Really, though. I think you make a great point and are truly on to something here.
That said back to the aforementioned tiny counter example: Windows, for instance is coming up with things such as Game Mode[1] that are incredibly interesting. This is a feature that doesn't really compromise existing behavior, rather amplifies.
It's proof that some degree of continuous iteration can be healthy, it isn't necessarily healthy all the time. But saying development should stall outright is always a hard sell, there's always a way to improve anything.
[1]: https://arstechnica.com/gaming/2017/03/how-game-mode-will-ma...
The "enterprise" model of software development often has a base product that you install with all of the features but individual features are enabled or authorized by specific authorization keys. This was something VMS did and I'm surprised that Dave Cutler didn't make it part of NT unless it was protected by patent or something back in the day. But generally game mode is a wonderful thing for gamers, turns your PC into an "x-box" like experience, and makes it easy to use from across the room. As a gamer would you pay $25 to enable that? $50? For me a killer new feature is Windows System Linux (WSL) which I use all the time now. Would I pay $50 to enable the "developer edition tools"? Probably. But then I want bug fixes for those tools essentially forever. Pay for features, bug fixes for free.
Even there however you end up with odd cases where things like NFS support, is that "enterprise" software or is it "developer" software or both? You know the "enterprise" guys have no problem paying $500/seat to get "enterprise" support but if they only want NFS client support and they pay the $50 "developer mode" upgrade instead Microsoft might feel like they are leaving money on the table (they aren't but that is a longer post).
Want more RAM for that mainframe? Have some IBM service rep show up, pop the doors open, move a jumper, and bill your company a massive number of dollars.
Intellectual Capital consists of human, structural and network capital. You need to work out the important activities or ingredients and measure where you are and where you want to be, then you can adjust or change your activities to enable you to achieve your new goals.
Paul
I've been using Linux as my primary desktop in the form of Ubuntu since early 2009. What settings are screwed up by up by updates? I can think about init scripts becoming upstart and then systemd. Is it that?
I didn't notice anything else in these years unless we're regarding desktop environments as settings (Gnome 2 to 3 or to Unity). The change of desktop environment has been the most annoying change of the period. Luckily I managed to keep an almost Gnome 2 experience within the Gnome 3 environment using the Gnome fallback mode. However I wished they never spent time working on Gnome 3 and left it as it was. I have the feeling I had to work to undo all their work which is a pity.
Sadly in recent years Linux has taken on more and more of a Windows like black box experience.
What used to be clearly defined lines between parts have been blurred. And what used to be debuggable by running the same commands manually has turned into heisenbugs mediated over dbus.
Sad part is that while this makes me sound like an old graybeard, i am not much older than the people that are implementing all these changes.
Debian provides a generally well supported and documented upgrade process. However it's the sort of thing that you frequently want an expert on hand to do. Still, many people pay to have their car's oil changed.
I have both Windows and Linux desktops, and both annoy me tremendously during updates. On Windows its a new feature that I have to 'click to disable' and on Linux is a new feature that completely screws up my settings and I have to click to go back to the way it was, or construct my settings using the new paradigm.
Both of these annoyances are driven by software change. As the software changes, it changes the environment in which it runs, and symbiotically a bunch of other things change. Why do these things change?
Basically capital. In the free software world it is intellectual capital. Most people who want to contribute to open source, want to do so by "creating a new feature" or "improving and existing feature to be better". Few people want to contribute by "fixing compatibility bugs" or "fixing configuration confusion." They have intellectual capital to spend and they want to spend it on change, not on status quo. In the commercial software world it is worse, they are constantly paying their programming team, every day, every week, 11 paid holidays a year. They can only do that if they have revenue. And they get revenue by selling change not by polishing the status quo. So they change things and they get 'creative' about ways to get revenue.
The ugly truth is that the software business could in fact be "done" for a lot of things. You could just declare the kernel "done" and only allow bug fixes or new processor support. You can declare AutoCAD done and only fix bugs or add GPU support. You can declare graphics Done and simply build GPUs that do what the graphics library need, perhaps a bit faster.
The world changed in a fundamental way when computers got to be 'fast enough' for the imagination of people who wanted to use them. Upgrade cycles started lagging and now it is not uncommon for someone to have a computer for five years before an upgrade. Microsoft needs to make enough money on the OS they sold you once ever 5 years to pay for a crap ton of developers. The math doesn't work, so they are being 'creative' about other ways to make money on those users. Linux/FOSS needs new developers to survive but those developers mostly want to work on 'features' not bugs. So Linux systems suffer arbitrary feature churn.
I think the author's peeve is a symptom not the problem. The problem is that the money is leaving the software space and without it commercial software is not viable.