I'm not saying that all change should be avoided. I'm saying that all change is an annoyance to your users, all change has a cost. So if you do want to force your users to change, provide them with an escape hatch so professionals can avoid that, or be really certain that the change is genuinely making things better.
Imagine that every change you're making is like hitting your user in the face with a brick. If you're going to give me something amazing that makes my life better, I may let you hit me in the face with a brick so I can get it. But if you hit me in the face with a brick and then you give me something worse than what I had? Don't do that.
Learning the basics of Vim was a really big, hard change for me, but it was an adjustment to my workflow that was made for a specific reason, that made my life better and that made me more productive, and that (importantly) was a decision I made voluntarily. It was not change for its own sake.
> Would we still be able to have startups at all if products required thousands of hours of QA or perfect test suites in order to launch?
On the contrary, how many more interesting, better chat apps would we have if every one didn't feel the need to reinvent Markdown? Wouldn't it have been more useful if instead of rebuilding their editor for no reason at all, Slack's engineers instead added new API endpoints, or experimented with encryption, or added new search tools, or addressed any of the pain points that actually get raised by professionals using their product?
I would also push back against the idea that this is simply the cost of experimentation. Vim/Emacs are much more experimental editors than Word, yet even modern remixes of those editors like Spacemacs put more thought into user customization and consistency than Word does. Spacemacs' keybindings evolve -- but they never force you to accept that evolution if it would break something fundamental to your workflow. And Spacemacs is doing way more experimental, interesting stuff than Slack is.
To add further onto that idea, prioritizing future innovation over the productivity of real users is a very software-specific philosophy about how a professional field should work. New animation techniques come out all the time, but we don't look at people like Miyazaki and say, "that man is holding us back." It seems to be a software-specific scenario where a change is proposed, users say, "I don't like it", and then engineers get somehow upset about that fact, rather than just saying, "cool, it was an experiment. Let's revert and move on."
I did not get into programming because I love computer interfaces. For me, being treated like a professional means that a company makes me specifically more productive. I don't care if a change makes things better for someone else if that change is making it harder for me to do something I love. As a professional developer, it is OK to demand tools that work well for you. Again, this is the case for every other field -- no one expects a professional woodworker to switch to a new measuring system that they don't want to use, even if that system is popular with some people.
Of course no field gets this perfect, but virtually everyone gets it better than us. I'm not demanding a theoretical utopia I can only imagine. I'm looking at every other professional field and saying, "why can't we have the stuff that they have right now?" It's very much a feeling that's informed by how good it feels today to work with physical mediums as an artist, and how utterly crappy it feels by comparison to use Skype or Windows 10.
And if that means that software moves slower, who cares? It's not theoretical to me. Other mediums are better to work in -- so whatever they're doing, we should copy, because they're better than us.