I was also at Mozilla during the time period Andreas outlines--I worked on the Firefox OS test team, for that matter, almost from the beginning of it graduating from Boot2Gecko as a Labs project until right before it got killed.
While my perspective isn't as strategic or metrics-driven as his, I had a lot of time to observe and think as both a community member and Mozilla employee. FxOS also wasn't my baby, so there's that. Note also that I speak for myself here and my own observations and paraphrases--whatever I say that pisses someone off is something I'm saying, not that Mozilla said verbatim.
My primary takeaways were twofold:
(Long, TL;DR at bottom)
1) I agree with you. Desktop and mobile are two separate markets, period. The first mostly serves a workplace audience and the second a personal audience, but most people with a desk job at the very least will use a web browser as part of their day. Desktop may be a minority of the overall, but it's a minority that won't go away anytime soon and so will continue to influence HTML and standards disproportionate to pure market share.
That's important because Mozilla's gambit for preserving the open web was pretty simple (I say was because I think they're just not that focused at this point):
Have enough people using your browser that websites absolutely have to support the emerging standards that browser relies on-- and perhaps in doing so make it less attractive for site providers and browser providers to spend time on proprietary tech that isn't significantly better than those standards, thereby making other browsers move to those standards too.
Doesn't mean these people have to use it everywhere, or that it has to be a majority share (10%+ was what I commonly heard as "enough") or otherwise "win". But it does have to be enough that people will complain if the website doesn't support their browser and that testers are influenced to test the site against it.
(BTW, as a test professional, the fact that Firefox no longer appears in most test matrixes I encounter due to lack of a blip on analytics is very telling, and Firefox has a serious risk growing around site incompatibility or instability in their browser).
That brings me to my second takeaway:
2) The grand majority of people don't use a web browser because of the browser itself; they use it for one of a few reasons:
a) It's default on their system.
You will not win these people over because they're not there for any reason other than it being the easiest or most integrated path. Note that this is pretty much the whole mobile market, and why it was a dire mistake for Mozilla to conflate the two markets, decide mobile was more important due to combined market share, then go tilting at windmills.
It's also, any altruistic reasons aside, why the moonshot was to create an OS so Firefox could be the default mobile experience.
b) Ethics/Community. This was a relatively small but very vocal part of Firefox's userbase. Probably more people were there "against Chrome" than "for Firefox," but whatever. Firefox succeeded in the first place because of "against Internet Explorer" so it's a valid reason to be there. The nice thing about these people is they pull in more people.
Unfortunately, one side effect of Firefox OS as a project was working with proprietary partners who emphasized confidentiality such that you couldn't share with the community in the way Mozilla did before. When Mozilla diverted most of their effort to Firefox OS, it froze out a lot of the community efforts.
I think Mozilla-the-org also became less skilled at working with community, both for that shift and perhaps because they brought in a lot of people from the mobile and other sectors who didn't have that background.
Whatever the case, this base wasn't well-maintained, and I don't think operates as a core in the same way it might once have, at least for Firefox.
c) Customization, and this is really why Firefox (and Mozilla) is where it is today IMO:
The add-on ecosystem is the most compelling reason both to start using but especially continue using a non-default browser, and Mozilla chased theirs away.
Here's the thing:
You have one browser, whose job is to get the hell out of the way of the content and fade into the background, and several installed extensions whose jobs are to solve specific problems you have. 1 browser. Several extensions. 1 invisible, generic browser. Several hand-chosen extensions that make it your own.
Which bit are you sticking around for? Probably the extensions. The browser is a platform, a means, not an end--not just for the web but for its own functionality.
That's why all the complaints are performance and crashes and things that make you notice the browser, not generally native features or UI. It's the whole rationale behind browser-as-OS efforts in the first place.
Syncs and reading lists are nice, but with an ecosystem you can get them after the fact. With no ecosystem, you'd better either really like what you're handed or be trapped into using it a la iOS--and even Apple has figured out they have to be somewhat customizable or people will even switch OSes to get it.
That's where Chrome won.
Specifically, when Mozilla pretty much cargo-culted Google's rapid release program without first creating a version-independent extension store and add-on API, they seriously fucked up.
A ton of add-ons couldn't keep up with release and broke--Firefox's add-on "API" was to either essentially monkeypatch the internals or to create binary extensions, neither of which had any real abstraction from implementation. The need to manually update the compatibility version number and reupload to AMO for every single release froze more out (this was fixed server-side by auto-incing the number but it was just a workaround to a real problem as now extensions might not be really compatible). AMO always had a problem with turnaround on submissions, and this didn't help.
In general Mozilla sent the message of being out of touch with the add-on community, and that it wasn't a clear priority.
Jetpack/Add-Ons SDK was there as a not-really-mature portability measure to mitigate this, but it wasn't powerful enough to support a lot of the best extensions. It also happened in parallel with the rapid release shift, not before it when it was strategically needed. Then Jetpack kind of died off, probably partially because of the shift in efforts towards mobile/FxOS and partially because Jetpack never really became kick-ass enough to make the type of impact that gets more resources.
Meanwhile, right before Firefox stepped into this and never stepped out again, the Chrome Web Store opens in 2010, ready to receive any disenfranchised add-on makers. Talk about right place at the right time. Now Mozilla finally is adopting Chrome's extension API, both because it probably works better with multiprocess but maybe also to get some cross-platform extensions back. I imagine that'll work OK, but maybe too little too late.
AMO isn't what it used to be at all. Meanwhile, Chrome's Web Store is phenomenal. Combined with the (yes, unfair) Google Suite integration in Chrome there's a lot of friction to keep people on Chrome as a platform, and short of Chrome making a huge mistake that shakes people off like Mozilla/Firefox did, there isn't much reason to switch browsers for a cross-platform extension that's already on the one you use.
And when I look at the blog posts re: Firefox improvements, etc., I keep coming back to "it's a platform, not an app" and wonder if they'll ever take that ground back. Maybe if they somehow completely redefine the browser experience or otherwise stay deeply opinionated (Opera is sort of going this route) but otherwise, it'll just be parity and parity doesn't move people.
I'm thoroughly convinced Mozilla won't achieve much impact with Firefox until they understand and execute on building the ecosystem as the vast majority use case and never damage it again. However, that means them also understanding that people came to and left Firefox for any number of reasons other than Firefox itself or its native features, interface, etc., and I don't see that yet institutionally in their execution.
Some bright spots: Rust and WebAssembly are both fantastic. In general, I think Mozilla has been great at addressing the developer segment, both with the browser dev tools themselves and especially with MDN. Maybe what'll eventually preserve Mozilla is getting drummed out of the web browser business and into more of a core tech business. Regaining browser relevance is going to be a heck of an uphill for them.
TL;DR: Mozilla should never have gone full-tilt at mobile and made a grave mistake in assuming desktop was becoming irrelevant to their mission; and by far the biggest fuckup was alienating the add-on community since customization is the most compelling reason why people would use a non-default browser. It's entirely possible they won't come back from that. Their biggest recent impacts have been peripheral to or completely orthogonal to the browser, and might be their brighter future.