> We will no longer distribute source code for the entirety of the Solaris operating system in real-time
In the case of Open Solaris, the code never came out from that point onwards. For Android, the likely end goal is to do the bare minimum of distributing only copyleft code that they don't own copyright to. Until those get replaced with a closed alternative.
I'm trying to think of a mobile OS feature addition that has made me say "I need to upgrade my phone" and it just hasn't happened recently. It's more like, damn it, the dastardly thing stopped receiving security updates and now I have to replace it for no good reason.
Isn't Android done yet? What further development is required that couldn't be done by the community?
If they were to stop, the demand for someone to do it would still be there, and that demand wouldn't be getting met anymore, which creates the incentive for others to do it.
Meanwhile the point is that most of "it" doesn't actually need to be done anyway. You don't need to do everything Google is currently doing. Adding support for new hardware is important, but that has an obvious source of someone to do it because the hardware vendors want their new hardware to be widely supported so they can sell more of it. So all you really need is security updates, and a community can handle that as evidenced by the many instances of it actually happening for other code.
What stops the thing that makes Debian work from making this work?
You're right, if Google steps away from Android completely then there would be incentive for others to do it, another megacorp will step in. Maybe Facebook or Microsoft or Samsung.
Meanwhile there are hardly any devices that come with it, because Macs come with macOS and Microsoft exerts pressure on PC OEMs, so all of the people running it are people who explicitly did want Debian over anything else, as opposed to many millions of Windows users who have no real preference or an active dislike of their operating system but got it by default with the hardware and may not even realize that anything else is available.
If Google stopped developing Android, it would still be one of the two major incumbent platforms and people would continue to use it. It might even get better because third party apps would have to stop depending on proprietary Google APIs/services and then the community could strip out the Google spying code without worrying about losing access to those APIs. So then the question isn't how to get a critical mass of users -- that's already there -- you just need basic maintenance of a stable code base, which is a thing the community can demonstrably do.
Anyway in the world we both live in, if Google abandoned Android then most people would instantly switch to a megacorp fork of Android with that megacorp's own proprietary APIs and services because people will follow the proprietary things they care about like for eg fast & battery-efficient centralized notifications, an out-of-the-box app store with popular apps like Instagram, and tap to pay.
But I didn't have to explain this to you, you already know this. You know this because "millions of people" is derived from you knowing that the peak Linux desktop marketshare is like 4% out of billions of people. You know this because you said users are "worrying about losing access to those [proprietary, spying] APIs" which is why megacorps who provide these proprietary spying APIs will actually win over users. You know this because you know friends or family or colleagues who are aware of Debian and still don't choose it because they rely on some proprietary service or API that Debian's community developers have never given a rat's ass about.