Indeed, my sense is that Windows Phone 7 was picking up some decent momentum, and then it was scuttled by the Windows 8 initiative. There was probably some guilt by association with the ill-fated desktop OS. But there was also the confusing compatibility story between Windows Phone 7 and Windows Phone 8, which I found to be confusing and slightly intimidating both as someone who was shopping for a phone at the time, and someone who was considering developing an app for Windows Phone.

I get what they were trying to do, and it was certainly an interesting and ambitious goal. But they made the same mistake I repeatedly make when I'm trying to kindle a campfire: putting the fire out by throwing way too much wood onto it at the first hint of a visible flame.

Microsoft couldn't get over the hurdle that US phone carriers refused to sell a 3rd option at all, no matter how much money or time they threw at that problem. They could invest in physical retail Microsoft Stores all they want, but they were locked out of the stores US consumers actually buy that hardware form factor from, and in many cases they couldn't even get SIM cards for most of the US networks even if they did manage to convince people to buy their next phone from a Microsoft Store.

Against that kind of gatekeeping, of course there is no third option in today's market! It's harder to sell three things than it is to sell two things, and AT&T/Verizon/et al decided that they didn't want to sell three things.

(Sure, Microsoft could have trundled on in other markets than the US. They had such a sizeable India marketshare against the troubles in the US, for instance. But when you've already lost your "home" market, it's hard to keep people's confidence. Especially in that case where Microsoft got the biggest marketshare gains from cheap hardware from a variety of hardware partners, those thin margin partners are nothing but confidence driven.)

> It's still a great mystery to me why it seemed like the swipe keyboard (Word Flow[0]) on Windows Phone was so, so good, and yet even SwiftKey which is owned by Microsoft doesn't measure up

I worked with the guy who lead the swipe keyboard on Windows Phone. Amazing machine learning engineer who lead an amazing team. Some of the work he did for my team was out of this world (ML models to do predictive touch inputs to decrease perceived touch screen latency and improve touch recognition accuracy, on a Cortex m3)

Unfortunately for whatever reason his code was never adopted outside the mobile org, and I believe he was moved off that project after its initial version was completed. This is also why the desktop Windows swipe keyboard for Windows tablets at that time terri-bad compared to Windows Phone's swipe keyboard! Also this is possibly why the keyboard got worse, not sure it was long enough ago that I remember it was mid-Win8.1 mobile -> 10 transition but I don't recall all the details and the timeline.