It failed in perception, no one developed for it and why would they when MS could break their app compatibility any second with just a single update. It was already even less cool to use than Blackberry, to much enterprise and too little user focus. At the time geeks were unhappy about how the things were going with Windows on PCs. Partnering with MS and taking Elop as their CEO was the final nail in the Nokia coffin. They really should had partnered with one of the asian companies like HTC, they were fierce competitors at the time and guess who bought their mobile division and continues to make pixel phones there. It's easy to say that after the fact, but they could have stayed a niche player, instead Elop chose to completely devalue their brand.
By the time Elop took over at Nokia they were already doomed. The best they could have done was simply become another Android OEM.
The final iteration of Windows Mobile was excellent, what killed it wasn't marketing but ecosystem. MS wasn't competing with just Apple or Google. If they were they could have stood a chance, they're just as big, just as technically adept, had just as good a platform (in the end) and just as well capitalised. They could go toe to toe with any company in the world.
They weren't just competing with Apple and Google though, they were competing with them, plus all the existing handset manufacturers that had invested in Android, plus all the developers that were developing or had already released apps on Android or IOs, plus all the companies providing services for those platforms. The ecosystem provides ~10x or more market power than the platform owner on their own. That is what you're actually competing with when you go up against a market incumbent platform owner. It's the crucial difference between a platform and a product.
Sinfosky was responsible for tanking Windows Phone.
Windows Phone 7 could have evolved from Silverlight/XNA models, while evolving the support, maybe adding C++ support in the mix (On WP 7 C++ was only available to selected partners).
Instead WinDev just killed it (similar to how they torpedoed Longhorn), came up with WinRT using an incompatible variant from .NET, that not only dumped Silverlight/XNA, it required multiple reboots (WinRT, UAP and then UWP) always asking us to rewrite the applications between each reboot.
MSIX, WinUI (desktop UWP), .NET 5, C++/WinRT are just the long roadmap of fixing those issues, while trying to make everyone happy again.
Windows Phone 7 was still based on the old Windows CE kernel which couldn't support multiple cores, didn't have true multitasking or support background services, had a limited network stack, the list goes on. Windows Phone 8 finally came out with a true multitasking NT kernel based OS that could compete with the iPhone, 5 years after the iPhone launched.
This is why developer access to the system was so limited, it simply didn't have the sort of process isolation capabilities and the system services this enables that you need in a modern app platform. There's no way they could just open up low level programming to anybody to develop for it and everything would be fine, the skills and knowledge needed to develop for it efficiently and safely were highly specialised.
It’s not true that CE could not support multitasking. Originally CE up to 5.0 had a limitation in the number of processes it supported (32), in order to support more performant task switching (no TLB flushes for one) - though it did not have such a tight limitation on number of threads.
However the version of CE (6, released way back in 2006) that was released on Windows Phone did not have this restriction and version 7 released in 2011 also supported SMP.
Windows CE was really not an entirely incapable OS (and I personally despised it). Its real annoyance was having a completely non-standard bastardized/limited Win32 API which made it both an odd ball embedded RTOS and only vaguely similar to the API of Windows NT. Despite this it had considerable adoption in industrial computing — never a huge market though.
It’s hard to say this was a real problem though as no one considered OS X to be a phone OS before 2007 either. The bigger issue is that clearly MS wanted out from under CE before the phone was even released.
The limitations in Phone 7 were higher level, not due to the underlying kernel.
Sure. They could have pivoted and partnered with automotive industry, Maemo legacy isn't even completely gone yet and it sparked a whole new generation of open source developers. Also as far as I remember Qt were still used in cars.
> By the time Elop took over at Nokia they were already doomed.
They had a lot of money and a lot of market share. They could have culled a large part of the organisation, especially due to the insane policy of self-competition resulting in a lineup of almost-but-not-quite-identical handsets, either put more into Maemo/Meego or jumped ship to Android.
Instead they had an MS plant installed at the top who seemed determined to devalue the company then sell off everything to MS, who then just ran the brand into the ground while failing to get Win Phone to launch in any meaningful way.
> By the time Elop took over at Nokia they were already doomed.
Yeah, someone in Nokia engineering had made some company destroying miss calculations.
It's not like the didn't know what they were up against - it was race between engineering teams. And it's not like Nokia the company couldn't fund that race - they were a gorilla in the ring when it came to resources. They even saw the train coming at them, knew what it would do to them, did some awesome things like buy QT in preparation but it was nowhere near enough.
Blind freddy could see the pace of development happening on MeeGo and is predecessor (I forget what it was called) was nowhere near fast enough. Given the resources I expected Nokia to be throwing at it, either the team behind it was tiny, or they were tied up in some titanic tide of red tape go-slow goo.
It's not like it was an impossible ask. That was back in the days of Android Honeycomb and Gingerbread which were clearly something Google has slapped together in a rush. They were IMO barely usable. iPhone was barely more than a phone + ipod back then, granted with the best UI on the planet interface but even then the gulf between what Android would allow you to do what would be possible in Apple's walled garden was apparent, so there was space in the market for a different mix.
> The final iteration of Windows Mobile was excellent, what killed it wasn't marketing but ecosystem.
Well in that case it was obvious what happened. I've lost count of how many "throw it away and start again" iterations Microsoft went through. I do recall using one of their earlier attempts when they had very little competition. It was a Win95 interface crammed into tiny resistive touch screen, complete with start button. It was unusable without the stylus they provided. It had an uptime measured in hours - literally far worse than Win95, which was an amazing "achievement".
They threw major bits of it away and started again, and again, and again, each time with something that was 100% binary incompatible with the previous version so they never built up a customer and app base. Eventually they ended up on the NT kernel with an amazing GUI toolkit that as you quite rightly say was the best in class by any number of engineering metrics, but by that time the network effects of their competitors building up an enormous customer and applications by keeping compatibility completely annihilated them. Numbskulls.
Regarding Windows Phone, on one hand I was pleased to see what should have been .NET in first place, managed runtime built on top of COM, but the way it was rolled out initially incompatible across mobile, tablet and desktop (first WinRT iteration), it was just a mess.
Not only were Windows Phone 7 devs being asked to throw out their beloved Silverlight/XNA tooling, they were being told to rewrite the app three times with #ifdef, and in XNA's case to also move into C++ bare bones DirectX (here is when DirectXTK was born as XNA-like for C++).
