Why did Symbian lose out to bigger, slower, less-efficient OSes?
Symbian was OK. EPOC, its progenitor, was in some ways better. (I write as a Psion owner, user and -- TBH -- fan.)
AIUI, and I do not have good solid references on this, EPOC was a very early adopter of C++ as opposed to plain old C, and as a result, it did many things in extremely nonstandard ways compared to later C++ practice. Its string handling, error handling and all sorts of things was very weird and proprietary compared to the way that the greater C++ community ended up doing.
This meant that by the time EPOC turned into Symbian and entered the wider, more competitive market of cellphones, its "proprietariness" counted strongly against it. If you wanted to develop on Symbian, you had to learn the weird Symbian ways of doing things, not the ways that worked on Windows or Unix or Mac.
Also, everything was very in-house - a bad case of NIMBY-ism, but they had no choice, because when they started out, the standards didn't exist. So one of the reasons for Nokia's acquisition of TrollTech for Qt was that Qt on Symbian would have been a more standard way of building apps for Symbian than using native Symbian tools - and since Qt is cross-platform, in theory, you could port a Qt app from Window or Linux or MacOS to Symbian (or /vice versa/) far more easily.
However, this meant that Qt-based Symbian apps were much bigger and slower than "traditional" Symbian apps. This made them uncompetitive with native Symbian apps, and meant that Symbian phones had to become bigger and more powerful in order to compete with Android and iOS and so on. Traditonal Symbian was tiny and its apps were tiny. So, vicious circle: Nokia had to tempt developers to Symbian. This meant making it easier to port to, or write for, Symbian.
They chose Qt, over internal dissent. Result, more apps, but bigger less efficient apps. Meaning they needed to up the spec of the phones. Meaning that they threw away Symbian's core advantage - that it was tiny, used little power and could run quickly and responsively on low-spec phones. Which meant cheaper phones and kept Symbian competitive in the market.
The pre-Symbian smartphone OS, inasmuch as there was one, was GEOS - a Commodore-64 derived [yes, seriously] multitasking layer on top of MS-DOS [yes, really]. This meant a 386-based phone.
Symbian scored over that because it ran on the much smaller, much more efficient ARM chips.
Then once ARM chips caught up, Unix came to ARM chips - Apple's iOS is a Unix, Android is Linux which is a Unix, some of the early Chinese smartphone (and PDA) OSes were Linux-based too. So the ARM hardware suddenly expanded and got more RAM and more storage and more MHz, in order to run Unix and Unix GUIs well.
Result, easier development of pocket-platform apps, using existing desktop Unix tools.
Result of that, Symbian is now competing with a bigger, more capable OS with decades more development and a huge range of tools - many of them free, as were many of the OSes.
With hindsight, there's a sort of majestic inevitability to it all, but it was not really possible to see it coming.
For example, I had a Sharp Zaurus - the SL-5500, with its weird little slide-out keyboard. That would be about 2003, 2004 when I got it. (I had 2 actually.)
They were clever, but dreadful. Linux on a 200MHz CPU with 64MB of RAM, no swap, and just 16MB of permanent storage in the form of Flash - so, yes, less "hard disk" than RAM! - was not a good platform. It was slow, it was unresponsive, with so little memory it was unstable, and it *ate* batteries. Pocket sized device, laptop battery life: 2hr or so.
At that time, it was "obvious" that this wasn't competitive and that Symbian had compelling advantages.
By 5y after the first Linux Zaurus appeared, the first iPhone appeared, essentially running NeXTStep - i.e. FreeBSD on Mach - in your pocket, and it was smooth and fast and gorgeous and really stable.
By the time that was out, it was just a matter of counting the days for Symbian.
What Apple did, Android copied, but with FOSS tools. (Early Android was a Blackberry ripoff. That was hastily dumped and replaced with a finger/touchscreen UI that was clearly directly inspired by iOS.)
Once the pocket hardware caught up, there was no need for a weird, proprietary, lightweight, paid-for pocket OS when you could use a free one and free tools for a better development environment and a better user experience.
So Symbian had its time in the sun and it did OK.
I believe Symbian phones continued to (and maybe still do) sell well in the developing world, where people don't have landlines or indeed mains electricity and they are all jumping directly to cellphones and mobile-phone Internet. These markets are /extremely/ price-sensitive and thus the fact that a Symbian phone could run well with a quarter of the CPU speed, RAM and Flash storage of an Android phone meant that you could sell them at a profit for US$100 or less.
But the gap is narrowing. Look at the Raspberry Pi. Half a gig of RAM and three-quarters of a gigaHertz sold at a profit for $35. That's usable with Android and more so with lower-end pocket Linux OSes such as FirefoxOS (which in a way is Palm WebOS but done with more mature, all-Free tech.)
AIUI, and I do not have good solid references on this, EPOC was a very early adopter of C++ as opposed to plain old C, and as a result, it did many things in extremely nonstandard ways compared to later C++ practice. Its string handling, error handling and all sorts of things was very weird and proprietary compared to the way that the greater C++ community ended up doing.
This meant that by the time EPOC turned into Symbian and entered the wider, more competitive market of cellphones, its "proprietariness" counted strongly against it. If you wanted to develop on Symbian, you had to learn the weird Symbian ways of doing things, not the ways that worked on Windows or Unix or Mac.
Also, everything was very in-house - a bad case of NIMBY-ism, but they had no choice, because when they started out, the standards didn't exist. So one of the reasons for Nokia's acquisition of TrollTech for Qt was that Qt on Symbian would have been a more standard way of building apps for Symbian than using native Symbian tools - and since Qt is cross-platform, in theory, you could port a Qt app from Window or Linux or MacOS to Symbian (or /vice versa/) far more easily.
However, this meant that Qt-based Symbian apps were much bigger and slower than "traditional" Symbian apps. This made them uncompetitive with native Symbian apps, and meant that Symbian phones had to become bigger and more powerful in order to compete with Android and iOS and so on. Traditonal Symbian was tiny and its apps were tiny. So, vicious circle: Nokia had to tempt developers to Symbian. This meant making it easier to port to, or write for, Symbian.
They chose Qt, over internal dissent. Result, more apps, but bigger less efficient apps. Meaning they needed to up the spec of the phones. Meaning that they threw away Symbian's core advantage - that it was tiny, used little power and could run quickly and responsively on low-spec phones. Which meant cheaper phones and kept Symbian competitive in the market.
The pre-Symbian smartphone OS, inasmuch as there was one, was GEOS - a Commodore-64 derived [yes, seriously] multitasking layer on top of MS-DOS [yes, really]. This meant a 386-based phone.
Symbian scored over that because it ran on the much smaller, much more efficient ARM chips.
Then once ARM chips caught up, Unix came to ARM chips - Apple's iOS is a Unix, Android is Linux which is a Unix, some of the early Chinese smartphone (and PDA) OSes were Linux-based too. So the ARM hardware suddenly expanded and got more RAM and more storage and more MHz, in order to run Unix and Unix GUIs well.
Result, easier development of pocket-platform apps, using existing desktop Unix tools.
Result of that, Symbian is now competing with a bigger, more capable OS with decades more development and a huge range of tools - many of them free, as were many of the OSes.
With hindsight, there's a sort of majestic inevitability to it all, but it was not really possible to see it coming.
For example, I had a Sharp Zaurus - the SL-5500, with its weird little slide-out keyboard. That would be about 2003, 2004 when I got it. (I had 2 actually.)
They were clever, but dreadful. Linux on a 200MHz CPU with 64MB of RAM, no swap, and just 16MB of permanent storage in the form of Flash - so, yes, less "hard disk" than RAM! - was not a good platform. It was slow, it was unresponsive, with so little memory it was unstable, and it *ate* batteries. Pocket sized device, laptop battery life: 2hr or so.
At that time, it was "obvious" that this wasn't competitive and that Symbian had compelling advantages.
By 5y after the first Linux Zaurus appeared, the first iPhone appeared, essentially running NeXTStep - i.e. FreeBSD on Mach - in your pocket, and it was smooth and fast and gorgeous and really stable.
By the time that was out, it was just a matter of counting the days for Symbian.
What Apple did, Android copied, but with FOSS tools. (Early Android was a Blackberry ripoff. That was hastily dumped and replaced with a finger/touchscreen UI that was clearly directly inspired by iOS.)
Once the pocket hardware caught up, there was no need for a weird, proprietary, lightweight, paid-for pocket OS when you could use a free one and free tools for a better development environment and a better user experience.
So Symbian had its time in the sun and it did OK.
I believe Symbian phones continued to (and maybe still do) sell well in the developing world, where people don't have landlines or indeed mains electricity and they are all jumping directly to cellphones and mobile-phone Internet. These markets are /extremely/ price-sensitive and thus the fact that a Symbian phone could run well with a quarter of the CPU speed, RAM and Flash storage of an Android phone meant that you could sell them at a profit for US$100 or less.
But the gap is narrowing. Look at the Raspberry Pi. Half a gig of RAM and three-quarters of a gigaHertz sold at a profit for $35. That's usable with Android and more so with lower-end pocket Linux OSes such as FirefoxOS (which in a way is Palm WebOS but done with more mature, all-Free tech.)