First of all, you're wrong about insignificant numbers buying Symbians for the smart features, the sales of Nokia Communicators alone are the same as Palm or PPC/MSS. No one would buy a Communicator and not want to use it as a PDA, it has a full keyboard for heaven's sake! Add in the minority of smartphone enthusiasts from all the other varieties of Symbian (S60, UIQ, MOAP) and it's clearly the biggest mobile computing platform even if you only count people who actually use their device for its PDA functions and are aware of what OS it's running.
Second, you're missing the point if you're saying "what percentage bought a smartphone as a smartphone?", for several main reasons:
1. Percentages mean less than numbers. Even if 100% of Palm users bought their device as a PDA, that isn't as big a number as those that bought a Symbian as a PDA.
The manufacturers (of both the device and the operating system) make almost all their profits from the sale of the device. Once the sale has happened, they've succeeded. The sale then provides the money to cover the cost of making new models and developing a better OS, so there'll be an even better iteration soon with even better hardware and software features. If PDA sales (or even market share) are going down, it doesn't matter if 100% of their customers bought it for "smart" functions, they'll simply have less money than their smartphone rivals to create better devices and a better OS.
The bigger selling OS will (probably) get better and more numerous models than the lower selling OS, so both the OS enthusiasts and those totally ignorant about OSes will get something out of those bigger sales. The enthusiasts would be riding the coat-tails of the "ignorant masses", which has been the case in computing for a long time now.
Many, if not most, PCs now seem to be bought largely as a way of accessing the internet, and bundled pre-installed software like word processors might get used too, but a lot of people are totally uninterested in installing anything themselves. My parents for example had a broken CD drive on their last PC but they never noticed, it was only when I visited them once and tried to install something myself that I realised it didn't work. Had I not tried, they might well have had the PC for years and never known about the broken drive.
Does that mean the PC format isn't a good one for software developers? Of course not, it just means that those who make full use of their PCs are being subsidised by those who don't. The enthusiasts are getting better and better machines partly because non-enthusiasts are buying them too.
2. Just because someone buys something without realising its "smart" doesn't mean they aren't potential customers for "smart" software. There's already a mass-market for Java games and apps, and I've seen an increasing number of mainstream smartphone titles sold in exactly the same way (UltraMP3 for Symbian was being advertised by Jamster/Jamba in TV commercials on MTV, for example). It's a proven business model for Java, why wouldn't it work for the even more impressive games and apps you get on smartphone OSes?
Note that none of these Java-style downloads would be via complicated sites like Handango.com where you're required to know the operating system (including its version) that your phone uses. None of these Java-style downloads require you to transfer a program to a PC, then to a memory card or wireless connection, then to your phone.
This has been a big barrier for a lot of smartphone users, that you have to mess about with a computer (and many of them don't have a computer believe it or not, or don't have a card reader or wireless adaptor) rather than just entering a simple phone number or web address into the phone itself.
I know this because when Nokia gave away the Snakes game for N-Gage, the instructions only covered transferring it by memory card or bluetooth. Loads of people asked about how they could transfer it straight to their phones like Java games, they couldn't transfer it by computer (some of them didn't have a computer!). In the end I set up an unofficial site myself, and it got hundreds of downloads despite a big warning saying they'd have to pay GPRS charges. They mostly didn't care, the convenience was worth more than the few dollars the download would cost.
As GPRS charges go down, internet speeds go up with 3G, smartphones become wi-fi capable, and smartphones make up an increasing chunk of the general phone market, Java developers can transfer seamlessly to the smartphone market and publish Symbian/Windows/Palm games instead.
Instead of saying "do you have a Symbian/Windows/Palm" they just list compatible models of phones and the number you text or site you have to go to to get them, no one has to even know they have a smartphone for this to be a viable business model. They might even carry on calling them "Java" games and apps even though they're not, because that's just the common word for a program you download onto your phone.
4. Smart features don't have to be advertised as smart in order for them to be appreciated by the customers and the manufacturers. Smartphones let you have 3D games, let you have high quality photo and video editors, or high quality MP3 player software. Smartphones let operators or even individual retailers pre-install features to add value to a device, in exactly the same way PC shops do. If there's a particular niche market, they're very easy to cater for simply by installing the right software, and if the market changes you just change the software.
If you say "this phone lets you do word processing and email" or "this phone works like an ipod" or "this phone works with instant messages" or "this phone lets you shoot video, edit it and publish it on the internet" or "this phone lets you do all these things at once", at least one of those things would probably interest someone who wouldn't have been interested in mobile computing in general.
-"Would you like a pocket computer?"
-"No."
-"Would you like an ipod phone that lets you buy music straight onto it?"
-"Yeah!"
The two things are essentially the same device, yet their appeal is totally different. One is aimed at technophiles, the other at music lovers, and music lovers are a far far larger group.
Customers don't have to know it's a smartphone, all they have to be told is how the features would make their life easier or more fun.
How many people would buy a PC if it didn't come with a pre-installed word processor and browser?