There's a lot of interesting commentary by David Wood on Symbian OS vs Linux here. David is one of the original inventors of the Psion and Symbian operating systems, and I've met him on numerous occasions. I agree with quite a lot of what he says - particularly around the idea that creating Linux phones is neither as easy, nor as cheap, as many propose.
He's also right to debunk the idea that most Linux phones are actually smartphones, just because they are built around a Linux kernel. My own forecasts predict Linux featurephones to outsell them several times over. (The difference is that a smartphone generally allows developers to write 3rd party software for the platform). That said, if you take a stricter definition of "smartness" - ie the user can buy & install 3rd party software, rather than it all being preloaded & supplied by carrier or retailer - then quite a lot of the Symbian devices in the market are too locked-down to be considered fully-open.
His arguments against Microsoft and Windows Mobile are a bit more oblique, and, to my mind, he skirts an awkward issue for Symbian:
Increasingly, higher-end mobile devices are not going to live as standalone devices. They will have be "good citizens" of an extended ecology spanning PCs, notebooks, consumer electronics, IP-PBXs, WiFi phones and IPTVs. Various emerging services and applications will span several categories of products. What this means is that while a handset-optimised OS is ideal for a handset-optimised service, it is much less clear that Symbian has a strong role to play in applications which are "federated" - perhaps across a handset, a PC and a games console, for example. It may well be the case that Windows' (or maybe Linux') ultimate deficiencies in terms of handset performance may be offset by superiorities in interoperability and cross-platform software portability.