I'm a professional compositor and I can assure that shake is still the de facto no. 1, but it won't be for long. I'd expect half the jobs I go for to be nuke-based by the end of the year. In the US, there are lots of competing compositing platforms, in which Shake has longsince had the most marketshare, but here in London there tends to only ever be one, so when nuke takes over, it'll be a more pronounced shift. I'm about to start learning it now on fxphd (the best online place to learn vfx software, if you're that interested - not sure about how much it costs, but it's cheaper and better than the gnomon dvds).
Ron Brinkmann, the guy behind shake, left apple to go to the foundry and work on nuke, and their latest version (v5, out 3 months ago) looks a lot more like shake. All told tho, whilst shake is old and buggy, it's popularity is down to the fact that it's by far and away the easiest software to get your head round. Whether you're an amateur filmographer or vfx hopeful, shake's the best place to start. Then if you want to do all the more intense 2.5D stuff that nuke/fusion offer, you can segue fairly easily into one of those (I'm only paying for the course cos its quicker than bluffing my way through the transition).
My guess is "phenomenon", or whatever they call it, will be a part of fcs3, if it ever sees light at all, and it will be a snappier version of shake with fresher code, but little or no new features. Whilst often used on big features, the fcs family is really geared towards semi-professionals and tv. Apple obviously isn't interested in the low-volume high-margin $2000/licence sector of the market anymore, or else they'd have made more of an effort to maintain their position as market leader. I think Brinkmann intimated as much but I can't remember where I read the quotation.