Waves and foam can be tough to do well, but there was a good paper on Gamasutra on it: http://www.gamasutra.com/gdce/2001/jensen/jensen_01.htm
(that'd be registration required).
The Unreal engine as a whole isn't incapable of doing fairly large outdoor areas - examples can be seen in U2-XMP. The unreal engine is pretty damn flexible and powerful, so this really isn't about them failing to focus on providing developers with the engine tools to produce great gameplay. Blame the individual developers for their own design choices.
There's also nothing wrong with taking advantage of pixel and vertex shaders in water; if you particularly think they're the things holding everyone back from producing the next and best wet stuff, I'd say you're mistaken. Take a look at some of the Cg (the NVidia PS/VS language) demos that people have done to see what sort of power Pixel and Vertex shaders have. The current technology that's the one up of using polygons is volumetric particles, but frankly, it's not worth it. Grab a high end 3d workstation and render good smoke with those and you're usually talking several minutes and upwards of render time, they're simply not very practical for the graphical payoff in realtime games. Smoke and gas can be fairly well faked using polygons at a fraction of the cost.
I'm still waiting to see something that makes good use of Unreal facial (?) and lipsyncing tech which they've mentioned in the past. That would certainly be one of the things that I'd like to see really well done by the engine in future. The faces in the Deus Ex:IW demo (still not out here) *really* sucked, I've seen better looking faces in much older games than that of Alex Denton in that, so it's certainly something I'm becoming more demanding about.