Aluminum cases are like designer jeans. Don't work any better than Levis (case temp the same given same airflow. Find an identical design in aluminum and steel, put the same parts in it, same conditions then measure and I bet you wouldn't even find that 1*C), not as durable, but there's a certain status thing to it.
They are a bit lighter which is the upside to the lan box idea, but they also are easier to damage so in that respect I'd stick with a steel case if I were moving a lot, unless I didn't care about looks, but then most people with aluminum cases do care about looks, and this sentence got to be rather long.
I like the IDEA of using the entire case as a heatsink. But some direct contact is in order, simply trying to exchange heat through a smooth aluminum surface with indirect airflow inside and little or none outside doesn't accomplish it. My old Realistic (yes, Radio Shack) cassette component used the chassis for the heatsink, transistors were mounted directly on it. Rather clever I thought, no little sinks stuck haphazardly inside, but rather the case and guts were designed to work together. PC's aren't, we have to use off the shelf parts and the designer of one never met the designer of another, muchless worked with her/him. Back in the Pentium days I was hoping for a new spec which flipped the CPU socket onto the other side of the board where the CPU would then contact the side (or bottom in a desktop design) of the case and use the entire panel as the heatsink. And I wonder if one could use the case to dissappate heat in a watercooling setup to reduce fan speed on the radiator, maybe even eliminating need for a fan except after periods of load, but that'd require big long waterblocks mounted to the inside of each panel, huge expense for little gain. So yeah, I like the idea, but it has to be implemented a lot more efficiently to do (any?) enough good to justify the price.
--Mc