Great points Mike.  Although it's hard to find small applications anymore, as you mentioned tongue in cheek, I do think that expected application size should have an immense bearing on the architectural decisions of an application.  For example, I did a couple of small eCommerce stores a couple years back and the use of Active Record with little separation of concerns was a terrific way to get the application up and running very quickly.  With a more recent project, which turned into an enormous CRUD application which tons of reporting needs, using S#arp Q3 "out of the box" wasn't enough to keep the controllers layer easily maintainable and the reports performant.  In hind sight, if we had leveraged application services more appropriately, keeping the controllers very clean of workflow logic, and employed CQS more for reporting needs, the application would have been simpler to maintain with much better performance.

Changing gears slightly, it's funny how often people put down most new and emerging ideas as being faddish or those-which-came-from-developers-with-too-much-time-on-their-hands.  When it comes down to it, new ideas are (hopefully) generated in response to a need or challenge.  CQS is simply naming a good idea which has been used frequently in the past, allowing it now be more effectively and concisely communicated to others who may be facing similar problems.  With that said, there are certainly ideas which do not solve problems as well as others; it's our job as a community to discuss these ideas, exactly as we're doing here, to allow natural selection to take due course to have the best ideas survive.  (Obviously, a corporation like Microsoft with millions of dollars to back specific ideas can usurp natural selection of ideas generated via group-think, but they still come up with something pretty good now and then. ;)  The trick is figuring out which ideas are worth keeping around without experiencing too much pain in the process. ;)

Along these lines, there's a great book about the evolution of ideas called "The Meme Machine" by Susan Blackmore.  After reading it, it makes the blogosphere much more interested to observe.