The real-world metaphor that app-driven, walled garden environments were designed to mimic is Television.
It was a way of bringing Television to the computing environments so that you could create interactive variations on television. In order to do this, they went with a secure run-time operating system in order to completely lockdown 'the network' from the rest of the system, and vice versa. In this way, a company could broadcast their content in an interactive manner using actual source code and programming, while the user would have no fear of viruses or worms traditionally associated with the transfer of code.
Microsoft took this and ended up creating the Modern or Metro styling of making everything full-screen and allowing you to quickly 'change channels'. If you're looking to broadcast an interactive app as the new television, you want it to have a world of its own. Thats why they insist upon it taking up the full allotted area, no matter its size. Likewise, its no error that their 'multitasking' involves allowing the user to configure their screen into multiple 'televisions' that look exactly like something you'd see at a Vegas casino security room. Simply set the televisions side by side and you can view multiple 'channels' you're interested in.
This is also why Microsofts plan for replacing the desktop with a television-based app environment simply won't work. The desktop is largely a content creation mechanism, and in that paradigm you have multiple streams of data merging into and the user uses to create something new. Even constructing a document, you may need information from the web, from email, or as data being sent to you from various sources in a dynamic fashion. In order to manage these streams, the television paradigm of an app, simply falls apart because its designed to have content flow to the user, not from other apps into the app.
Something as rudimentary as Netflix would do well to take note, and to reimagine what they're doing and allow their app to actually contain multiple forms of data streamed from the servers that are unrealistic in a web-based environment, but fairly easy in an interactive television-based app environment. Commentary tracks, extras, and pretty much anything you can find on a Blu-Ray. There is no reason all that stuff can't run out of their interactive app.
Now this isn't to say that the two paradigms can't coexist, and there isn't any reason they shouldn't co-exist. But the desktop's strength is content creation, while the television-based app environment is based around content consumption. Two halves of a coin, and if you keep them separate the whole makes a lot more sense. This is the one thing Microsoft doesn't seem to understand, and until they stop trying to put a round peg into a square hole(content creation into a television-based consumption environment), they will fail miserably.