It was nice educating people about the Dynabook, but since that was one of the most seminal writings in computer science, I am not sure how you missed in in your career in IT.
I agree that the app oriented model is limiting, but Apple tried (with OpenDoc) to promote the object model of software distribution, and failed. The problem was not technical, as OpenDoc worked fine, but it was rejected by the marketplace. For software vendors, the problem with selling software components is that the business model is murky, because applications are well understood, but components are tiny, not well understood things. That is, you can sell people a spreadsheet or a word processor or a game, etc., but it is far from clear that you can sell them components to provide those functions in documents that are composed of bits of functionality from multiple vendors. So OpenDoc was never supported by any of the major software companies. And for users, it is confusing to get a document and not be able to read it because it contains pieces written using tools that you do not have, or with comparable components from competing vendors. And some things, such as games, do not fit the document-component model at all. So while the application model is limiting, it is a simple, well understood model.
If only Apple would approve Squeak (or Scratch) for the iPad. That would be, while unlikely, a wonderful combination.