I will tell you what I would like it to be. Maybe it gives you a Visual Studio-style editor for websites, which let you easily lay out a page, edit CSS, add event handlers to elements, link to other pages, and make and use data bindings (which could be configurable: persistent to DB, associated with user account, validated with this function, etc.)
Then you could use the data model created by the bindings, modify it if you want (data model specification is probably integral to everything), add and configure whatever extra stuff is supported (accounts, websockets, whatever).
A cross between Weebly and Visual Studio. It would be great for demo apps and internal utilities.
The issue is... you're stuck with Lightswitch once you start using it. There's not an "export to standalone C# project" button, I believe; you'll always be dependent on the Lightswitch libraries, which Microsoft will probably decide to drop or radically change in the future, at least before .Net itself is dropped.
And that's going to be sort of the problem with all these systems. Unless it can output pure, maintainable C# code (or whatever else), not dependent on significant libraries specific to the RAD toolkit, there's a very high risk that the toolkit and libraries become unsupported well before the application stops being used.
Truth is there are a ton of CRUD apps out there, that are 80% boilerplate in reality, it's just different boilerplate. I know HN tends to focus more on startups and bleeding edge hobbyists but enterprise CRUD is what drives software development, and creates most of our checks. And we can still do much better.
Whoever can come up with an elegant, fast, stable solution to crack out CRUD apps and not make all the mistakes of the past attempts can revolutionize things.
In my opinion the LAMP stack did this great for a time. Rails does a pretty good job now. The .Net framework also does this very well, at a cost. It's not sexy, hip or cutting edge but there's a big demand for it.