What I was trying to convey was that an idea from the days of centralised processing seemed to be relevant again, that the heavy lifting might be done on one machine, which communicated with another to provide an interface that is perhaps unique to one individual. There are all sorts of possibilities for further simplification if we could start from a clean sheet, but for now let's think about is practicable with a current OS such as Windows running on the main machine.
My apologies also. I probably also over-thought it while under-thinking it. (In the hopes of trying to grasp the concept better)
I guess it just confused me because it does seem like such ideas as "all server apps" were re-directing the course in other uncharted aspects that go against the already "Brave New World" concepts of cloud computing and complete online OS's. (Well... online because 40hz was alluding to a virtualized OS which is kind of an opposite but similar ideal to just virtualized server apps and virtualized OS's)
For a concrete example of the keyboard/mouse replacement, think of a reduced set of keys with no labels, and a pointing device such that you can easily key any combination while using it. Current key assignments are displayed either on a controller display, of in a window on the main display. Any key may serve as a shift key, displaying a new set of assignments, so for any task you have the correct keys available, where you want them to be.
The problem I see is that for this to work with existing software is that the pointer needs to be a 'real' mouse pointer, so when you use it to select something different, how do the key assignments know to change?
Well, I don't really know about your idea and I don't know anything about the backbones behind making technologies work but why can't it be a simple plate like what KeyStrokeCE does on the PPC?
What I mean by this is that similar to visual keyboards or switching to a new window with a different program or even pressing shift then mouse click mouse click to select a subset of icons in Windows... wouldn't all you need to do is to have a software that ignores the mouse click completely but simulates it anyways?
For example, on a simple scale, in order to switch typing tools on a PPC --- you don't try to fit the model onto another program --- you separate it and create a software that ignores but mimics the model.
In the case of a PPC, you don't uninstall or worry about two typing software colliding because a well-made software would merely add an entry under the input model of your PPC which would then lead you to choose between which input panel you prefer via a drop down arrow on the lower right hand side.
In the case of Desktops, it's more complicated but something like a Launcher software skips between the hotkey and mouse pointer model because it didn't try to simulate the previous model but created a way where you can go so far to a point until you can ignore the previous model. In that case, you can ignore the keyboard once you've opened the application and therefore it makes it less scarier because you don't have to memorize a million hotkeys or drop the mouse for keyboard typing but also allows more advanced users to set specific commands to the program that a mouse wouldn't be able to do.
From a desktop KeyStrokeCE model, I was assuming you can have the software toggle and make the Numpad control the mouse keys thus skipping the mouse pointer dilemma. Obviously, I'm speaking in lay man's words since I don't know how to develop the actual software nor even know what complications that make that concept problematic.
Also, of course there are things which numpads can't do even if they control the mouse pointer but that is where I envisioned the software to create alternatives to make it work within that framework. (Much similar to how a tablet pc software changes the dynamic of the pen by allowing for hotkey commands to be changed but in a much more different way where the program should select stuff via a software mechanism that makes the numpad more like a circular clock selecting/switching app that has a sub-dial like switch much similar to a software magnifying glass which makes one see and control what selections to omit, fine tune and combine to mimic commonly used goals like switching windows + switching commands.)
Yours is a different model though and one that I don't understand (but seems like I should understand) which is why I didn't try to overwhelm you with questions and opted for the one you quoted in the hopes that it will eventually make sense to me as this thread moves along.
I'm still answering the question though because it seems like a legitimate question and not a rhetorical one and because I do find it weird that you can envision this "complex" concept but find it troublesome to envision a mouse pointer alternative and just settling for the concept of a real mouse pointer. Hopefully even if my answer horribly misses the point, it will lead to a better understanding of why you opted for this question.