Unification of UI makes sense where modalities are similar. But touch vs keyboard vs mouse are very different modalities. The first demands a few large areas with gestures; the second demands something very much like a terminal; the third, much more densely packed areas without gestures (because large mouse moves are far more imprecise).
I have Windows 7. But I never use the start menu's search functionality, because I have a terminal; if I want to start an app by typing, I'll do it from the terminal. That way, I get completion for application arguments, I can write scripts to shorten more complicated launch sequences, etc. That's because the terminal (+ shell + friends) is set up for typing; a search box is laughably primitive in comparison.
Meanwhile, my start menu doesn't look anything like yours. I'm running Classic Start Menu, with programs classified by function: Development, Work, Games, etc., just containing the main icon. I use these menu items when I'm being goal directed, but more importantly, it saves me the effort of having to remember any names at all. Menus are optimized for scanning lists of text; in this way, they're far superior to ribbons when looking for an unfamiliar command.
?
If I want to start Google Chrome, I can open the start menu, type "go" or "ch", wait half a second to see that it's the first result, and hit return. To do this from the terminal I need to switch to a terminal, know the name of the executable (not just the display name of the program), know what directory it's in (or make sure to put it in my PATH), etc.
I don't agree with your characterizations. The mouse allows density but doesn't demand it - both more and less dense layouts have pros and cons for mouse and are appropriate in different contexts. A terminal is good for input but limiting for output - I would like to see a command shell that took a text command language, but produced graphical output.
The only time I want to launch Chrome (it's not my main browser) is if I want to visit Facebook (which is blocked every which way in my main browser).
Now, if I wanted to make visiting Facebook quick, I'd create a shell script called 'fb' which would start Chrome with facebook.com as the argument. But I don't, so I haven't.
Switching to a terminal? I'm always in a terminal. I have 4 open on my desktop right now, not including the virtual terminals hosted in screen sessions. I'm pretty much never more than an Alt-Tab away from Cygwin in mintty.
Less dense layouts don't work well with the mouse, in my experience - I'm talking about screen-sized layouts, the kind you see on phones and tablets, only scaled up to multiple 30" monitors[1]. Big mouse moves can cause you to lose track of the mouse altogether for a second.
Graphical output? I'm not really sure what you're getting at. What kind of graphical output would you want, apart from starting graphical applications based on command-line arguments or piped output? I've written many of both. Most of the time all you want is to view a graph, picture or video. All are trivial to integrate to the command line with appropriate glue.
[1] This is actually a bigger concern of mine over Windows 8. I'm normally multitasking across 6+ apps at all times: my custom notetaking app, vnc, several terminals, browser, documentation and two email clients. It's why I have multiple big monitors. Only being able to multitask with two applications at a time sounds like a bad joke.
I agree 100% that the idea of Metro style apps with their severe limitations (on multitasking and otherwise) replacing desktop apps is a bad joke. However, I don't really interpret them as replacements, but just as another model with its own advantages ( isolation, suspension/controlled backgrounding, guaranteed declarative install) and disadvantages (multitasking limitations, walled garden, etc.) I think it's very stupid that metro style apps are set as the default for file types and on the start screen etc. However, as a supplement to the desktop I like having them available.
Unification of UI makes sense where modalities are similar. But touch vs keyboard vs mouse are very different modalities. The first demands a few large areas with gestures; the second demands something very much like a terminal; the third, much more densely packed areas without gestures (because large mouse moves are far more imprecise).
I have Windows 7. But I never use the start menu's search functionality, because I have a terminal; if I want to start an app by typing, I'll do it from the terminal. That way, I get completion for application arguments, I can write scripts to shorten more complicated launch sequences, etc. That's because the terminal (+ shell + friends) is set up for typing; a search box is laughably primitive in comparison.
Meanwhile, my start menu doesn't look anything like yours. I'm running Classic Start Menu, with programs classified by function: Development, Work, Games, etc., just containing the main icon. I use these menu items when I'm being goal directed, but more importantly, it saves me the effort of having to remember any names at all. Menus are optimized for scanning lists of text; in this way, they're far superior to ribbons when looking for an unfamiliar command.