I personally think we ought to reconsider our whole desktop paradigms from scratch. Some of the ideas we're using were designed they way they are primarily beca

I personally think we ought to reconsider our whole desktop paradigms from scratch. Some of the ideas we're using were designed they way they are primarily because of display size constraints. For example, popup/tree navigation based application menus would probably seem silly when we could just dedicate a fraction of one side of your screen to always show the whole tree. I don't think we're anywhere close to the limit where something like Eagle Mode won't still benefit by adding more pixels (provided you can also supply the memory and processing capabilities for it to work responsively). The same kind of thing could apply to many tree structures, or graphs of complex relationships - being able to visualize large collections at once without the context-switch of scrolling is useful. Being able to view the whole source code of an application at once would be incredibly useful. Another example might be reading or writing some research paper. Frequently you end up scrolling back and forth through a paper to cite or connect the bits together, which is pretty cumbersome, particularly when "back" buttons don't always have the expected or desired behavior (e.g, many viewers are missing a "push my current scroll position onto a stack before navigating" feature). Being able to view a whole document at once would ease some of that. It'd be interesting to see some studies on the difference between the efficiency of using spatial layouts compared with sequential actions. It seems to me that spatial layouts quickly become rote actions that don't require much thought, but having to execute a set of commands is repetitive and error prone. (e.g, how many times have you clicked the wrong application in a task bar accidentally when you have a brain fart?). It seems the industry is obsessed with heading in the opposite direction. I find mobile platforms a pain to use and I generally avoid them. Things like Gnome Shell's task switcher are also a regression from task bars, as the sequence to switch an application now has extra steps.