I am not a user of KDE, so I am not sure of what these activity are usable in practise.

There was a talk at KDE Akademy 2013 (I don't like KDE, but the people being it are great !) about task centric UI. I didn't found the video of it, but found the first part of the slides : http://www.slideshare.net/rcolomar/talk-task-centric. If anyone knows where to find the videos, please let me know.

What I am proposing is related to the principles exposed there, except that the contexte is different.
In KDE talk, they want to produce content, and are proposing how to have predefined workflows to reach it.
In my mind, they workflow would be constructed on the fly, and the flow would be more a matter of automatically organising things in a multitasking environnement (so that the multitasking view would shows tasks, not applications).
This would solve the problems that an app can be used in several simultaneous tasks, but the current UIs are not designed for this, and thus require a lot of going back to switch between things.

You can see a start of this integration in several apps, but it is not generalised. Best example would be camera : You open camera, take a photo, then switch to the gallery, select a photo, send it by mms, write the mms, and send it without leaving the application.
This would normally take 3 apps if it was not integrated, but there you never exit the current one. It also only takes one slot on the multitasking view for this whole chain of event to do this single task. Very Good !

However, in the middle of writing the MMS, you kid/cat/.... does something cute and you want to take another photo. It is not possible !
We have a phone where multitasking is the selling point, but where you cannot take a picture while sending an MMS...
There you should be able to start a new task (a new slot on the multitasking view), which would consist of taking a photo, and then other things that you may decide on the fly, and then getting back to your first MMS.

It is a lot more important on a mobile device than on the desktop, as the tasks done are usually smaller on a phone (not working with an IDE, file broswer and 10 stackoverflow tabs, but simple messaging of data searching), things are mostly full screen compared to multiple windows/screen/virtual desktops on a PC, and finally input system is a big limitation (no mouse and keyboard makes everything slower).


Initial question was to disrupt mobile, not desktop. Putting desktop paradigms in mobile is for me an error, as they are not fulfilling the same needs (even my laptop is not comfortable to do heavy work compared to when it is docked, I don't image doing anything heavy on a phone).