This and the following tweet
fairly encapsulates my perspective on it.
Currently the worst example of a mouse-based interface I use is Netflix's "closing credits" screen. I've learnt the interface, I know that I have n
seconds before it'll auto-play the next episode (or even a trailer for another programme!), and that to prevent this and go back to "Browsing" I must click on a button in the top-left corner of the screen.
To do this, though, I must go through this whole ceremony of mode-switching from watching the video to locating the mouse cursor on screen (admittedly an improvable skill, potentially eased by using a more prominent cursor image), moving the physical mouse until the on-screen cursor is over the image of a button (also improvable, as an old work colleague proved), then clicking on the physical mouse button.
This ceremony (besides the mode-switch) could be bypassed by just binding, say, Backspace to "Go to Browse". Then we'd only need to locate and press the Backspace key on the physical keyboard. It could also be bypassed by using a touch screen, hand gesture (do we have the tech for this yet?), voice command, blink of the eye. Basically any means of communication that doesn't involve a mouse or depend on a prolonged feedback loop of moving something and verifying the discrete position of something else.
The issue with the Netflix interface isn't in the learning of it (especially once you've learnt how to use a mouse and read a GUI), it's in the execution and the fact that mastery of the interface enables barely more effective use of it than a cursory understanding.
The way I'm approaching Cinera's interface is to have both mouse and keyboard controls, with the keyboard offering extra functionality (mostly surrounding filtering) that regular users may want, without littering the screens of occasional users. You can see that kind of thing in action here
Suppose you're a regular who only wants to see which programs were run during the stream. You can do this by filtering off all the media except "In-Game" and switching the filter mode to "exclusive". A beginner may likely do this all with the mouse. A keen keyboard-focused intermediate user may perform the mouse controls one-to-one with their keyboard equivalents. And a master may use the keyboard's extra stuff by filtering off
the "In-Game" medium and inverting the media filter (v).
Hopefully this doesn't come across as a Keyboard Master Race kind of pitch, stunting casuals and rewarding masters. It's motivated by the appreciation that there are various styles of user, and the desire / need to understand, respect and tailor the interface to those styles.