In fact, all those interface changes have been as minimal as possible given the technological constraints and consumers have consistently rewarded manufacturers that simulated old controls with new tech whenever possible. Even radical changes like taking pictures from a screen and not a viewfinder have been smoothed over by providing both for years to give users time to adapt.
Interface changes are a cost not an opportunity. Radical interface changes are only seen in concept cars and quietly dropped before going into production (and when designers are too prideful to remove enough of the concepts from the concept cars in production models the result does not sell).
Consumers do not like interface changes in real life they like solid no-hassles and no-surprises execution. Gadgets with radical interface changes succeeded in spite of those changes not thanks to them (the iphone built on ipod familiarity, and the ipod tried to build on Apple computer device looks at a time no two mp3 players had the same buttons in the same place). Radical interface changes only work on TV commercials. Users do like bling and surprising looks, but only on non-functional pure decoration parts they don't have to interface with.
A fugly app like LibreOffice is getting slowly adopted because it gets the work done reliably (and is cheap). winamp and xmms had a terrible interface but this interface was stable and the software worked and that led to wild adoption at the time. GNOME 2 got happy users when it stopped trying to impose new UI paradigms and focused on fixing bugs. The most loved Windows release of all times was the NT version that got delayed, forcing developers to fix bugs for months because the scheduled feature and UI changes were already finished. Apple made a comeback thanks to Steve Jobs insistence on fixing every little thing (not because he had some magic vision, and in fact his vision changed several times, from color imacs to black-and-white ones, but because he made sure each time his people executed cleanly without cutting corners). All the server-y unix stuff Desktop people have denigrated for years have been increasing its market share in the past decade because it just worked and didn't eat your data. Working, not eating your data, and being predictable is much more a seller than half-finished software that tries to follow some abstract vision at the cost of execution.
GNOME people will get some praise and happy users the day they stop looking after the rainbow for a way to win market share instead on focusing on fixing bugs and adding features while changing the interface as little as possible. And it won't matter at that stage what interface design is in place. What makes a Rolls Royce is not the paint colour but the insane number of paint layers that ensures Rolls Royce owners do not have to bother with paint scratches ever. Software is no different.