Potentially unpopular opinion: There is no money to be made in consumer desktop operating systems in 2018, so there is little funding for innovation in the consumer desktop operating system space. Until a business model emerges which challenges this, the status quo will remain roughly the status quo.
The two major commercial producers of proprietary operating systems are basically giving away their operating system at this point. Free Linux distributions offer basically the same functionality for most users; the only things preventing migration from one OS to another are existing applications. For non-gamers, that's often moot, because the world exists in a web browser now. Yes, there are other applications causing OS lock-in too, but they're mostly on the business side, not the consumer side.
There is still plenty of money to be made from enterprise operating systems, but the innovation there is in a totally different direction than what consumers would find valuable.
Example: Linux containers are great! But who's spending the R&D money to make them a viable way to distribute games? There's plenty of work being done on containers and graphics chipsets for ML applications, but not for consumer use.
The two major commercial producers of proprietary operating systems are basically giving away their operating system at this point. Free Linux distributions offer basically the same functionality for most users; the only things preventing migration from one OS to another are existing applications. For non-gamers, that's often moot, because the world exists in a web browser now. Yes, there are other applications causing OS lock-in too, but they're mostly on the business side, not the consumer side.
There is still plenty of money to be made from enterprise operating systems, but the innovation there is in a totally different direction than what consumers would find valuable.
Example: Linux containers are great! But who's spending the R&D money to make them a viable way to distribute games? There's plenty of work being done on containers and graphics chipsets for ML applications, but not for consumer use.