About the revolutionary aspect of it: primary memory instead of files. Suppose we overcome the problems mentioned in the article, then it still doesn't do anyth

About the revolutionary aspect of it: primary memory instead of files. Suppose we overcome the problems mentioned in the article, then it still doesn't do anything for current software for normal tasks. E.g., how do you back up primary memory? The only way would be for the application that owns the memory to do it. How do you share files between two processes? By adding memory sharing into your application. So, many of the tasks that are normally taken care of by the OS would need to move into your application. That's not feasible. So it needs a rather different OS and run-time environment, which don't exist, plus a huge rewrite of your apps, along the services it needs, before it can be even considered. But not only that: it doesn't solve a problem big enough to make it commercially viable. It might be nice for something like Amazon's Redshift, but not for our mundane web servers and accounting programs, nor for the apps on our laptops and telephones. It's a solution in search for a problem.