You missed the point--he's saying that root access might one day no longer be necessary, ...
Actually, people have been claiming since the early days of unix (back in the 1970s) that root never has been "necessary". I've read a number of discussions triggered by such claims. They all reduce to the same conclusion: Yes, in a well-run computing environment, in which all vendors and users understood all the security issues and agreed on their solutions -- and implemented them all correctly -- the root id wouldn't be necessary. But we never have been anywhere near close to such an ideal. And until then, root is needed to cleanly fix the permission messes that our current practices so often produce.
I've been, uh, "discussing" an example of this on a web server where I'm the maintainer of one of the web sites. The site is actually replicated on my home machine and on another remote machine. I make changes on my home machine, then rsync the three machines when a change is working to my satisfaction. On the two remote machines, rsync has always produced a lot of bogus permission errors (while correctly copying the files). The reason is that some of the files are created by the web server, and are thus owned by the web server's id, not by mine. The code can enable world read/write permissions for everything, so the rsyncs all work. But due to the mismatch in ownership, rsync complains that it can't fix the permissions.
This is a problem for one important reason: Whenever the software gives such floods of bogus error messages, they bury the actual error messages, and teaches the users to ignore error messages (since all of them that you see are so bogus ;-). This isn't an ideal situation, if you want people to correctly spot problems and fix them.
It turns out that I can "fix" many of these problems if I spot them early enough. None of my login ids can fix them, since the logins don't match between the 3 machines and I don't have admin access to the others. But on my own machine, I can often use "sudo" to adjust permissions so that rsync won't produce so many bogus error messages. But I haven't stumbled across a way to fix them all.
I have occasionally persuaded (nicely ;-) an admin on one of the other machines to use a similar sudo to give me control of my own files, but they usually consider this a bother, and don't do it. I need to stay on good working terms with them, so I don't push it.
Anyway, I'd agree that root isn't, stricktly speaking, "necessary" right now. But it's often the least time-wasting solutions to all the annoying permission problems that typical machine setups produce. On a well-done server machine, owners of a web site would have group "www" permission, and could fix most such problems, but I don't think I've ever worked on a server that's run that way (except the servers that I run myself ;-). And so on.
I have an Android phone that at random times gives me what look like permission errors. I've investigated, but so far haven't found a solution other than rooting the gadget. I haven't actually done that, so once again, the machine's "security" setup is teaching me to ignore error messages, since they're usually sinkholes of time that I can't do anything about.
(I've never kept anything important on my phone for more than an hour or so, and treat it all as "transient" stuff that can disappear any second. I've occasionally worked on some apps, but tend to minimize testing on the phone itself due to the confusion of all the permission problems. Maybe this'll change some day. Or maybe I'll just stick with developing "apps" that run inside the browser, and continue with the mess that that "OS" is. ;-)