The results in the study linked in the updates to that NNGroup article on URLs are limited to users viewing links on a web search results page with the intent to click one of the links to visit that webpage. Clearly, investigating suggested webpages could and does involve inspecting the URLs of said pages for many users. The study linked, however, provides no guidance whatsoever on the eye fixation time of users on the URLs on intermediate stage webpages in a webapp or other web-based workflow with which they are familiar. That is, we shouldn't be generalizing an eye-tracking study that only looked at search result links to eye fixations on the URL bar on general webpages.
The web-browsing public is also more accustomed to unreadable strings of nonsense as URLs now than they have been at any point in the history of the internet. Just have a look at any link from a CDN, such as facebook's ubiquitous fbcdn or at any Google search result URL. Even your bank's website, a trust-critical application, uses URLs consisting of a bunch of flags and ?= options that aren't particularly human-interpretable, even if some of then are dictionary words. Not that this is "good design", per se, but that the effects of bad URLs on the browsing public may not be as large as some fear.
URL clarity is still important, especially if you expect people to share or bookmark your pages. But in the context of an app's internal pages, URLs aren't as important. As the article Jurjis' linked indicates, URLs can help with You-Are-Here, but on the other hand, do you really want people trying to URL-navigate themselves to the middle of some app workflow you didn't design to be started in the middle?
Of course, certain kinds of URL clarity are important to another class of user: developers. As this article points out nicely, you want your URLs to make sense, at least at the routing level, to the devs, if no one else. But in the context of a security-sensitive application, you also don't want your URLs to be so specific that they leak information by existing.