Actually, it's a learnability problem: once you learn how to use its features, iOS 7 is a vast improvement over 6.
If you mean people were complaining that the controls were too visually complex and they attempted to differentiate themselves too much, I guess personally I'd have to disagree with those people.
But that can be said about anything which involves "knowledge in the head" vs "knowledge in the world": Once you learn how to use it, it's superior to simpler systems restricted to only what can be operated by knowledge in the world alone.
And so far everyone has argued that iPhones are "better" by virtue of being "simpler" and "more intuitive" than Android because you don't need to figure out how things works, i.e. it only depends on knowledge in the world.
And now with iOS7 the tune is that you just need "knowledge in the head" and all is good. That's not a very consistent story.
They're not, and their relative ease of use is only because users have learned over years that a few words looking markedly different (but not emphasized) might just be links.
A web developer can override both those conventions and when they do, their links become less recognizable and their apps become less usable. Since there's no cursor on iOS touch screens, losing the visual cues is a similar problem.
[edit] You might argue that just tapping on something serves as the way to discover if a visual element is tappable on iOS. But it doesn't, as the original article states, it just makes the user wonder if something might be broken.
Also in this case the problem is if a site has some text that's blue & clickable, and some text that's blue & not clickable. That's more or less what iOS 7 is doing. Colored text doesn't mean it performs an action. It might, it might not. Similarly, black text might be clickable, it might not. It might have a border, it might not. And these are all mixed together in the same app, or even a single screen such as the case of the alarm that the article is talking about.