Manovich does say that clicking the hypertext links is not an interactive process, and to some extent I agree, but not completely. It also depends on what type of hypertext we mean. If we mean a small website, then clicking links on it as a hypertext isn't interactive (at least not any more interactive than any reading process) because the text doesn't fundamentally change by being clicked just chooses the order and areas to be read, which is like the interactivity of skipping around in books (like Barthes' tmesis, I think). But, some of the storyspace hypertexts do change the links available based on the areas read and the path the reader comes through to get to the other areas. In these cases, the story is one way at the outset and different options programmatically open and close based on how the reader reads. In these (much rarer) cases, the text does respond to the reader, albeit in a limited way.
I think these do constitute some level of interactivity and I think this gets to part of the issue with the term 'interactivity'. If we define it too strictly or too openly, it's pretty useless. The open and closed interactivity aren't established terms in video game, or new media studies, as far as I know, but these could help eliminate this problem. The closed system could be one where the majority of the interactivity is anticipated by the designers (as with hypertexts made with storyspace and small websites) and the open systems could be emergent interactive systems, where the designers put in place the main structure and then the play is interactive (novels could also fit into both of these based on their structure and so this wouldn't conflict with reader response theories). Is this what you mean?