There is nothing magic with the hyperlink except that it intensifies the magic of the ordinary paper reference pointer.
Not only the hyperlink, but every link, and not only links that are actually followed, but already the encounter of a link may have great effects: The reader may have associations of their own stimulated, or their own thinking may be intensified.
The link anchor text represents thoughts that are presented elsewhere, as if the link text were a deputy, a proxy, a place holder, a door handle for these remote thoughts. The anchor represents the choice to turnoff from one’s reading flow, and facilitates one’s thoughts to branch, towards an association, a recall. If there are several branching possibilities, the starting point contributes to increase orientation.
All these effects of non-linearity are, of course, more speculative, less tangible or observable than the effects of a followed link that yields a concrete click sequence (which is linear again, BTW). But the effects are present, with both paper and electronic links.
Now then, the electronic version of this “potential branching of thought”, the hyperlink, is often a bit superior to its paper counterpart. Mostly just bagatelles are involved, gradual differences, a small increase of a tendency, or enlarged attractivity of leveraging affordances that were present before, and in part, they are merely about optics.
1. An excursus in fine print does not appear equally clearly as a skippable alternate branching option as a hyperlink does. On the other hand, the link appears still more integrated into the main text than a reference on paper, because it is easier to follow, and this impression has its effect also if the branch is actually not taken.
2. This is not only true for the integration of the click text into the link source context, but also for the link’s target context. In principal, also a paper reference could sufficiently provide the guiding “sign-posting” towards the target context, if clear title information would signal the quoted section as well as its context already at the jump-departure spot, for instance using “see x under y”. But the immediateness of the target text that is seemingly improved by the hyperlink, makes its context appear closer even if the click text is annotated more sloppily saying, for instance, just “see x” or “see under y”.
3. Footnotes are the only acknowledgeable predecessors of hypertext and they most closely measure up to hyperlinks in terms of the optical impression of potential branching and easy following. Also here, an optical bagatelle makes a difference: if they are put at the end of a word or of a sentence they do not explicitely enough mark the phrase that is supposed to serve as deputy for the branched thoughts, i. e., that would become the click text of a hyperlink. Even if a hyperlink is not at all typed or annotated and only consists of a single clickable word, at least this word is clearly designated as the deputy for an association, while a footnote behind this word could as well apply to a phrase or, if it is located at the end of the sentence, apply to the whole sentence and thus to the predicate that may be far away from the end of the sentence. At least the optical confusion may arise even if the footnote is correctly put before the period, in this case.
4. To experienced hypertext readers, a blue underlined spot has a stimulating effect even by its mere existence, nowadays.