I never bothered to research the history involved here, but for some reason, English has a braindead rule that says that "this is incorrect English". Rather, you must do "this." Notice how the punctuation mark goes inside the quoted space, even if it's not logically part of the quote. That sounds harmless enough, right?
Now, let's learn a bit of HTML. If I want to create a hyperlink pointing to Google, "<a href='http://www.google.com/'>this</a>" is the standard way of doing things.
Now, let's try putting our knowledge together to create a hyperlink like "this". Wait, the period needs to go inside the quoted space (English 101), like "this." Ah, man, the period isn't part of the hyperlink (HTML 101). We can't take the period outside of the hyperlinked space, though, without taking it outside of the quoted space. What can we do? The obvious answer is that we can't achieve what we want without violating at least one of (a) English; or (b) HTML. The question still stands, then: what _can_ we do?
Well, we can take advantage of parentheses to "push" the punctuation mark away from the edge of the quoted space, so that English no longer requires the quotes to conquer our punctuation mark. Using that technique, we end up with things like "this" (this comment intentionally left blank). Don't you just love "linguistic evolution" (this comment intentionally left blank)?
Postscript: I developed a language called MIL (My Interesting Language) more than a decade ago with the eventual goal of replacing English completely. (It's a precise language, with a logical structure. In particular, it's compatible with HTML.) Unfortunately, MIL was effectively limited to written forms, with no useful verbal mapping. Fortunately (I guess?) MIL is still MIA from the dave2 catastrophe, and so I've taken to rethinking it (like most of my lost dave2 stuff) from scratch. I revived some of my earlier experiments in this arena, HIL (Human Interaction Language) and FOL (Fact Oriented Language), and I'm working on a new dictionary. The beauty of the FOL/HIL approach vs. the MIL approach is that FOL provides the grammatical basis for an efficient and extremely precise (precise enough to describe multivariable calculus problems and solutions in an unambiguous way, for example) language, while HIL provides a generic framework for arbitrary bidirectional mapping between "Human Interactions" and "Facts." In other words, FOL ensures that the language has a Mathematically sound foundation, while HIL allows us to produce a variety of different written, oral, and other (sign language via body gestures, or a direct Braile mapping, as a couple of practical examples) forms to suit different audiences. (An English-like mapping might gain popularity in the US, while an Italian-like mapping might gain popularity in Italy, for example. An Italian who migrates to the US would simply have to learn the American oral mappings, in order to get along "on the street," and for anything written, he'd simply ask his computer to "remap" to his own favorite mapping. Now, let's say somebody in Greece writes an interesting paper, that an American wants to read. Normally, it'd be all Greek to him, but with the aid of an automated "translator," it suddenly becomes legible.) Since mappings are all relative to the Mathematically defined FOL, "translations" between different HIL mappings become a trivial matter for a simple computer program. Compare my "everybody feels like a native speaker" approach to the "Democratic compromise" approach undertaken by the Esperanto experiment, where everybody feels like a foreigner (and after all that, the language _still_ lacks precision). The only "downside" to FOL (and by extension, all HIL mappings) is that there's very little wiggleroom to "smooth things over" in a diplomatic way. (After all, the language was designed from the ground up for the specific purpose of highlighting differences, not masking them with ambiguity. In fact, I've always stated that it will be a source of great shame to me if a best-selling novel is ever written in MIL, and the same statement holds true for FOL.)