Links are Not About the Relationship Between Texts

By | 2014/02/03

One of the most important distinctions we need to make when writing for the Web is the difference between how texts are related and how subjects are related. If that sentence made you say “Huh?”, let me explain.

In the book world, the principal reason for one book to refer to another book or article was for purposes of citation. A citation says, for support for this claim, see this work. The citation is used either to prove an assertion about another text, or to support an assertion about a subject with reference to an authoritative text. Any old text on the subject won’t do. It is the authority of the specific cited text that is being invoked.

In many cases, the citation comes with a direct or indirect quotation of the work cited. The citation is about the relationship between this book and that work, and its purpose cannot be satisfied by consulting any other work. Citations exist to bolster the authority of the current work, not to lead the reader in a new direction.

Most references in books are citations. Most Web links are not.

It is true, of course, that most Web links lead to specific works. (Not all do, because some launch queries that return a current result or set of results.) But those works are not chosen because the current text is making an explicit reference to the content of the linked work (a citation), but because the author is referring the reader to more information on a particular subject. The purpose of such a link could easily be satisfied by dozens of other links. We very often use links to Wikipedia for this purpose, when we don’t have a work in our own content on the subject. Their purpose is to lead the reader in a new direction (should they wish to go) or to allow them to fill in their background knowledge of a subject (should they need to).

I call such links, subject affinity links. A citation is a link to a work for the sake of the authority that work confers on the current text. A subject affinity link is a link to a work, or to a query, for the sake of the reader’s possible interest in that subject.

If this does not seem like a distinction of great consequence, watch this video in which Ted Nelson, the father of hypertext, demonstrates his vision of what hypertext ought to be. Would you want the Web to work more like Xanadu? Would Xanadu work better for the kinds of things you typically do on the Web?

The key thing to observe here is that the Xanadu project is all about the relationship between texts. It is interesting to note some of the anachronisms in Nelson’s thought. He talks about marginal notes several times. Electronic texts, of course, don’t have margins (except as an artifact of their presentation on a particular device). Marginal notes are, specifically, commentaries on the text. If you are making notes on a subject, you do so in  a separate notebook, in which you combine information on your subject gleaned from many texts. Marginal notes are something you make when you study a text, not when you study a subject.

Nelson wants hypertext to show us the relationship between texts by showing the quoted work alongside the work that quotes it. If that seems a highly academic way of looking at a text, it is. It belongs to the academic study of texts. Most of the reading we do on the Web is because we are looking for information on a particular subject. If we follow a link, it is because we hope it will bring us closer to our subject. Being shown the relationship between the source and destination texts, as Xanadu does, would not help us in the pursuit of our subject. It would be clutter and a distraction.

The difference between studying texts and studying subjects is huge. When you read Dante’s Inferno or Aristotle’s Ethics, you are studying a text. If you are examined, you will be examined on the text and its interpretations. When you are reading Learning Python by Mark Lutz , you are studying Python. If you are examined, you will be examined on your knowledge of Python and your ability to write Python programs, not on your knowledge of Lutz’s text.

Classical Western education was largely the study of texts. To be educated, in the Western tradition, was to know the great works, to have studied and learned the required texts. The validity of such an education is a topic for a different debate in a different venue, but while this curriculum is seldom taught anymore, its focus on texts still affects the way we think about individual texts and the importance we attach to them.

Regardless of the virtues or deficits of a classical education, a knowledge of text and their relationships is not what most people who use the Web or technical communication are interested in. They have no interest in the texts for their own sake, and therefore no interest in how the texts are connected. They are interested in subjects, and in how subjects are connected.

Wikipedia is a particularly interesting specimen to examine when exploring the difference between citations and subject affinity links. Wikipedia is meant to contain no original opinion or research. Every statement is meant to be backed up with a citation to some other work. At the same time, Wikipedia articles are rich in links to articles on related subjects. These subject affinity links are not citations. (Wikipedia cannot cite Wikipedia as a source.) They are intended to help the reader to get more information on any subject mentioned in the text they are reading.

Wikipedia, therefore, is rich in both subject affinity links and in citations, but it keeps the two strictly separate. Subject affinities are expressed through links on words and phrases in the text. Citations are expressed through footnotes. This subordination of citations is important. Wikipedia is structured (from the bottom up) around the relationships between subjects. Citations (the relationships between texts) are important, but they are not structural. They are treated as secondary and moved out of the way.

Most readers are not interested in exploring a universe of related texts such as Xanadu provides. We know this from the repeated failures of many such projects. Where all these visualization techniques have failed, Wikipedia has conquered all. There is no map or plan to Wikipedia. There doesn’t need to be. There are only flat pages with basic structures, some headlines, some pictures, and lots and lots of links along lines of subject affinity. This lets people pursue subjects. And subjects, not texts, are what matter to readers.

For technical communication, and indeed for content strategy and information architecture in general, this is an important lesson. The reader is not interested in the relationships between your texts. They are not interested in how your manual is laid out or how your help system or website is organized. They are interested in the problem they are trying to solve, the subject they are trying to learn. They don’t want to have to learn the relationships between your texts in order to pursue these interests. They want to follow the relationships between subjects. Help them to do so. Link your content richly along lines of subject affinity.



One thought on “Links are Not About the Relationship Between Texts

  1. Samuel Danielson

    Thank you for sharing. I discovered you blog several hours ago and it has now consumed my evening. The following quote got my attention because, though it is true, there is a strong case for transclusion/citation in electronic documents.

    “The reader is not interested in the relationships between your texts. They are not interested in how your manual is laid out or how your help system or website is organized. They are interested in the problem they are trying to solve, the subject they are trying to learn.”

    Very true. But who maintains this manual? As an instructor at my airline I am very much interested in the relationships between texts. There are top-level legal and certification documents for the airline, laws, FAA approved maintenance manuals for the aircraft from the manufacturer, aircraft operating manuals, etc. From those we derive company specific operating manuals. From those we derive instructional materials. From those we derive informal gouge. There is a hierarchy of authority in these documents and the most usable information is often the furthest from the most authoritative. It takes a whole department to keep the FAA approved mid-levels updated and most of the training staff to keep the training materials updated. Good gouge goes out of date. Front line employees are confused and behind the times.

    Can transclusion and citation help? I think so. At the very least it would make inconsistencies more obvious, if not to the user then at least to the manual maintainer. When a figure is included in a powerpoint slide, is the included figure still current? When text describes a quoted block, is the quoted block current? And when a rule derives from an invisible citation, is there still a reason for the rule? Perhaps there is no end-user value in citations/transclusion but there is a business case in reducing the costs associated with keeping current manuals.

    BTW, transclusion and citation are the fundamentals of online advertising. Adblock is nothing more than a transclusion and citation blocker! You must be on to something when you say the average user is not interested in these features.


Leave a Reply