In Defense of Footnotes

by Jason Eisner (2015)

As has been noticed, my academic papers often include footnotes. This is not by mistake.

Reasonable people may differ in taste, and moderation is wise in all things. However, this page explains why my co-authors and I find footnotes indispensable.

Why is any explanation needed?

Apparently there's some prejudice against footnotes. Within my own field, the journal Computational Linguistics says: “Whenever it does not impede the logic or readability of the article, footnote material should be integrated into text.”

To my mind, discouraging footnotes is as peculiar as discouraging hyperlinks, section headers, appendices, bulleted lists, tables, captions, or italics. These mechanisms were invented to keep text organized and readable. Why not use them wherever they help?

What are footnotes for?

A footnote is a parenthetical remark—one that has been removed from the flow of the main text to avoid distracting the reader.

A footnote allows the surrounding text to retain a concise and graceful form:

Pat should have hurried home.7 However, she thought Sue was waiting elsewhere.
The underlined words refer back to the previous sentence. Interrupting that flow would be confusing:
Pat should have hurried home. (Recall that the characters in this story are robots, so their notion of “home” is governed by ISO Draft Standard ISO-13482.) However, she thought Sue was waiting elsewhere.

Footnotes can even be attached to shorter phrases or list elements. The footnotes themselves jump off from the point of reference and, being elsewhere, are free to digress for as long as they like.

A reader can skip a footnote without guilt and still follow the paper. Ordinarily you needn't consult a footnote unless you are closely studying that part of the paper. Isn't it nice when an author takes the trouble to distinguish the main ideas from the esoterica for you? Actually such two-tier systems are common:

Tier 1Tier 2
Main textFootnotes
Actual movieBonus features
EmailSocial media posts
Main routineSubroutines
Required coursesElectives

Some readers reportedly find footnotes distracting. Well, I find the Internet distracting, but I don't argue for shrinking it. Instead I try to resist the impulse to follow every pointer. These readers should learn to control themselves. They will find themselves reading wonderfully short, focused papers in nice large type.

But why are parenthetical remarks needed at all?

Because academia.

Academics are people who have an impolite amount to say. (Or is that just me?) Lingering over fine points, exploring connections, and acknowledging counterarguments are the badges of the academic. In the humanities, digressions are considered scholarly, and footnotes have a long and distinguished history.

An academic writer is faced with the task of laying out a complex argument in some order. Yet arguments are not naturally sequential. Mathematical proofs and computer programs are structured as directed graphs. Similarly, a technical paper is a network of interrelated statements. Unfortunately the statements must be serialized into some order, but an author can signal their relationships using discourse connectives. Parentheses and footnotes provide a useful stack mechanism.

The fundamental reason for footnotes is the problem of multiple audiences. Ideally, a paper would be an interactive teacher, always answering the reader's next question by offering frequent choices about what to hear next. Footnotes, sidebars, figures, and cross-references offer a more meager form of personalization: detours along a fixed trail.

Anyone reading an engineering paper should be able to come away with the main ideas. What was done, why was it done, and is there some evidence that it worked? However, papers are not merely advertisements for a product. They are the actual product. Thus, the standards of scholarship require a paper to include other information as well, information that will interest only a minority of readers:

Footnotes are sometimes the most convenient place to put this secondary material. Like tables and appendices, they are a resource to be consulted as needed. If you're reading the main text and start wondering about something, a footnote mark indicates that your question might be answered right at the bottom of the page.

The most engaged readers—who are studying a paper deeply in order to master, replicate, or extend it—will probably want to read it multiple times and eventually follow all the footnotes. It is crucial to serve this audience. But it is also important to serve the 90% of readers who just want the main ideas plus “details on demand.”

Couldn't you just rearrange the text to avoid footnotes?

Not necessarily. Rearranging the text would have side effects. Linearizing an explanation is a difficult optimization problem that involves tradeoffs. Often there are many points to make and multiple strands of argument going on at once. Sometimes the best solution involves footnotes.

Yes, parenthetical comments could be moved into separate paragraphs or sections that the reader could skip. But then they would be presented far from the material on which they comment. This would make them harder to find and harder to understand.

But come on, don't you just use footnotes to save space?

No, that's not why.* My dissertation had no page limits but it still had 214 footnotes. The conventional small size just serves to set the footnotes off visually and emphasize their peripherality. An equally valid way to accomplish this would be to use endnotes in full-size type (which I'd be happy to do), but most readers find footnotes more convenient. They're not even that much smaller.

When I write a paragraph, the footnotes are there from the start. They are not an 11th-hour layout edit. When even a space-hacked 11th-hour version still exceeds the page limits, we start regretfully deleting these footnotes. After all, they have already been designated as the most skippable parts of the paper. It's true that their slightly smaller size does allow me to keep one or two more of my little darlings.

Did you come up with the title of this page?

Yes; but so did Herma Hill Kay, in an entertaining article in 32 Arizona Law Review 419 (1990). I discovered this when using google to pull up my own screed. Prof. Kay has outdone me by demonstrating how to write an article consisting entirely of footnotes. For more footnote fun, see other links on this page.
*Not that I'm above saving space. Check out Smith & Eisner 2004, Figure 3. Kudos to Noah Smith for pulling this off.

This page online: http://cs.jhu.edu/~jason/advice/in-defense-of-footnotes.html
Jason Eisner - jason@cs.jhu.edu (suggestions welcome) Last Mod $Date: 2017/02/07 21:55:22 $