Experiments in delinkification

A few years back, my friend Steve Gillmor, the long-time technology writer and blogger, went on a crusade against the hyperlink. He stopped putting links into his posts and other online writings. I could never quite understand his motivation, and the whole effort struck me as quixotic and silly. I mean, wasn’t the hyperlink the formative technology of the entire World Wide Web? Wasn’t the Web a hypermedia system, for crying out loud?

My view has changed. I’m still not sure what Gillmor was up to, but I now have a great deal of sympathy for his crusade. In fact, I’m beginning to think I should have joined up instead of mocking it.

Links are wonderful conveniences, as we all know (from clicking on them compulsively day in and day out). But they’re also distractions. Sometimes, they’re big distractions – we click on a link, then another, then another, and pretty soon we’ve forgotten what we’d started out to do or to read. Other times, they’re tiny distractions, little textual gnats buzzing around your head. Even if you don’t click on a link, your eyes notice it, and your frontal cortex has to fire up a bunch of neurons to decide whether to click or not. You may not notice the little extra cognitive load placed on your brain, but it’s there and it matters. People who read hypertext comprehend and learn less, studies show, than those who read the same material in printed form. The more links in a piece of writing, the bigger the hit on comprehension.

The link is, in a way, a technologically advanced form of a footnote. It’s also, distraction-wise, a more violent form of a footnote. Where a footnote gives your brain a gentle nudge, the link gives it a yank. What’s good about a link – its propulsive force – is also what’s bad about it.

I don’t want to overstate the cognitive penalty produced by the hyperlink (or understate the link’s allure and usefulness), but the penalty seems to be real, and we should be aware of it. In The Shallows, I examine the hyperlink as just one element among many – including multimedia, interruptions, multitasking, jerky eye movements, divided attention, extraneous decision making, even social anxiety – that tend to promote hurried, distracted, and superficial thinking online. To understand the effects of the Web on our minds, you have to consider the cumulative effects of all these features rather than just the effects of any one individually.

The book, I’m pleased to say, has already prompted a couple of experiments in what I’ll call delinkification. Laura Miller, in her Salon review of The Shallows, put all her links at the end of the piece rather than sprinkling them through the text. She asked readers to comment on what they thought of the format. As with Gillmor’s early experiments, Miller’s seemed a little silly on first take. The Economist writer Tom Standage tweeted a chortle: “Ho Ho.” But if you read through the (many) comments her review provoked, you will hear a chorus of approval for removing links from text. Here’s a typical response:

Collecting all the URLs into a single block of text at the end of the article works very well. It illustrates Carr’s point, and it improves the experience of reading the article. It also shows more respect for the reader – it assumes that we’ve actually thought about what we’ve read. (Which is not to say that all readers merit that level of respect.)

Now, Neuroethics at the Core, the fine blog published by the National Core for Neuroethics at the University of British Columbia, is carrying out a similar, informal experiment. As Peter Reiner explains, at the end of a lengthy, linkless post:

So here at the Core we are embarking upon a small experiment. For the next little while, we will try not to distract you from reading our blog posts in their entirety by writing them without hyperlinks in the main body of the text. We will still refer you to relevant posts, papers, etc., of course, but we will do so at the end of the post. Oh, the horrors, you might say, but really it is not so bad. One of my favourite science writers, Olivia Judson, regularly writes lovely articles for the New York Times in which she cites the relevant literature at the end of her article, and rarely includes links. If you have not read her posts, I highly recommend them. It would be great if you could share your experience of reading sans hyperlinks. Do you find it irritating? Does it allow you to read an entire blog post without skipping off to some other corner of the internet? Do you jump to the bottom of the post to get at the links anyway? Feel free to let us know.

My own feeling, in reading these works, is that I much prefer the links at the bottom. I do find that the absence of links encourages more concentrated, calmer, and more enjoyable reading. Of course, I’m biased. Try it yourself. You may be surprised.

And here, patient reader, are the links:

Salon review

Neuorethics at the Core post

Standage’s tweeted chortle

The Shallows site

UPDATE: Wow. This post really seems to have ticked off the Self-Appointed Defenders of Web Orthodoxy. Jay Rosen, the NYU journalism professor and ubiquitous web presence, even accused me of wanting to “unbuild the web.” Don’t worry, guys, no one’s going to take your links away. If you’d taken the time to read the post, you’d see that it is about some simple experiments (note headline) aimed at improving our understanding of the Net’s effects on attention, comprehension, and reading.

I don’t want to unbuild the web, but I do want to question it. Is that allowed, Jay?

UPDATE2: And now the king of the linkbaiters, Jeff Jarvis, accuses me of writing “that piece about links to get links.” Yes, Jeff, whenever I write a post with the craven intent of harvesting a lot of links I always make a point of publishing it on the morning of Memorial Day.

69 thoughts on “Experiments in delinkification

  1. Ajkohn

    It’s interesting but by putting all the links at the end of a post you’re actually training people to ignore your content.

    Research shows that people very rarely read word for word. Instead they scan content. And one of the ways you can anchor users and ensure they’re engaging with your content is … links.

    People aren’t patient and while it would be nice if they were, I don’t think that’s going to change. Fighting against this instinct doesn’t seem to be a winning strategy, nor is it entirely bad.

    Different mediums dictate different styles. A novel versus a haiku versus grant writing. They’re all very different in style, syntax and structure. Contextual links are simply a part of the style, syntax and structure of web content.

    Unlink At Your Own Risk

    Yes, I see the irony of putting my blog post link at the bottom of my comment.

  2. Morgan Warstler

    It promotes bad writing.

    Half way through your article I was bored and wanted to go to the CORE site.

    If you were a better writer and put the meat in the window (didn’t bury the lead), you’d have started with, “One way to force readers to read shitty writers online is to remove all the links… so that’s what I’m going to do.”

    Yes, yes I know you take your own time to write these words, it is up to you to decide… blah blah.

    You aren’t doing anyone any favors by pretending we can’t just just skim you to get the point, or see you TOTALLY miss the real point.

    You aren’t that deep, we can use the opportunity to get away from you. And since most of your readers are “one time shoppers” you OWE it to them, to put the knife in your own back and become a better writer.

    Here’s a hint: Far shorter paragraphs. It rewards the reader psychologically. Direct response writers have known this FOREVER. If you re-spaced this entire post never having a paragraph greater than 5 lines high, you’d get FAR MORE completes.

    Trying to help!

  3. Joan W

    Half way through your article I felt the need to skip to the end to see if you had provided a link to Laura Miller’s article. You see, I’ve become accustomed to opening hyperlinks of interest in new tabs as I read articles online. I have to admit that there is a down-side to my approach; I end up having dozens of tabs open at the end of the day, waiting for my attention. (Wow, it feels strange to write more than 140 characters at a time!)

  4. Thomas Hawk

    “UPDATE2: And now the king of the linkbaiters, Jeff Jarvis, accuses me of writing “that piece about links to get links.” Yes, Jeff, whenever I write a post with the craven intent of harvesting a lot of links I always make a point of publishing it on the morning of Memorial Day.”


  5. ibinsnet


    just 3 things:

    – could you please discuss this in context of Steven Johnson’s “Interface Culture” chapter on hyperlinked writing?

    – Ockham’s Razor: cut all “brain science”-crap out. is something lacking? (i’d strongly argue: no. this brain/cognitivism-stuff applied to culture/semiosis is just making us dumber.)

    – probably we should differentiate between 2 types of texts: one that is working better without links, and one that is depending on being hyperlinked. i’m not sure which one makes us smarter/dumber and would opt for the coexistence of both types.

  6. Nmannan

    Why don’t we get rid of all links? Like the ads in the sidebar that link to your books. I know I could concentrate better if I didn’t have to try to ignore the sidebar. Here’s an idea, include a link at the *beginning* of the article that allows users to view a text-only version of the article. I support experimentation, but this one seems a bit disingenuous to me.

  7. Nick Carr

    Craig, there are lots of studies of hypertext vs linear text dating back (at least) to the 80s. A good recent overview, which examines the results of about 40 studies, was published in 2007 in Computers in Human Behavior: DeStefano and LeFevre, “Cognitive Load in Hypertext Reading: A Review.” Conclusion: “the increased demands of decision-making and visual processing in hypertext impaired reading performance,” particularly when compared to “traditional linear presentation.” Lots more citations in The Shallows.


  8. Jason Treit

    Big ups to KK and Karlstrope above (click!), whose alternative should sap the last of anyone’s endnote nostalgia.

    Text is made of paths elsewhere. These paths summon anxiety, or as Kierkegaard would say, “the dizziness of freedom” to follow associations. The more I grasp an expressed idea, the more I feel the inadequacy of my grasp ā€“ while the cosmos of that idea bounds in so many directions at once, my thinking can’t. Such is life.

    Nick, if I judge Engelbart’s 1962 “Augmenting Human Intellect” rightly, he would agree that links shouldn’t jump out of text. Yet he recognized latent links in their full abundance. Words link to their broader uses; claims link to prior or forthcoming claims; figures of speech link to logical or playful relations; references link to outside texts. To him the job of computer-aided text was to enqueue as many links (and link kinds) at once as practical, invisibly, to be delivered at the speed of the viewer’s curiosity.

  9. Karan Bavandi

    Getting rid of links is like getting rid of references in a book. Links give articles credibility and show the depth of research by the author. Now making a decision to display embedded links obviously depends on the context. Who is your audience? What are you trying to achieve? So the decision to link or not to link is not one shoe fit all, obviously there are different scenarios and objectives. One solution would be to publish your links separately as an independent resource like http://www.kbucket.com. Your links can tell as much of a story as your article does. Linking is one of the main advantages of the web.

  10. Kroberts39

    If only there was a way to get hyperlinks into printed books. It’s so boring with all those words (on every page!) and no pictures and stuff. Oh, wait, I spoke too soon. A company called Ubimark “embeds books with two-dimensional codes that work as hyperlinks when photographed.” Cool! Here’s a visual (pictures!) demonstration:


  11. Tomforemski

    I agree. Putting in links is a pain, but I do do it if I’m blogging someone’s article otherwise at the end is better. I’ve also started using Pearltrees which is even better because it offers visual representation of the links in a post and also additional sites worth exploring.

    You can see here: http://www.pearltrees.com/foremski/

  12. Jack Damn

    But the only way for me to share this thought provoking article with my friends is by posting a link to it in my blog or Twitter account. Otherwise, my friends miss out on it and their ignorance compounds exponentially.

  13. Nick Carr


    My considered opinion is that if your communication consists of 140 characters or fewer, considerations of reader distraction need not enter into your calculations. Link away.


  14. dfarber

    Too many links can be distracting (especially those awful Bing links on MSNBC for common entites that pop up on mouseover) but let’s not over-rotate on link litter. Links allow people to experience what the web does well–make information webs, connect things, provide the option to drill in any relevant direction for another layer of info. As to studies that show cognitive issues with link litter, it may be that some pages are too rich with links but there is a whole generation of young people who are adept at filtering out noise and expect to be able to move quickly via gestures (swipe, for ex) and links to get the info they want.

  15. Nick Carr


    I generally agree with you, except for the assumption that “there is a whole generation of young people who are adept at filtering out noise.” The current evidence points in the opposite direction: the more time we spend zipping about the web and multitasking online, the less adept we become at filtering out noise. See Clifford Nass’s work at Stanford, for instance. And that goes for young and old alike.

    Also, there are deeper modes of thinking than “getting the info you want,” but these are being pushed aside by the Net. Increasingly, we are limiting our intellectual lives to “getting info.”


  16. Kevin v

    I find the impulse – is it a “reactionary” one – to remove the hyperlink perhaps historically parallel to those that felt that REAL writing was dis-served by illustrations in text, or even movable type. While I appreciate all the claims to cognitive yanking, the hyperlink works as one of the most dimensionally expansive ways of textually representing your thought. Certainly they are more than “footnotes”. Footnotes move from book to book, essay to essay, and back, but hyperlinks provide landscapes, whole topologies of investigation, and if a reader is so undisciplined to not resist their pull, well…perhaps there are reading skills waiting to be discovered.

    Hyperlinks need only be more quietly anchored, in text, perhaps. But not at the end. Putting such links in the foot is like putting the engine of the car only in the front because that was where the horses were.

  17. Mindsurgery.wordpress.com

    This is something I’ve been thinking about a lot: http://mindsurgery.wordpress.com/2010/04/30/the-bookpage-versus-the-webpage/

    Reading on a computer screen, my attention is splintered, and I tend to skim-read much more than I do with a printed page.

    It seems to me that hyperlinks are only part of the phenomenon, though – it’s also to do with the whole browser navigation system. When these days does anyone look at a single web page without other tabs open in the browser, and the email running in the background, and documents open, and twitter… and, on the desk alongside, a mobile phone, fixed phone, and every other mechanism of communication?

    There’s no such thing as single-mindedness, these days, I think.

    Perhaps someone needs to invent an app that makes it impossible to multitask.

    An anti-distraction plug-in?

  18. alqpr

    “People who read hypertext comprehend and learn less, studies show, than those who read the same material in printed form” deserves (needs!) at least a footnote (but a hyperlink would have been much better). The phrase “studies show” always sets off alarm bells in my mind because it doesn’t take much cleverness to design a “study” which will “show” just about anything. Regarding the issue in question, it would be easy to make up a document in which links are obstructive, but it seems that you have just demonstrated by example that it is equally possible to make one in which the absence of links is even more so (since I will not have been able to “comprehend” your post until I have had a chance to look at the sample documents used in those “studies”).

  19. Jason Tarre

    Every so often, an article will come out in the Wall Street Journal or the New York Times about how with digital media and digital devices, we will finally experience the multimedia revolution in education.

    Is this revolution backwards and twisted? Will a digital revolution short circuit the brains of young learners? Is a print education “best”?

  20. Laura Miller

    I’ve been pursuing this experiment with the articles I write for Salon by continuing to include any links to external sources in a final paragraph.

    An interesting result of this practice is that in generating the text for that closing paragraph, I’ve had to include a little more information about the sources themselves — i.e., “Here’s an abstract for the study mentioned,” and “Here’s an article about the study that offers a little more detail.” (Not all substantive research papers are posted online for free, by the way.) It’s a bit more work, but I believe it offers more value to readers, too, since part of what I’m offering is an evaluation of the source.

    Even more fascinating to see has been the response to this experiment. My readers have been overwhelmingly enthusiastic about the change, but Web punditry seem to regard it as a kind of blasphemy. There’s been hysterical exaggeration (pretending that the proposal is to eliminate external links entirely) and other misrepresentations, intentional or not, which surprise me in a community that’s always presented itself as embracing change and flexibility. I wonder if the prospect of overturning a single longstanding tenet of the digital punditry is threatening because it undermines the prophetic powers of Web pundits as a whole (i.e., if they were wrong about this back in 1995, maybe they’re wrong about other things today)? The whole discussion takes on the quality of a doctrinal war.

  21. Laura Miller

    One addition question, re: people who don’t feel that in-line links are an interruption because they open the source links in new tabs. Is this really so different from links at the end of the piece? I mean, if you’re reading the main text all the way through, and then moving on to the linked sources through a series of tabs, then it’s not as if you’re retaining the original context of the link. You’re simply reading a piece, then proceeding to its sources after you’re done. If you’ve got a lot of tabs open, as many people seem to do, then chances are you don’t remember *exactly* where the original link to it was, so why is it so important that it be embedded in the text?

  22. Steer9

    When reading books, I much prefer footnotes on the bottom of the page than at the back. That way if you want to read them you can, as if they were written inline. And if you don’t want to read them, just carry on as normal.

    You don’t have to consider a hyperlink a disruption of the linear reading process, any more than a phrase in parentheses is. Of course some hyperlinked text can be read in an in-line frame of mind better than others. It’s up to the author of the original piece to choose carefully.

    It’s also useful to distinguish between classes of linked information. A (paper) author would consider carefully when to put text in parens, when to write it in full as a footnote, and when to simply cite another work.

    For reference or factual work, links are still invaluable (although a decent search engine can semi-replace them). For more creative or abstract work, footnotes or traditional references may be a better mechanism

  23. HCIR_GeneG

    While your desire to streamline reading is a worthy goal, there may be compromise solutions that strike a better balance between deferred link following and preserving anchor context. See an FXPAL Blog post on hypertext anchors for more details.

Comments are closed.