Hyperlinks and Less Wrong
Just prior to learning about Less Wrong, I read an article that discussed internet usage, hyperlinks, and reading comprehension. The focus of the article was on numerous studies showing that people’s reading comprehension goes down when they are searching the web, or viewing multimedia presentations. People reading an article with straight text and no distractions were able to answer more questions about it than people viewing a page that contained the same information, but which included video and hyperlinks. Even if you don’t click on the hyperlinks, every time you see one you have to consciously make the decision whether to click, which is distracting.
I will link the article at the end of my post. If you want you can go read it now.
In my personal experience, I find it not only distracting but frustrating. Each link is a tantalizing window into interesting-sounding-new-information, and I know that if I don’t click on it immediately I probably won’t bother to go back to it later, but that if DO click on it immediately I’m probably going to lose track of what I’m currently reading. It can be fun to link-crawl through Wikipedia, starting out with an article about prepositions, and somehow ending on an article about animal sexuality. But what’s fun is not the same thing as useful for education.
Enter Less Wrong. My initial reaction was “This is Wikipedia on Crack.” Not only do a lot of articles here feature a bajillion hyperlinks, but each link often goes to another lengthy article full of fascinating information that I don’t know, some of which is necessary to understand the first article, but none of which is easily summarized. With Wikipedia, if I run across a new word with a hyperlink it’s at least possible for me to glance at the hyperlink, get a quick sense of what it’s about, then return to my original reading. On Less Wrong that is often impossible. My first foray into this website felt like drowning in amazing ideas. It was a lot of fun, I definitely learned some things, but it was very confusing and hard to focus. Eventually I realized what the Sequences were and started reading them, but even they have a fair amount of interconnectedness that makes it hard to focus. (A related issue is that oftentimes I WAS already familiar with the introductory stuff, so I skipped ahead to the later articles, and then THOSE would link to various articles that I wasn’t familiar with).
I’ve been meaning to post about this for a while. I was particularly motivated to after the Scientific Self Help article, which was a) really interesting and important to me, and b) filled with an absurd number of hyperlinks, with varying degrees of usefulness. The problem is that even among the useful ones, I don’t have time to read *all* of them and still be able to really read the original article. I think the web in general and Less Wrong in particular would benefit from a new standard technique to provide relevant information that takes advantage of the web’s interconnectivity, without being as distracting. Part of that can be accomplished by including better summaries of information instead of linking entire articles. Part of it might require fundamental changes in site coding, which is more work than I am capable of doing. (And yes I’m aware of the “hey guys there this awesome idea oh by the way I don’t have time to do it” issue, but I think there’s legitimate work that non-programmers can contribute and I plan to if other people are interested).
Solutions?
I’m not sure what the ideal alternative is. Hyperlinks are used mainly for two reasons, both useful enough that simply saying “don’t do it” isn’t practical. One is to provide easy access to content the reader might not be familiar with but which is outside the purview of your article. The other is to craft a sort of joke, where the content isn’t actually important, it’s just funny that you pretended it was. Which isn’t strictly *necessary*, and probably is overused and would easily be replaced with alternate ways to be funny or poignant, but I’m not going to dismiss it out of hand.
The main alternate approach to hyperlinks is footnotes, and those have the opposite problem: by the time I get to them, I often forget what the footnote was about. What was interesting to me about the Scientific Self Help article was that it contained a LOAD of links to amazon.com books in the middle of the article. Then at the end it had a References section which listed a lot of books but DIDN’T include links to acquire them. (There were *some* links there, but they seemed to go to actual webpages, not places where I could acquire books).
I can think of some possible browser apps/extensions that might alleviate the issue. (Maybe let you click on a hyperlink and store it, along with the surrounding context, in another window, so you can store the links you wanted to remember and then read them all after the initial piece). But forcing people to acquire a particular app isn’t the best of solutions, especially when it’s not clear to a visiting user that they are supposed to do so. It probably wouldn’t actually solve the measurable reading comprehension problems either.
Edit: Yes, I know about tabbed browsing. But when you get ten+ hyperlinks per article, by the time you get to the end it’s hard to remember the context for the pages I opened. With ten+ links it also becomes difficult to read the titles of the pages so finding the ones I’m looking for is a pain.
A fairly simple solution (for some cases) is to replace or supplement hyperlinks with mouseover text. I’m not sure if that’ll actually improve the reading comprehension issues. (Probably not, so long as hyperlinks are the norm, so your brain will struggle every time it sees something that even *looks* like one). But Lesswrong could probably benefit from an established norm wherein top level posts come with a *brief* summary (no more than a paragraph) that can be easily referred to.
Facebook has a feature that goes a step further (and presumably requires some effort to program) wherein entering a link automatically grabs relevant information and attaches it to your post. This is something that works specifically in the context of Facebook though, where posts are short and you don’t need to wait before reaching the end for the relevant information. I feel like there should be some way to accomplish something similar for longer posts.
That’s all I have for now. I don’t know if other people are as bothered by this as I am, but assuming there is interest in this issue I’ll continue to think about it.
Relevant links:
The Web Shatters Focus, Rewires Brains (Wired Article)
Scientific Self Help (Example of over-hyperlinked Less Wrong article)
Upvoted. A lot of articles are overusing hyperlinks in my opinion.
To be clear, hyperlinks are fine, to some extent. They can provide necessary information, and I can open them in new tabs and read later, which I usually do. On the other hand, starting from certain frequency they compromise the readability, especially when the reader cannot assume what is linked to. I can confirm that the discussed (otherwise very good) lukeprog’s article can create impression of a “hyperlink abuse”. It starts Some have suggested that… and all four of those words are hyperlinks.
What I find acceptable or even beneficial:
Links to definitions of specialised terms which the audience may disinterpret (the term makes the link).
Links to Wikipedia (or similar) page about a not well know person (the person’s name is the link).
Links to original sources of statements or quotes. If the quote is longer than two or three words, please don’t transform the whole quote into a link; rather append “(source)” or something similar.
Links to really relevant articles, if those are online.
What I find distracting:
Links to unknown territory. I really appreciate if I know where the link will lead me. In the lukeprog’s article mentioned above the phrase “instrumental rationality” links to Anna Salamon’s LW article which does not include the string “instrumental” at all. It is not unreasonable to expect that such link would explain what “instrumental rationality” is; it doesn’t.
Links to Amazon pages about a book. Maybe I am mistaken, but I suppose people don’t buy a book suggested by an internet article before they have finished reading the article. Also, the combination of the book’s name and author is fairly unique and to find it on Amazon is quite straightforward. If books are linked to, I prefer doing it in a bibliography section.
Long links. The Latin alphabet is kind of optimised for reading, but not so much when underlined. Long stretches of underlined text are hard to read.
Lists of links disguised as normal text. Sometimes it is indeed appropriate to include a lot of similar links. If so, do it in a properly formatted list (preferably first briefly describing the link’s content followed by the actual short link). It is difficult to navigate through seven links on the same line.
Jargon definitions (sometimes). There are several phrases with more or less fixed meaning established often in the Sequences, and often when used they appear as a hyperling to the “defining” Sequence article. But consider: If the reader did read the relevant part of the Sequences and is an old LessWronger, she already knows the meaning. If not, using the jargon is not the best way to convey information, even with hyperlink included.
Anything in the first list with too high frequency. For example definition of terminology: If a term is really important, its definition should be included, rather than linked to. If it is not important, it can be linked to, but it may be omitted. If the author thinks that the article can’t be written without using fifteen concepts non-familiar to the readers, he would better not write the article at all, because before the readers pass through the fifteen links, they are likely to forget what the article is about. If the article doesn’t desperately need to use all those concepts, it will be most likely more comprehensible not using all of them.
Edit: One more point: The author is supposedly more informed about the topic than the readers, and so he can more easily decide what further reading is indeed relevant. If the article contains eighty unsorted hyperlinks, the odds are that really relevant pieces are lost among distractions.
I think this puts it more succinctly than I did, with nice bullet-list specifics. Thanks.
Middle-click on a link, or click on it while holding down the control key. This opens it in a new background tab for reading later. It neatly solves the problem. Sorry if this sounds overly blunt, but I thought everybody did this. When you see potentially interesting links, just middle-click them to queue them up for later.
(Also, regarding the links at the end of your post: please, describe what the links point at. Don’t just copy-and-paste the URLs. How are we supposed to know whether to click on them or not? This is an important part of good web writing.)
EDIT: Upon re-reading this, it still sounds a little harsh; tones of voice don’t carry over easily into text. So for middle-clicking practice and cheerful feelings, here’s a link to a cute picture of a cat looking at a duck.
This is the way that I browse, but I think it works differently for us than for others, since I know a lot of people who get lost doing this. I read link-heavy sites tab by tab, left-to-right, opening more as I go. The way Google Chrome opens tabs, this ends up being a depth-first search, and means that if I have fifty tabs open, the last one is almost always a child of the page I started on, while I might be eight levels deep in an entirely different branch.
My code-style explanation is this. When I click a link, I subconsciously allocate space for the new information. I store pointers to it, and start duck typing it, integrating it into my existing knowledge before I even have it. All of these connections are referenced by the word or word-phrase that was hyperlinked. When I get to the page, I fill in the knowledge, create back-references, clean up any dangling pointers that didn’t get addressed, and erase the tag phrase from my short term memory.
On the other hand, if I couldn’t efficiently reference content by word phrase tags, this wouldn’t work. I’d need context to remind me what the page I’m reading relates to, and why I opened it in the first place. If all pages had at the top the sentence or paragraph that contained the hyperlink that the user clicked on, this context would be apparent, and I think this is the kind of thing that Raemon is looking for. It might take some javascript-fu to get that working, but it would be a cool browser extension.
Yes I am aware of this. But by the time I get to the end of a single article, I have anywhere from seven to ten tabs open, and if I’ve taken time to read any of the linked articles it gets even more ridiculous. When you hit a certain threshold of tabs you can’t read what the titles are, and even after seeing the article it’s not necessarily obvious what was originally relevant to what.
(I admit my gut reaction is to feel a little condescended to here, but it was entirely possible, perhaps even likely, from your perspective that I really didn’t know that, and I can’t think of how you could have worded it any differently)
I would like more information than the Wired article provides. However, it doesn’t seem to me to support your hypothesis especially well; the studies it mentions (I would say cite, but it doesn’t) seem to be relatively unwilling to make value judgements about web-style content, focusing instead on the fact that we read and process it differently.
Personally, I like hyperlinks a lot. I have often (especially recently) thought that the main problem with textbooks is a lack of hyperlinks. I think hyperlinks can be distracting, but to me a hyperlink says “this is important background knowledge”. If I have it, it’s a little like an annoying pop-up window, but I don’t have the knowledge, that signal is really helpful to me. I also frequently read sites like Wikipedia several links deep, and feel like the mental of cost context switching is more than made up for in being able to learn something the exact moment before it’s necessary to understand whatever else I was about to read. Obviously that’s a difference between us.
One last point: I’ve seen multiple studies that demonstrated that people get information more quickly and more completely from bullet-style lists than from paragraphs. That style of emphasized and chunked information reminds me a bit of hyperlinks; the similarity might be entirely my imagination, but it seems relevant.
If people can find more detailed articles that discuss the science better, I’d appreciate that (I did some minimal digging, didn’t find anything particularly helpful). I know it’s not the most rigorous of articles and am open to the possibility that I’m worrying too much. But it did specifically state that reading comprehension and memorization drops when reading articles with a lot of hyperlinks. That’s not just different, that’s pretty explicitly “worse” assuming you value reading comprehension. I don’t think I made any claim that wasn’t also made in the article.
I do think hyperlinks accomplish good things. But I think there should be a way to accomplish similar things without the distraction. Using them sparingly to highlight important information may be a net-benefit. But when every other word in a paragraph is a link (as was practically the case in the Scientific Self Help post), and most of those links are not really important, I think you’re breaking even at best and probably at a net-loss. I don’t have enough data to know if/where a measurable point is at which reading comprehension loss outweighs extra knowledge. I would definitely like to see a more detailed study on that.
Less Wrong is particularly bad because the links take you to new, huge posts with no way to know what’s actually important to the original topic. Even if overall use of hyperlinks on the web wasn’t a problem, I’d say it’s a problem here. Because often the original topic was pretty complex on its own, and fully digesting it requires more attention than the average wikipedia page.
I think an ideal compromise might be to break long posts into (relatively) short pages, and include necessary links at the bottom of each page. That way people can read individual sections without interruption, but still have the bonus info while relevant topic is fresh in their mind. (In fact, this is how footnotes actually worked in old-print-media. You didn’t have to wait till the end of a 10 page article to find out what that little number meant, you’d find out by the end of one page.)
I completely agree with this. In fact, I tried to find ways to break this post down into bullet-lists, but I had already spent a few hours working on it and was getting tired and just wanted to post it and be done. I also don’t think it necessarily translated well into bullet-lists at all. But yeah, it’s a lot easier to learn that way for me and I try to do that where possible. However:
I think it’s your imagination. They look similar, but whereas a bullet list actually breaks stuff down into easy-to-digest chunks, a hyperlink brings you to an article that is NOT an easy to digest chunk, which then interrupts your digestion of an already-not-easy-to-digest chunk.
The similarity that appears to me is that the existence of a hyperlink (the visual cue) condenses text around it into a chunked thought. Too many hyperlinks, or hyperlinks placed or named badly fail to do this, and in that regard I agree with you. I sometimes read hyperlinks after the content, and sometimes I interrupt the content, but in either case the hyperlinked word usually serves as a mental tag that allows me to correlate the sentence it was in with the linked content, and helps me remember the sentence as a whole.
I think the “tagging” concept is why I can comfortably have fifty Wikipedia pages open at once, but I get lost only a few links deep when a link called “Carnot” takes me to a page that fails to use the same word and just uses “efficiency”. As long as the terms are consistent, I can usually get to the page even hours later and usually integrate it into the hole I had prepared for it while reading the original text.
On Wikipedia, go to preferences, click on “Gadgets” at the far right and tick Navigation Popups. It pops up a preview of whatever article is at the other end of a wikilink. The cure for mystery meat navigation, at least internally.
Thanks for the suggestion. I just tried playing with it, and will continue to for a little while, but right now I don’t think that it will help me.
1) It’s too slow. I normally read around 350 wpm (90% comprehension 1 day later), and read quickly at around 900 wpm (70% comprehension 1 day later). I usually read Wikipedia at my faster speed (they are discrete modes for me, rather than a continuum). On my connection, it takes about a second for a popup to appear, but even moving the mouse is an interruption. In my test just now, that reduced my reading speed by about half.
2) It requires concentration. This might change once I get used to it, but the way I pre-allocate an expectation for new information, and tag what I expect to go in there is subconscious, and happens easily at any reading pace. It seems to revolve around the association of a single word or word phrase. Just now, I had to focus on the popup and mentally note “Butte du Lion = Lion’s Mound” (and that’s an easy one!).
3) It’s only helpful for the corner cases, but I can’t predict those ahead of time. Wikipedia is pretty good about labeling links with the title of the page they link to. It seems that (a lot?) less than 10% of links violate this rule, and I can’t easily predict which ones; it’s only useful for me to see the popup when the page it links to is a redirect.
Ah ha! I just realized I could write some javascript to replace all links on a given site with the title of the page they link or are eventually redirected to. I wonder if that would be helpful? I would have to write different scripts for Wikipedia and LessWrong and TvTropes (maybe), but I don’t use that many link-heavy sites. On the other hand, it would ruin jokes and make some text nonsensical. Hmm.
Where/what do you do to check your reading comprehension like that?
Good question. By hand, and I don’t know how my results match up against more “official” methods, but here’s how I did it (a few months ago, so I remember it well):
1) I time myself reading 100 pages, and randomly sample from those 100 pages until I have a tight confidence interval for the word count of those 100 pages. This gives me my wpm.
2) The following day, I generate 100 random numbers between 1 and the number of lines per page in that book. For each page, I check 3 lines starting at that number (may bleed onto the next page). I count it correct if I remember the content of those three lines, and incorrect if I don’t. It’s subjective, but I’m strict. If the lines are
I count that as the facts: 1) Todd Kessler was a Nickelodeon producer 2) He worked on Sesame Street 3) but left dissatisfied
If I get any wrong (including failing to remember Todd Kessler’s name), that counts as a strike against the whole phrase, not just 1⁄3 of it.
Thus, my 90% comprehension rate means out of 100 samples, I got 7-13 wrong (most of those were in fact forgetting a person’s name, location, or organization).
3) Repeat for at least two more books to see how different writing styles affect my reading rate.
You could append the title of the page they link to- i.e. joke(Aha! Jokes: Clean Humor and Funny Pictures!)- and that would just make the jokes faster and the text interrupted. My concern would be that many titles are overlong. Do you want every Less Wrong link to have the phrase “—Less Wrong” in it?
Ah, I can see that. And I agree, if used responsibly, it probably accomplishes that to some degree. Again, what Less Wrong is lacking here is good summaries that allow the single word to actually encapsulate a relevant idea. (Or, as I noted in another comment, a glossary, so people can quickly learn new terms without reading entire articles)
The web usability guru is Jakob Nielsen.
One possible solution would be a button present on every page that would toggle hyperlinks. If pressed, all hyperlinks would disappear. If pressed again, hyperlinks would come back. A ‘reading mode’ toggle.
As you wish: Drag the link on this page to your browser’s bookmark bar. Clicking it on any page will turn all links black and remove the underlines, making links distinguishable from black plain text only through changes in mouse pointer style. Click again to get the original style back.
It’s like magic!
That would be useful.
That exists: Both Readability from Arc90 as well as Instapaper improve your reading experience. The first one additionally features the conversion of hyperlinks to footnotes (with backlinks), the latter one helps you with eBook creation (I load most of the SEP articles I want to read on my Kindle, for instance).
(edit To make it clear: It’s not exactly what has been proposed above, as they also severely alter the layout. A feature, IMHO. /)
Isn’t this somewhat solved by tabbed browsing? This has the problem you mention with footnotes, of course (though I find it a bit odd that you read footnotes at the end instead of immediately), but it’s simple.
The question in this post is more: Are Tab-Explody Sites to be Considered Harmful?
It’s a reasonable question to raise and it’s been raised well (and hence I’ve upvoted). Tab-explody sites like Wikipedia and TVTropes and LessWrong may not be a good idea. I like them and you may like them, but Raemon clearly doesn’t.
At this point, the appropriate thing to do, after a certain amount of typical-minding to get a notion of the problem domain, is to do some serious proper usability testing.
I think tab-explosion is the thing that distinguishes the web from other media. If you’re not going to link to relevant material, is the web really the best medium for what you’re doing, or would it be better presented as a journal article, book or similar?
There’s no reason we have to go from “no links” to “tab explosion.” There’s a near infinite amount of tab-possibility-space out there. Most of those possibilities won’t be ideal, and many of them may be worse than traditional print.
It’s worth noting that I actually DO “like” them, insofar as they are fun. I enjoy getting lost in TVtropes for hours. But it’s empty mind-calorie fun. TVtropes is pretty meaningless. It looks like you’re learning and feels like you’re learning, but at the end of the day you haven’t accomplished anything useful.
Now, Wikipedia DOES, theoretically, actually teach you stuff. It’s possible that at the end of the day you have benefitted from your eight-hour-link-foom. But I’m pretty sure that the neural impulses responsible for most of your “yay I learned something!” feeling at Wikipedia are the same ones responsible for TVtropes, and this says to me that these impulses cannot be trusted. How much you actually learned that you will remember a week later that will actually benefit you is not necessarily directly correlated to how much your “yay I learned something” receptors are firing.
Less Wrong is the only website where I viscerally felt like the information was too dense and the links too distracting to actually learn something, which is due in part to overuse of hyperlinks, in part due to the fact that almost everything on this site was new information to me, and in part because that information was all pretty dense.
I agree that Wikipedia is mostly empty mind-calorie.
Typical example for me: yesterday I had a question about hypobaric combustion. I found the answer, and because I had a question in mind before I looked for it, I will probably remember that answer for a very long time. However, by the time I finished my search, I had another four tabs open, and by the time I closed my browser, I had probably read through around a hundred pages. I remember their content pretty well, and will probably remember it decently next week, and probably remember some of it well in a year (the page on Howard Hughes, I expect).
However, I can’t absorb that much information in two hours. If I want to keep it in long-term memory, I need to put it into my Anki decks, but that would probably take more than an entire day. 20 hours is probably an underestimate, actually. That’s a clear sign that I’m running on the “yay I learned something!” impulse rather than actually trying to learn things. Another clear sign is how often I go back to the same pages, still without bothering to put it into Anki. It’s an impulse society is not very judgmental about, so I have difficulty feeling ashamed of it and correcting my behavior.
When you put it that way, I could see how many people wouldn’t be interested in spending a good three months of eight-hour days inside a Tab-Foom.
My previous office mate had Firefox store his tabs from session to session, and had a special tab manager, because he routinely kept between 200 and 1000 tabs open at a time. Some of them were more than a year old, and he still hadn’t gotten to them. He likes it that way!
Of course, he wasn’t spending solid days going through them, but I know of no clearer example of Tab-Foom.
I am your previous office mate :/. I don’t like it that way; I wish I could get the information I want to keep in my head into it, but somehow those tabs just keep sitting there—after you pass a certain amount, the “out of sight, out of mind” effect kicks in, and various other akrasiastic effects—and I only start afresh when I can’t recover a tab-set, which feels mildly catastrophic and mildly cathartic.
You had me quite confused until I looked through your other comments to assure myself you were being metaphorical.
Most tabs he gets through within a week. The ones that stick around longer are mostly references, articles he’s liked, sort of an alternate bookmark space. He actually prefers tabs to bookmarks most of the time, and has treated an unrecoverable tab set as a catastrophic event.
Maybe I’ve gaged his feelings wrong, but we’ve talked about it, and I don’t think he views it the same way you do.
If you don’t like it, what have you tried to do to change it? I’ve felt similar to the way you describe before, and solved it by forcing myself to shut down my computer at night (and not save tabs between sessions). If something is really important, I’ll get it over with before I shut my computer down, or I’ll bookmark it, but it stops plaguing me either way.
I hit something like 700 tabs the first time I fell over TVTropes. A testament to the powers of Firefox 3.0 at the time.
Did you read them all? TVTropes takes me a long time, because I generally read through the examples of the trope at the bottom of the page. 700 TVTropes pages would probably take me...a 40 hour week? Doing it in small chunks so as not to get bored...three months? I can’t easily conceive enjoying that much TVTropes.
I was using it as a spare-time background idle reading thing. A couple of weeks to skim that many pages :-)
More hyperlinks! Seeing a whole page of text without hyperlinks in it makes my eyes hurt!
I try to use links instead of googlepating if I’m (1) worried about inferential distance issues but (2) think a point is worth making anyway and (3) don’t think the backstory is worth spelling out, boring those who got it in the first place.
Another use is to gesture towards evidence in support of a controversial claim so that if I’m wrong there are fewer trivial inconveniences in the way of some reader finding out about that fact and calling me on it where the other readers can see, which is a component of epistemic hygiene.
If we’re counting, I disagree about having too many hyperlinks. WHen digging deep I open links in additional tabs as I need to and check them either immediately or later depending on context. I don’t trust tests of reading comprehension with and without unfollowed links as being relevant, obviously the text that experiment was designed for was a test of flat text reading comprehension. Finally I don’t believe the tested were limited to people who were incredibly facile with reading hypertext. I love the fact that these articles give lots of in line hyperlinks.
I really like hyperlinks as well. What I use them for (and like them for) is to keep track of the flow of an idea.
As I develop an idea, I often remember some places where that idea came from and I’ll link to the ‘seeds’ of that idea. The ‘seeds’ can be my own comments or someone else’s, and I feel I am pretty good at organizing information in this way.
Reading through a set of threads in a page, there is often an obvious vacuum where some other ideas should be linked. On Less Wrong, the same conversations occur over and over. Sometimes a conversation that is very interesting to me fizzles out and when that conversation spontaneously picks up again, it seems important to connect them with links.
Ultimately, I model Less Wrong as a brain, where more links between ideas corresponds to fast, more complex, non-linear and higher order thinking.
I agree it can be annoying and frustrating. In response to SarahC’s comment here where she writes:
I agree and feel the same way but it would be worse if the links weren’t there. While there’s a nagging sense of incompletion if you don’t link, without the links there is a real lack of completion. (Sometimes, knowing that humans can’t realistically follow, keep track of or care about all these interconnections, I imagine I am doing this for a future artificial intelligence building a model of how humans develop ideas or for a graduate student writing a thesis about the development of physical materialist viewpoints in the 21st century).
We can definitely improve hyperlink use with a list of best practices. For example, while referring to SarahC’s comment above I also copied the sentence I was referring to. I think this was helpful (you don’t need to follow the link, I already included the relevant information) and I did this consciously in response to criticisms I’ve read about hyperlinks in this post.
I would definitely like to see a more detailed analysis of the studies. I’m sure the truth is more complicated than “hyperlinks are bad. Always.” And the article doesn’t actually cite anything, so it’s possible some things are exaggerated. But I think the evidence is compelling enough that if you don’t think hyperlinks are bad for your comprehension, the proper response is “hmm, well, these experiments might be flawed, but I should think about what experiments wouldn’t be flawed and how I would respond in a least convenient world where it turned out that hyperlinks are worse than I think they are.”
And it seems to me this should actually be possible for us to run an experiment on, tailored specifically for Less Wrong members. It would take a fair amount of work to set up but it’d be doable.
My proper response is “I am going to continue to want to read heavily hyperlinked posts anyway, my confidence that this study is irrelevant to my particular case is high enough that neither studying this study nor fixing this study is concievably worth my time. And oh, I better post that opinion in case in the absence of posts to the contrary, some posters who currently do a great job putting hyperlinks in their posts are motivated to stop doing that thinking this new ‘information’ is valuable.”
But a proper response for someone who either takes the former study more seriously or takes the whole question more seriously might be to propose a study which looks for the difference.
Meanwhile, I’m still going after dualism vs materialism, the implications of newcomb’s problems for how I should think about the world, and whether or not free will really is trivially solved on lesswrong. I prefer to do that with plenty of hyperlinks as I cast about widely.
I think it’s fair to consider this enough of a nonissue to not care. But I honestly don’t understand where the confidence that the study is completely flawed comes from.
I am willing to put in effort to work out an experiment relevant to Less Wrong, but only if other people are willing to actually participate.
I don’t believe the original study was completely flawed, and indeed as far as it went it might not have been flawed at all. As I read it, the study showed
Experienced web surfers use a pile of neurons when surfing the web
Reading a sequential story where the links to further pages were somehow intermixed with the text of the story was slower and more confusing than reading a sequential story where the links to further pages were at the end and labeled “next”
Reading comprehension of a flat article which had not hyperlinks in it was better than reading comprehension of that same flat article when hyperlinks were added to it, whether or not the hyperlinks were followed.
If you want to make a test that shows hyperlinks are better, test the readers of the article with hyperlinks on some of the things the hyperlinks went to. We will then find that readers of the hyperlinked articles comprehended more than readers of the same articles without hyperlinks.
I imagine when I read hyperlinked articles on something that my alternative to a few hours of that is a few days of reading textbooks, and having to hope I remember the connections from one book to another and can find information in the other books when I need it. Even before hypertext I used book indexes and tables of contents extensively when I was learning something.
it’s called middle click.
Edit: I see the OP mentioning he already uses middle click. might I suggest tree style tabs available as extensions for firefox as an alternative to left-to-right tabbing.
tree tabs
I like your suggestion of adding mouseover text to hyperlinks. I know how to do it in HTML, but how do you accomplish it in LW comment markup?
To my mind, an author who does not provide links is pretending to know exactly what his reader’s background is and exactly what the reader wants to learn. So, when that author meets his ideal imagined reader, they together achieve a masterpiece of efficient communication.
An author who does provide links, on the other hand, can create a very satisfactory communication experience for a large variety of readers. A communication experience which can be enjoyed more than once by a single reader. An experience that teaches the reader things that even the author didn’t know.
I like an article with links.
There are a variety of add-ons out there, and they work for the reader without any special effort by the author. So there is no question of “forcing people to acquire a particular app”. I use Read It Later on Firefox.
Like this.
Which you get from this code:
Cool!
I actually didn’t figure out how to do it in Less Wrong Markup, I just went into HTML mode, which doesn’t seem to work for comments. So thanks for that.
The one issue with mouseover text is that it takes a pretty long time to appear (often longer than it takes to load a webpage). I’m not sure if that’s a setting that can be changed in browsers, or is unique to individual sites, or what.
I’ve actually seen it used pretty effectively on some forums, where mousing over a thread title shows you the first few sentences of a thread. That would be a great feature for Less Wrong, if we made it a convention for the first few sentences to summarize the article.
Or… does this site have a glossary? That might help dramatically.
I do think linkage is important. My issue is not that references should not be provided. Just that the way we currently provide them is at odds with how brains tend to actually process information longterm.
A suggestion for deeper links.
Maybe lesswrong could use Emphasis, for linking to sentences/paragraphs instead of whole articles.
I’m a sequential thinker so I just don’t click on hyperlinks as I read. I’m left with a nagging sense of incompletion afterwards, and if that nagging sense is strong enough I go back and check the links. Too many hyperlinks can be slightly irritating, because I have to make a mental note to go check a lot of things, as I’m reading, which feels like having to carry a very long to-do list in my head.
I think links probably make communication better on the whole. If you quote part of an article, mention an obscure concept, or cite a source, I’d rather have a link to the original than not. The ethic of linking freely prevents people from making as many unsupported claims or presenting other people’s ideas as their own. In that sense, linking changes our thinking for the better. My very first internet forum had a catchphrase “No linky, no believey.” I think that’s a good thing.
My two pet peeves about links:
Don’t use too many links to the Sequences. The “relevant” Sequences post is usually only a little relevant to the post that links to it, and linking to the Sequences becomes a complete digression.
Use “open in new tab” or “open in new window” instead of “go to.” You want your readers to remember they want to finish reading your post. You raise your chances of keeping their attention if you keep open the page the post is on.
The Read it later Firefox extension lets you right click on a hyperlink and select the option to store it on an easily accessible reading list (which you can access in drop-down format from a toolbar icon). The downside is your list will get very long and you will never finish reading all of it.
Thanks for posting, I had the same thought (but not the same dedication and research backup to post about it myself).
I think footnotes with links would be a decent compromise; limit in-article links to explanations of terms readers might not know.
In general seems that links should not be an “green smatter overlay” which the article reads without, they should be explicit references (e.g. “and take a look at this article”).
I actually literally read the wired article the day before I started exploring Less Wrong, a few months ago. I wonder if the article primed me to be particularly annoyed. But I think the issues exist regardless.
The same could be said about citations.(1) I like it if there are a lot of hyperlinks (if you want to know why go to the end of this comment’). And without either hyperlinks or citations a post would have to go to great lengths to define its presuppositions or expect its readership to either look everything up themselves or be aware of the discussed subject matter.
(1) I don’t see what difference this makes, except that you’ve to go all the way down to the end of the page.
′ Because additional information allow me to understand unknown concepts without having to Google for the answer.