Link fanaticism

« previous post | next post »

It's a very small point, but it annoys me, this fanaticism on Wikipedia for providing links to every mentioned entity that has a Wikipedia entry of its own. My own Wikipedia page is just a stub of five short sentences, but it has ten links, to:

linguistics, Stanford University, Distinguished University Professor, Ohio State University, Morris Halle, MIT, Edward Sapir, Linguistic Society of America, UIUC, Language Log

(plus an external link to my homepage and a "see also" link to Recency Illusion).

The problem is that these links are visually obtrusive. They scream. And they point you to webpages that you probably don't want to see (because they don't really provide any useful background information about me) and could in any case be accessed by a simple search on an obvious phrase in the text.

Link fanaticism is not some accident. As I discovered some time ago, it is PRESCRIBED Wikipedia style. There are people who view any unlinked reference as a FAULT, and edit pages to insert the (I'm sorry to say this) missing links. What the editors are after is perfect consistency and uniformity. But the point of links is that they should be useful and helpful — which means that the writer of an entry needs to take the readers' likely knowledge and interests into account and use JUDGMENT in inserting links. Skillful linking is, in a way, like the skillful deployment of anaphors in writing or speech.

When I talk to friends about link fanaticism, some of them say, "Well, that's just geekiness". Maybe, though I don't want to be too quick to attribute motives or personal qualities to link fanatics, nor do I want to put down people who might be labeled "geeks" (truly, a great many of my friends might be characterizable as geeks, and in some contexts I might myself be so labeled). (Of course, Wikipedia has a page for Geek, which is itself dense with links, including one to a page listing events that take place or took place on May 25, occasioned by the mentioning of the remarkable fact that "Nerd Pride Day has been observed on May 25 in Spain since 2006"; there are also links in this sentence to Nerd Pride Day and Spain, but not to 2006, though the year does have its own page.)

My own practice in linking on Language Log is to be sparing, so as not to clutter up my postings. I insert links when I think they would be especially useful and would not be easily findable. The Wikipedia Geek page is very much a side issue in this posting, and it's easy to find, so I didn't link to it above. On another occasion, I mentioned two books by my step-daughter Emily Transue, by title, and a commenter supplied links to the (U.S.) amazon.com pages for the books, though anyone interested in buying the books could easily find them at the bookseller of their choice. On still another occasion, I referred to Mel Brooks, and a commenter provided a link to the Wikipedia page for him, in case some readers didn't know who he was, though any reader unfamiliar with Brooks could get information about him by searching on his name.

It was a judgment call not to give links in these cases. But giving links everywhere you can would be crazy. That leads to things like the preposterous link to May 25 on the Wikipedia Geek page. And even if SOMEONE might want a link, it might not make sense to provide one.

Consider the following fact: not infrequently, readers come to Language Log by following up a link from another blog. They then stumble into a context where they don't know the players (they might not even understand that this is a group blog, with postings from individual bloggers in their own voices), don't know the history, and don't know all the terminology we use. If I tried to take these readers' concerns into account, I wouldn't mention the other bloggers without giving links to their webpages, wouldn't mention eggcorns without a pointer to the Eggcorn Database, wouldn't mention "singular they" without links to relevant pages (it's come up at least 25 times so far), and so on. Every posting would be a forest of links, just like those Wikipedia pages.

Instead, I expect readers to do a certain amount of work in understanding what's going on here. And I try to give help where I think it's most needed.



70 Comments

  1. Rob Gunningham said,

    July 17, 2008 @ 1:58 pm

    In general I agree with this, but she's not quite as well known as Mel Brooks and so it was useful to have that link to your stepdaughter's books, I remember.

  2. Arnold Zwicky said,

    July 17, 2008 @ 2:31 pm

    Rob Gunningham: "In general I agree with this, but she's not quite as well known as Mel Brooks and so it was useful to have that link to your stepdaughter's books, I remember."

    But I gave her name, the titles of the books, and the publisher — more than enough information to let you go to the bookseller of your choice on-line (which might not be U.S. amazon.com).

  3. nsidestrate said,

    July 17, 2008 @ 2:38 pm

    I'm not certain that PageRank is what motivates them to do this, but I'm pretty sure that a side effect of this policy is to inflate the number of links that Google sees and to promote their article to near the top of any search for the term.

  4. K. said,

    July 17, 2008 @ 2:44 pm

    Wikipedia, like all encyclopedias, is best used as an initial source of information for things one doesn't know anything about, and it is natural to assume one might need to have some background information about the subjects one is looking up.

    Take your own stub, for example. Perhaps a webizen is looking you up because they were linked to a LL post of yours from some non-linguistics blog. If they're trying to scope out your credentials for some claim you make, it seems natural that they'd be interested in knowing more about your discipline (plenty of people don't know what linguistics really is) and about the schools you have worked at. The scholars you have worked with are also relevant, as well as what your title means (perhaps they're unfamiliar with professorial designations), and the associations you've been a member of. The article on Language Log seems self evidently material.

    Do we need links to Edward Sapir and UIUC? Probably not, but it's easy enough to ignore them. I've always suspected that the policy of linking all possible articles was largely motivated by the desire to generate traffic -thus, potential improvement- to less popular articles( not that UIUC or Edward Sapir are particularly obscure, but how much traffic could "Distinguished University Professor" possibly get?)

  5. Ryan Rosso said,

    July 17, 2008 @ 2:45 pm

    I personally disagree. I rather enjoy the links on the wikipedia pages and find them helpful. I would rather click the link than search for each item, sometimes struggling to find the correct entry in the search list. I do, however, understand how ridiculous it could really get, with every word linking to another entry. Perhaps wikipedia should offer a simple option for users to choose whether they want links to show or not.

  6. mae said,

    July 17, 2008 @ 2:56 pm

    I agree and I'm really happy to see someone else who is annoyed by this fetish.

    But there's worse noisy linking in the New York Times and some other papers. I suspect that these generate random links by computer, which often are TOTALLY irrelevant, sort of like puns. For example, if a person's name is the same as the name of a (random) city, a mention of their name will link to a page for that city. I can't provide any actual specifics, but I assume someone is making a collection.

  7. Harry said,

    July 17, 2008 @ 3:01 pm

    Count me as one vote in favour of heavy linking on Wikipedia (though I agree the May 26 link is just silly).

    I sometimes find links in blog posts irritating because I never know what the links are likely to be, so so I find myself clicking them or hovering the cursor over them to find out, and much of the time they're not relevant. But internal Wikipedia links are generally self-explanatory, and I find them often useful.

  8. Angelina said,

    July 17, 2008 @ 3:06 pm

    Part of the problem for me is that the internal search engine on Wikipedia is pretty terrible. As someone who has grown used to Google correcting typos in search strings I find it extremely frustrating when it takes a coupe of tries for me to find the proper spelling of a name or item I don't know much about in Wikipedia. Building those links in when they're relevant to the article at hand is really helpful, and I'd rather a few too many links there than too few.

    It may be different in your site, which is a blog dedicated to a particular topic, but in a general interest 'pedia-type site with a billion different topics the over-linking makes perfect sense. The whole point is that it's for reference, so an assumption that someone already knows about the topic seems a bit backwards!

  9. Nathan said,

    July 17, 2008 @ 3:07 pm

    Maybe it's because I'm a seasoned Wikipedia editor, but I disagree with you completely on this. I don't think wikilinks are visually obtrusive. They're highlighted, but it's a simple change in color, to the blue that has long been standard on the World Wide Web. They're certainly much more convenient then having to "search on an obvious phrase in the text."

    I also don't agree that "the writer of an entry needs to take the readers' likely knowledge and interests into account". I'm an Arnold Zwicky fan, but not every Wikipedia reader is, not even every Wikipedia reader that reads your article. Someone may be reading a particular article for many different reasons, including because he just hit the Random Article link. Just because an article doesn't "really provide any useful background information about" you doesn't mean readers "probably don't want to see" it. If I were editing the article (which I probably won't do now that I'm saying this) I would probably add several more wikilinks, at least to the technical linguistics terms used in the article. It's especially convenient for a reader to be able to click on an unfamiliar term and learn some of the basics.

    Wikipedia's style guideline on writing better articles includes the following:

    Wikipedia is an international encyclopedia. People who read Wikipedia have different backgrounds, education and worldviews. Make your article accessible and understandable for as many readers as possible. Assume readers are reading the article to learn. It is possible that the reader knows nothing about the subject: the article needs to fully explain the subject.

    Remember that every Wikipedia article is tightly connected to a network of other topics. Establishing such connections via wikilink is a good way to establish context. Because Wikipedia is not a long, ordered sequence of carefully categorized articles like a paper encyclopedia, but a collection of randomly accessible, highly interlinked ones, each article should contain links to more general subjects that serve to categorize the article.

    Obviously the article about you needs a lot of work, but I think every one of the links in it should stay.

  10. Colin said,

    July 17, 2008 @ 3:08 pm

    I disagree with the assertion that links somehow "shout" and find that I completely ignore the links in Wikipedia (or any other HTML text for that matter) unless I'm actually interested in information about the hyperlinked term. In those cases, the links are invaluable. Even links to dates can be of interest to certain folks. Of course, I have been using the World Wide Web sine I was a teenager. Perhaps this is a generational thing? No offense, but it feels like you're being a prescriptive usage crank here.

  11. M said,

    July 17, 2008 @ 3:16 pm

    Is it the fact that there are links that troubles you, or their prominence? It's very easy to code a web page so that hyperlinks are not bright blue; they can in fact be any color, even black like plain text. A site with many internal likes, like Wikipedia, might well benefit from a more subdued color.

  12. Sili said,

    July 17, 2008 @ 3:18 pm

    As the one who supplied the aforementioned links, my apologies.

    I'm afraid that I like links to Amazon even if I do not buy all that many books there. I simply find the reviews and blurbs helpful. I wrongly assumed that you had not supplied direct links out of a sense of modesty.

    Again: I'm sorry.

  13. Nick said,

    July 17, 2008 @ 3:21 pm

    I also have to disagree. I find the linking enormously helpful, especially when I'm delving into a subject about which I know little. The ability to quickly link to (and return from) pieces of information that are relevant to the article I'm reading is key. If I had to copy and paste those items into a search box it would take a lot more time and I'd be much less likely to bother.

    My only criticism is that all the links tend to increase the total time I spend (waste?) on wikipedia. Once I start down the endless link path I tend to lose much more time than I originally intended.

    One suggestion: perhaps wikipedia could prove a button to either (a) turn off the links or (b) change them to be the same color as the rest of the text. If the visually-jarring nature of the blue text bothers some users, option (b) would preserve the link functionality while making the text more readable. If you just hate the concept of extensive linking generally, then option (a) is for you.

  14. Robert Sharp said,

    July 17, 2008 @ 3:22 pm

    I love the multitude of links on the wikipedia. Sometimes I will just start with a topic I want to learn something about, and then spend three hours reading because I want to learn about some of the topics linked to within the text. And I end up with 35 tabs open. The only downside is generally i have to just quit for the day, because each new article sparks my interest further with all the new links in it. I don't think I've ever just dwindled to nothing while reading on the wikipedia; I always just have to decide that it's time to go feed myself.

    It is my opinion that well-placed links in wikipedia articles ignite a passion for learning more rather than distract a user unnecessarily.

  15. Thomas Thurman said,

    July 17, 2008 @ 3:32 pm

    Are you logged in? If so, when did you create your account? Older accounts have an older stylesheet which underlines links. Newer accounts merely make them all blue. Either way, anyone who wants can make their links look like anything they want by editing their own stylesheet.

    Richly-linked text was part of the original vision for the web, just as editable text was; it hasn't widely been used, but Wikipedia is an example of what might have been in both cases. Most people use the web as though each site had a tree navigation pointing to a set of simple pages of text; we might as well still be using gopher. (I remember as a child, well before the web started, reading about hypertext in a magazine and being filled with excitement at the possibilities it offered; Wikipedia is always tied up with those memories for me.)

  16. henitsirk said,

    July 17, 2008 @ 3:37 pm

    I've edited for Wikipedia and Simple English Wikipedia. My assumption is that it should be like a paper encyclopedia: if I look something up, and there is a term in the article I don't know, I will want to look that up too. In a paper encyclopedia, that would involve turning pages or choosing another volume. Online, that happens via links.

    It can seem out of hand, but then I've come across lots of fascinating and useful information from all those links. And with Simple English Wikipedia, it's all the more important because we are assuming that the reader doesn't know what the linked terms mean.

    And of course, anyone can do a web search for Mel Brooks or Arnold Zwicky, or a book they'd like to buy. But then there are people who have very limited computer and internet skills, who would be well served with a link rather than trying to do a search.

  17. Cameron Majidi said,

    July 17, 2008 @ 3:43 pm

    I wondered what you were talking about until I realized you were probably seeing the default skin. Change the skin on your wikipedia profile to something other than Classic, and you won't see links as visually intrusive anymore.

  18. Thomas Thurman said,

    July 17, 2008 @ 3:54 pm

    Sili: IIRC Wikipedia auto-linkifies ISBNs in page text anyway, and it goes to a page where you can look up the book in libraries or buy from dozens of booksellers.

  19. john riemann soong said,

    July 17, 2008 @ 4:06 pm

    I'm a Wikipedia sysop (under a different name).

    I have reason more superfluous wikilinking and I think your page is quite reasonable.

    In general, wikilinks should be used when the general public is likely not to know the material in question. This is because wikipages in the end are not isolated articles on specific subject, but a comprehensive network of articles covering a broad area. If you have read the xkcd episode on "the problem with Wikipedia," the typical user's experience of being engrossed in article after article because of an ever-expanding network of piquing links tends to be a good thing.

    Does the general public know about Edward Sapir or Morris Halle? And without comprehensive wikilinking, how would I know the interesting fact that both of Liberman's parents were psychologists?

    "though I agree the May 26 link is just silly"

    Actually this is just general policy — all dates are to be wikilinked, in a specific format. Often it's not for people to click on them: it's so that the dates show up on the database and and an assortment of individuals, including scripted bots, can maintain various lists, especially those concerning births, deaths or anniversaries of various events. But by policy each specific date is to be wikilinked once per article.

    Articles littered with blue IMO look cool. It's been that way for almost half my life now and it jars ME when I notice an article that is 90-95% black.

  20. Eric Bakovic said,

    July 17, 2008 @ 4:08 pm

    I'm of two minds about too many links. On the whole, I find them useful enough to balance out the annoyance factor, but then again I long ago changed my browser preferences to remove underlining and it's very jarring when I switch to a browser with underlined links.

    I do worry about the link (ahem) between link fanaticism and other Wikipedia cross-reference fetishes, though, as illustrated in this lovely xkcd comic:

  21. Shmuel said,

    July 17, 2008 @ 4:11 pm

    I was going to check out Arnold'z wiki (heh) but there was no link.

  22. mollymooly said,

    July 17, 2008 @ 4:15 pm

    There are firefox addons that will turn every word on every page you load into a link to Wikipedia, Google, or whatever. I guess Arnold won't be downloading those any time soon. I won't either, but nevertheless you can still count me among the heavy-linkers.

    Having used Wikipedia both instrumentally and for idle browsing, I have found I get more from reading through a series of articles which are related, each of which is short but well-linked to the others. Attempting to write a long survey article using Wiki methods will result in a big pile of collaborative goo, where too many interlinks is the least of the problems.

    The justification (inadequate I will agree) for linking "May 26" is that it allows the Wiki software to automagically change this to "26 May" depending on your locale.

    The justification for tons of articles about nerd trivia is "Wikipedia is not paper", i.e. they're not stealing space that would otherwise be devoted to worthier topics. In that spirit, I would say: Wikipedia is not a blog, an online newspaper, or any other type of website where the linking tradition favours sparseness. Since I stopped regarding each wiki-link as screaming "click me! click me!" they haven't bother me at all.

  23. JK said,

    July 17, 2008 @ 4:24 pm

    I will join the chorus of pro-hyperlink comments. Hyperlinks are the lifeblood of the internet, without them it would not be all that special. Lots and lots of links are especially appropriate on Wikipedia, which is an encyclopedia best used for basic information gathering. As opposed to this blog, which is a forum for commentary on a specific area of academic study. It is almost certain that the vast majority of readers find this blog through hyperlinks — whether you take those "readers' interests" into account in your linking practices is up to you, but I'd suggest that this traffic is vital to keeping this blog a thriving venue and one would be well served to cultivate and reward such readers. Also, perhaps I'm misunderstanding the post, but it seems to me poor form to blog about other bloggers without providing a hyperlink to their sites. After all, the proliferation of cross references is part of what makes the whole blogosphere work.

  24. Andy Hollandbeck said,

    July 17, 2008 @ 4:40 pm

    All those links have an economic purpose, too. Search engines live for links, and links to pages that are highly viewed get more weight in the search engine algorithms. Including links to real useful (not necessarily really useful) pages in the content of the page helps that page climb to the top of Search Results Pages, aka SERPs.

    And apparently it works for Wikipedia. Just about any Google search results in a Wikipedia #1 slot.

    And to follow up on what M first mentioned, in Firefox, you can change the link colors by choosing Tools–>Options. Find the Colors button on the Content tab. Make sure you deselect the Allow Web Pages to Choose Their Own Colors, Instead of My Choices Above option. If you find that links scream too much, I suggest changing the text to black and keeping the underline.

  25. Jonathan Lundell said,

    July 17, 2008 @ 4:42 pm

    I find the links useful but intrusive, so I'm happy that this thread induced me to find the Wikipedia preference (under the Misc tab, not the Skin tab where it belongs) to turn off link underlining altogether. I'm not feeling ambitious enough to rewrite a style sheet, though.

  26. Rob Gunningham said,

    July 17, 2008 @ 5:06 pm

    Arnold Zwicky said, "But I gave her name, the titles of the books, and the publisher — more than enough information…"

    Sure, but it takes a lot less time to click a link.

    I can't remember all the details, but I think I hadn't decided when I read your piece that this was a book I needed to buy. Had I decided, then your info would have been enough, but later I clicked Sili's link and found, 'oh yeah, this is really interesting, she's a doctor', etc…

    I thought, like Sili, that you hadn't linked it out of (misplaced) modesty.

  27. Marbury Apricot said,

    July 17, 2008 @ 5:11 pm

    I read Wikipedia like Robert Sharp does (opening a new tab for every link that strikes my fancy) and so I appreciate the extensive linking–in fact I'm annoyed if an obviously linkable concept is not already linked for me. You say, "the point of links is that they should be useful and helpful — which means that the writer of an entry needs to take the readers' likely knowledge and interests into account", but I suspect the Wikipedia editors are much like Robert Sharp and myself and enjoy being able to stream-of-consciousness their way through the encyclopedia, and thus the links are absolutely appropriate.

    Further down you compare your own linking style to Wikipedia's, but why should your judgment about what makes a good link in a weblog article inform a decision about how to decide what to link in an encyclopedia? I don't expect to see links to everything imaginable in a weblog post, but Wikipedia's extensive cross-referencing is one of its greatest strengths as an encyclopedia, I would argue. The aesthetic results may not make for the most efficient reading, but as has been adequately pointed out, there are many many possible solutions to that problem.

  28. kip said,

    July 17, 2008 @ 5:15 pm

    I am pro-linking, but I don't have much to add to what was already said. I do want to correct this:

    All those links have an economic purpose, too. Search engines live for links, and links to pages that are highly viewed get more weight in the search engine algorithms. Including links to real useful (not necessarily really useful) pages in the content of the page helps that page climb to the top of Search Results Pages

    And apparently it works for Wikipedia. Just about any Google search results in a Wikipedia #1 slot.

    Unless I am mistaken, search engine page rank is not boosted by links within a site to other pages within a site. (Otherwise people would generate websites that link to themselves thousand of times.) The reason Wikipedia is the first hit on most searches is because lots of other sites link to Wikipedia.

    Still, having lots of links does improve the search engine's ability to find everything on the site.

  29. rone said,

    July 17, 2008 @ 5:15 pm

    However, a link on the sidebar to the Language Log Glossary, which would contain the explanation of terms such as eggcorn and snowclone, would be a valued addition to the site template.

  30. David Scrimshaw said,

    July 17, 2008 @ 5:16 pm

    Arnold, I share your view that linked text "screams", and I also don't like using it in my web writing because it encourages people to jump away in the middle of reading it.

    My solution is to put useful links to things referred to in my postings in a list at the end. (Although I'll still put links in the posting if they are funny.)

    If I am telling people about something I want to support, I try to make it easy for them to find the pages that will let them purchase, participate or learn more. I figure the casual reader is more likely to follow up if they don't have to take the trouble of doing their own search for more info.

    I figure that the effort it takes me to add a relevant link is less than the effort it takes a person to do a search for that link and so if one person clicks, I've created a net gain.

  31. Andrej Bjelakovic said,

    July 17, 2008 @ 5:20 pm

    When I was writing a short stub for Wikipedia, started with having only two links and I liekd the way the article looked. Then, a little later, I added some more just to see how it would look like. And, much to my surprise, I liked it. It was only then that I realized that all those links not only don't annoy me, but actually feel strangely comforting. I am used to Wikipedia looking that way, and I usually find it useful, what with the awful search engine someone already mentioned.

  32. Mark P said,

    July 17, 2008 @ 5:31 pm

    I vote "present" on this issue, but I did notice that following the link to "recency illusion" yields a mention of "truthiness". There was a link to Stephen Colbert and the statement that although the coinage was attributed to him, the word had actually been recorded in 1824. I really wanted a link to that citation, but it was not there.

  33. Ralph Hickok said,

    July 17, 2008 @ 6:53 pm

    Mae said:
    But there's worse noisy linking in the New York Times and some other papers. I suspect that these generate random links by computer, which often are TOTALLY irrelevant, sort of like puns. For example, if a person's name is the same as the name of a (random) city, a mention of their name will link to a page for that city. I can't provide any actual specifics, but I assume someone is making a collection.

    Yes, the NYT, Washington Post, and others use computer-generated linking. This was being discussed a few months back during Gene Weingarten's chats on the WaPo site and a lot of people submitted examples. One I recall is that a story referenced a woman whose first name was Avis and that was linked to the Avis car rental site.

  34. john riemann soong said,

    July 17, 2008 @ 6:57 pm

    "My solution is to put useful links to things referred to in my postings in a list at the end. (Although I'll still put links in the posting if they are funny.)"

    But then you have the problem of people /not/ reading them. I rarely open lists of links, because the lack of context (e.g. a lack of having links in the middle of a sentence) make the links less inviting.

  35. Sili said,

    July 17, 2008 @ 7:05 pm

    I too like to open links in new tabs (wonderfully simple in Opera) – often I do it in the background to return to them later. Which incidentally means that I much prefer the links to be in the text where it belongs, rather than as a list fo scholarly notes at the end.

    This also means that the 'snap view' in WordPress and a certain papers, the add-on on Wikipedia that allows you to get a look at a linked article by just hovering the cursor over it and the Firefox add-on that turns every word into a link are very much not to my liking, because they become intrusive.

    So … relevant links to other blogs or back to old LL postings are good for instance. But I prefer overlinking (hyper-hyperlinking?) – in some cases it's nice to be directed to a good definition of a concept rather than having to ferret it out oneself.

    Still – style is question of taste, and I hope I've never complained about a lack of links in anyone's posts. It is easy to just highlight a troublesome word and right-click > search dictionary/WP. But sometimes being lead by the hand out into the world of exploration is nice. After all – noöne has to follow the links even if they're given.

  36. Jarek Weckwerth said,

    July 17, 2008 @ 7:08 pm

    I'm decidedly pro-hyperlink, too. I'll second Thomas Thurman (among others, of course): linking is obviously one of the fundamental ideas of the interwebs, and in fact I often feel there isn't enough of it on many websites, rather than the other way round. It's all just a matter of making them graphically non-intrusive.

    What annoys me is links lacking from places where your default expectation is to have them. The extreme case is non-linking URLs, like this:

    You can find more on http://www.wikipedia.org.

    (Of course, Opera, for example, has a right-click option for this, called "Go to web address" and I'm sure there are add-ons for Firefox etc. But the less savvy will not know…)

    Most of the time, this happens when documents meant for printing are put online, but not always. One of my pet peeves, really.

  37. Jahi Chappell said,

    July 17, 2008 @ 7:22 pm

    I think I would say links "shout" on the page if they were STANDARDLY CONNOTED LIKE THIS or something, but not otherwise. I'm not all that arsed about it. On the other hand, Zwicky's examples of things he ISN'T linking to are a series of links I'd actually like to see as a newbie reader and non-expert in the field.

    What I do like in Wikipedia to moderate between the extremes is when a link to a novel or important concept/person/etc. is given the first time it is mentioned in each section or simply once in the whole article. After all, Zwicky is right that I could go search for the different things he's talking about. But just as he is blogging such that his opinions can be heard by a wider audience (I assume), linking makes people more likely to learn about the work you refer to. "Making them do the work," is useful in some pedagogical sense, but that strikes me as a more cranky than practical take on how people actually read web content. I know I often don't feel it's worth the trouble of independently googling something while reading an article or post, but if it's already linked, I'll look at it.

    Indeed, some bloggers *don't* link to some things as a distinct smack-down, to say "I'm responding to this but it is too heinous or insipid to bring a greater audience to." The balance (imho) isn't between whether a reader likely already knows or can easily find what you're talking about or not, but whether the link is specifically useful to understanding or not. I.e. in most cases, year or day links in Wiki aren't useful ("Gee, What else happened in 2005?"), or other periphery concepts (like a link to "audience" two sentences ago), but I may, for example, in my own blogging link to examples of some of the aforementioned "smack-downs" in the area I'm talking about, to illustrate. I'm pretty far from Z's camp.

  38. David Eddyshaw said,

    July 17, 2008 @ 7:26 pm

    Talking of xkcd, surely this neatly summarises the problem:

    http://xkcd.com/214/

  39. Jeff McCune said,

    July 17, 2008 @ 7:56 pm

    Links are the whole point of the web. The Machine is using us, after all.

  40. Evan said,

    July 17, 2008 @ 8:47 pm

    I agree with Arnold. Of all the problems with wikipedia, this is not the biggest one, but it is frustrating because when you see that a certain phrase is linked and another is not, it makes you think that the linked phrase is relevant to the topic at hand. Frequently, though, this is not true—the choice of what to link seems arbitrary in many cases.

    I wonder how editors choose between conflicting links. For example, if a sentence contains the phrase "national anthem of Germany", you could link to 'national anthem', 'anthem', 'Germany', or 'Deutschlandlied', or some combination of these. But the editor cannot link to all of them. What's a fanatic to do?

  41. john riemann soong said,

    July 17, 2008 @ 9:01 pm

    Usually the most specific. If [[Deutschlandlied]] did not exist, one would link to [[national anthem]] of [[Germany]]. If [[national anthem] did not exist, one would link to [[national]] [[anthem]].

    And of course there are always piping tricks, if somehow you felt the need to make two links in one, e.g. [[national]] [[Deutschlandlied|anthem]].

  42. Adrian Bailey said,

    July 17, 2008 @ 10:02 pm

    I disagree on AZ's general point, but some of the linking on Wikipedia (e.g. to dates, years, countries) is silly.

  43. Mark said,

    July 17, 2008 @ 10:09 pm

    Arnold writes: "Link fanaticism is not some accident. As I discovered some time ago, it is PRESCRIBED Wikipedia style." I suppose it is true that Wikipedia style calls for more links than he likes, but it does have a concept of overlinking. The style guide mentions well-known place names as things not to link to, and has a more general suggestion to avoid likes of "low added value."

  44. jagorev said,

    July 17, 2008 @ 10:21 pm

    I too favor the heavy linking – it often leads me down fascinating pathways in a way that a more traditional encyclopedia never could

  45. Karl Voelker said,

    July 17, 2008 @ 10:33 pm

    Many comments have already been made about the value of links in general, so I will refrain from repeating them and instead go after a few specifics from your article:

    This blog and Wikipedia are two distinct websites. Links on Wikipedia are generally predictable: you know that when you click on a link you are going to get an encyclopedia entry about the topic you clicked on. If you already know enough about a linked topic, you ignore the link. In contrast, links on a blog could point to anything, because there is rarely a fixed policy saying what links should exist. This uncertainty makes it harder to ignore links on a blog, because you might be ignoring something worthwhile.

    You argue that even without links, it is fairly easy to follow up on an interesting topic, so providing links is unnecessary. Here's the thing: computers are unnecessary. (You can mail letters, buy books, and do longhand arithmetic.) Given the existence of computers, keyboards, mice and video displays are unnecessary. (You can interact with the machine through a panel of toggle switches and lightbulbs, after all!) Even given all our modern hardware, every software convenience is unnecessary. There is always another way that's just a little bit harder.

    Clearly, the entire purpose of computers is for them to be convenient, and links serve that purpose. Now, I am not saying that the writers of LL should be obligated to provide as many links as possible. After all, it would be quite a chore for you (and, as I previously mentioned, copious links are not as good an idea for blogs as for Wikipedia). The software behind Wikipedia is, however, quite marvelous in that linking to other Wikipedia pages is trivial. So, on Wikipedia, it is worth the minimal effort to provide the convenience of links.

  46. Nathan said,

    July 17, 2008 @ 10:41 pm

    @Adrian Bailey: As mollymooly started to say, the real reason for linking dates (but only the full date, [[July 17]] [[2008]], not just [[July 17]] or just [[2008]]), is so the software recognizes them as dates and outputs them in the format the user has set in preferences. It's a hack, but I like it.

    Yes, some of the linking on Wikipedia is silly. Many editors are trying to improve that.The policy is not merely to link everything you can; it's supposed to be relevant and not common knowledge. So most of the time you don't link hat, lunch, or plastic. But I would think that any article that mentions phonology or clitic should have wikilinks to them, and an anonymous editor seems to agree with me.

  47. Stephen Jones said,

    July 17, 2008 @ 10:43 pm

    It's with short pieces that the problem arises most often. With longer pieces the ratio of links to plain text goes down. The links for countries, days, and obvious other stuff (such as 'history', 'language', should probably be formatted in a separate way from others, but it may well be too late to make the change).

    The reason has nothing to do with search engines.

  48. dr pepper said,

    July 17, 2008 @ 11:05 pm

    It seems to me that there should be link categories that can be turned on or off separately. The most basic category being links chosen by the author as relevant to their thesis, versus links that some well meaning meddler added later.

  49. Nathan said,

    July 18, 2008 @ 12:18 am

    @dr pepper: Your idea is unworkable. Many Wikipedia articles aren't written like that, with one main author, and with small editions by meddlers later. Every article really has a unique history.

    Anyone who wants to see the links in a given Wikipedia article change is invited to make those changes, or at least to discuss them on the talk page. I'm excepting obvious stuff like Arnold Zwicky editing the article about himself; I wouldn't recommend that.

  50. Andy said,

    July 18, 2008 @ 1:44 am

    "[…] wouldn't mention eggcorns without a pointer to the Eggcorn Database, wouldn't mention 'singular they' without links to relevant pages (it's come up at least 25 times so far), and so on."

    I didn't know such a thing as the Eggcorn Database existed, and I still don't know what it is, because without a link, I have no idea whether the first result in a Google search for the term would be the same one you are referring to. A reader also does not necessarily know whether you have ever discussed the singular "they" before, and it certainly wouldn't be worth their time to manually check every topic of interest to see whether it has been discussed before on this blog. Even if they suspected so, they might not know the easiest way to use this blog interface to find all such references.

    By leaving such things unlinked, you are making it hard for newcomers to acclimate, and still slightly inconveniencing long-time readers who may have been reminded of an interest in those things but now have to get to them the "long" way.

    If your page's style makes links radically different from normal text, that's already a problem. If not, then links that a reader is not interested in can be easily ignored. The cost is small, and the benefit to some is significant. Had you linked either of those two items, I would read further about one or both. As it stands, I still won't, because I'd have to go too far out of the way and I don't even know for sure that I'd get to the right place.

  51. Mike E said,

    July 18, 2008 @ 2:02 am

    I don't think linking in an online encyclopaedia should be compared to linking in other types of Web pages, such as in blogs.

    Although it is appropriate to use linking more frequently in an online encyclopaedia, the sentence with the word "PRESCRIBED" might give readers the impression that Wikipedia "policy" is to link everything possible. This is not the case, as can be seen from the Wikipedia guideline at http://en.wikipedia.org/wiki/Wikipedia:Manual_of_Style_%28links%29 — in particular the section on overlinking and underlinking.

    As regards the linking of dates, this may also also be done because the same syntax that creates internal links also permits the date to be displayed in the order "May 26 2008" (American style), "26 May 2008" (British style) or "2008-05-26" (ISO style) according to user preferences.

  52. Joe said,

    July 18, 2008 @ 2:17 am

    It's funny, because when you capitalized all those random words, I kept wanting to click on them as if they were links. Which I suppose was your point, even if my desire to click them undermines it.

    Also, the counter-point to the XKCD version of the Wikipedia entry on wood (which, incidentally, had more than a few attempts at adding a "wood in popular culture" section after that), is the other XKCD comic, showing how someone can wander through a forest of links from one thing to researching many others. This phenomenon has also been seen on TVTropes (though they do not have a "Lesbianism in erotica" page like Wikipedia does, which some people have a habit of wandering towards).

    And while someone certainly *could* Google for, say, eggcorns, most are too lazy. Anyone who has ever answered questions online could tell you that 80% of their questions are in the FAQ, but still asked by people who do not *read* that FAQ, and another 15% or more could be answered by a simple Google query. If you can't figure out the acronym JFGI, though, that might be outside your experience.

    In other words, they're useful if sometimes ugly.

  53. Paul Donnelly said,

    July 18, 2008 @ 2:35 am

    I'm, of course, in favor of heavy linking on Wikipedia. As an encyclopedia the point is to put all [notable] related information on your plate. The rest of us do need to think twice before linkifying words, because outside of an encyclopedia linking conveys a suggestion that you might want to read the link. Worst is when it's an outlet for the author's off-topic free association—really, anyone who wants to read xkcd has already seen that one.

    Dr pepper's idea is a good one, although not workable as is. Coding links by relation to the topic would be a good idea though. Underline links clarifying vague references, overline those you think are interesting but not crucial to understanding, and strike through those that are just so funny I *have* to see it. And no adding "urbandictionary.com" links to the idioms you use. Ever.

  54. Rick S said,

    July 18, 2008 @ 3:26 am

    I want to elaborate on what Stephen Jones said. Because wikilinks are only supposed to be attached to the first use of a term, the bulk of them occur in an article's introduction; they become sparser as you get further down into the meat of the topic. Short articles like Prof. Zwicky's might seem overloaded, but in fully fledged ones the links tend to get out of the way as you get more involved.

    Having all those links in the first paragraph or two is a big advantage when you're hunting for something but you don't know exactly what it is you're looking for, or even just don't know what it's called. Just start with something vaguely related and jump links to narrow down on it. It's like having a big signpost at every turn; there's often no need to read down into the article searching for unadorned mentions.

  55. outeast said,

    July 18, 2008 @ 4:04 am

    I'll add that linking has a major (potential) advantage over googling (or, in WP, using the clunky search option) in that if chosen appropriately the target page will actually contain the needed information. Googling can easily lead to bad information: if I want to find out what some linguistics term means, for example, I'd rather be pointed directly to an authoritative source than google and end up finding 6 conflicting explanations by students, journalists etc who have simply failed to grok it correctly.

    WRT Wikipedia itself, as it is open-access it is inevitable that there is going to be bad linkage in some instances (and some outright silly stuff – I agree that the links to dates are annoying as hell). I generally find the linkage more useful than not, though.

  56. Rob Gunningham said,

    July 18, 2008 @ 6:30 am

    Isn't it possible for those who are distracted by links to click their screens into black and white?

  57. Ellen K. said,

    July 18, 2008 @ 8:02 am

    I don't mind most of the links. But some of them don't link to what you expect. Not an actual example, but they are things like the text saying "Mexico City" and having separate links to "Mexico" and "city", those are the most annying links.

    Or (actually example this time), in the entry for "Blog", the following list where all are links, though "artlog" is a "no entry" link: art (artlog), photographs (photoblog), sketchblog, videos (vlog), music (MP3 blog), audio (podcasting)

    Like, do we really need links for "music", "videos", "photographs", and "art"? That linking to articles for quite ordinary words I don't like.

  58. Rob Gunningham said,

    July 18, 2008 @ 8:42 am

    @Ellen K., there is a discussion of this above, by John Riemann Soong.

  59. Jason Orendorff said,

    July 18, 2008 @ 8:55 am

    Yeah, the whole point of Wikipedia, if not the Internet, is the experience of wandering around in it, following links and learning about stuff you didn't know existed. It has to be richly linked.

    I think you're not understanding about blogs either. How it works is, you provide finely crafted, thought-provoking content for free, and you provide well-chosen links to other suggested reading and primary sources on the web. You're the one who has to "do a certain amount of work" around here. Our role is to nitpick, cavil, and kibitz. Hope this helps.

  60. Nick Lamb said,

    July 18, 2008 @ 9:03 am

    What ought to be happening here is generic links. These are a core concept of Hypertext theory and would make about 80% or so of the links in Wikipedia entirely superfluous, with a further 10-15% needing only an annotation so that the generic link software knows which "Steve Harris" the article was talking about.

    That is, any ordinary user of the hypertext system should be able to highlight a word or phrase, like "Daily Mail" or "linguistics" in ANY document and follow links from that to relevant information, such as a definition or an encyclopedia article. Systems that could do this existed long before the web, but the World Wide Web eclipsed all of them because it was so easy to get started.

    But the web doesn't provide generic linking per se, and web browsers have steered clear of layering on one of the myriad other hypertext systems that could do it. So you lose more than half the power of linking by only allowing links to be embedded into documents.

    You can get systems that work as hypertext was originally envisioned, but Wikipedia can't rely on your having them. Nor, since it isn't aimed at expert users, can it rely on you knowing where to get one, or figuring out that you can trivially guess the URLs of most entries (which is how I mostly use Wikipedia). So explicit links are the only way to achieve their goal of making the encyclopedia accessible.

  61. language hat said,

    July 18, 2008 @ 10:02 am

    I entirely disagree, and I think Nathan has it exactly right. Links are the currency of Wikipedia, which is an entirely different entity from Language Log. You can presume a certain level of knowledge on the part of your readers, or a willingness to do some legwork; Wikipedia does not and should not. The reason links to years are deprecated is that they are in fact useless; links to people, places, and so on are useful and important. Not to you in each case, but it would be the height of arrogance to take your own level of knowledge as a basis for how other people should be treated.

    To give an example of how a high level of linkage can be useful, I was just trying to remember the word for the piece of movable type with a given letter on it. I went to a Wikipedia article involving typography, knowing that if I clicked on the right links I would get to an article that would give me the word I wanted (such pieces of type are called "sorts"). Wikipedia is a resource, not a test or a tease.

  62. kip said,

    July 18, 2008 @ 10:40 am

    @Ellen K.:

    Or (actually example this time), in the entry for "Blog", the following list where all are links, though "artlog" is a "no entry" link: art (artlog), photographs (photoblog), sketchblog, videos (vlog), music (MP3 blog), audio (podcasting)

    Like, do we really need links for "music", "videos", "photographs", and "art"? That linking to articles for quite ordinary words I don't like.

    No, we don't need those links. Those fall under the "common words" category. They have now been removed. (Of course, you could have removed them too… anyone can edit and all..)

  63. Catanea said,

    July 18, 2008 @ 5:57 pm

    Recently my mother-in-law drew my attention to the existence of the film "WALL·E"…
    Televisionless, I had never heard of it. I looked at "IMDB", and at Wikipedia sites. At Wikipedia, I was told that the Macintosh start-up-tone [chord?] was significant in the film.
    There was a link. I clicked on the link – thinking vaguely of Windows users who might never have heard…
    There was a link. I clicked on it, just for some cultural validation – to hear that anyone clicking there would get to hear (perhaps for the first time?) The Macintosh Chord… HAH. It was only a ludicrously general link for those who had never heard of Macintosh. Analogous things have happened to me dozens of times in Wikipedia.

    THAT's my complaint. The Wikipedia links are ONE WORD AT A TIME; and thus don't just transport the clicker instantaneously to THE cited THING.

  64. Aaron Davies said,

    July 18, 2008 @ 6:02 pm

    @John Soong, @Ellen: those are the only type I find annoying. I'd much rather see a red link to, say, "[[Uruguayan Army]]" than two blue links to "[[Uruguayan|Uruguay]][[Army]]" which reveal their true nature only if you hover them and note the lack of a connecting underline.

  65. Matthew Cockerill said,

    July 18, 2008 @ 6:06 pm

    I'm hugely in favour of the extensive internal linking in wikipedia. One of the reasons I am so keen on it is actually a possible accidental side effect.

    What is great about Wikipedia links is that they identify all the important concepts/terms/entities which are related to an entry, and the link associated with that term actually disambiguates the sense in which it is used. This makes Wikipedia as far as I can see one of the most extensive and richly structured interconnected ontologies of concepts that exists. Informal it may be, but it's an amazing resource.

    To give a completely random example, the term Cancer could refer amongst other things to:
    Cancer, the disease
    Cancer (constellation), the constellation
    Cancer (astrology), the astrological sign
    Cancer (genus), a genus of crab

    In Wikipedia, the links make it absolutely clear (even to a dumb computer) which is meant.

    Some work has already been done on mining to create formal conceptual maps from this ever growing informal knowledgespace of concepts e.g.
    http://www.mitre.org/work/tech_papers/tech_papers_06/06_1028/06_1028.pdf

    Ultimately, I think this is where the long promised semantic web is going to spring from.

  66. adam j. sontag said,

    July 18, 2008 @ 11:07 pm

    I ended up at Language Log tonight because I'd been surfing wikipedia for too long and decided to peruse one of my old favourites. After several entries led me back into miniature wikipedia binges, I came back here and found this post. In the comments, where people allegedly frequently link to concerned matters, someone mentioned the xkcd strip "the problem with wikipedia," but didn't link to it. I googled "xkcd problem with wikipedia," and chose the wikipedia article on XKCD, and used a link there to find the XKCD strip I'd been looking for.

    Moral of the story?
    I'm not sure. But I think XKCD is closer to nailing the "problem with wikipedia" than Arnold was in this post. :)

  67. JakeT said,

    July 18, 2008 @ 11:26 pm

    I've always assumed that some of the links on the wikipedia to things like May 25, for example, were semantic jokes, ie, linking to something outlandishly general as a pretend reference for something very specific, the humor being simply that it's not only possible to do, but pretends to be meaningful, subverting the very thing that linking is supposed to do in the first place, namely to provide information.

    How's that for over-analyzing jokes?

    In my opinion, learning to link well, to provide meaningful anchor text that provides good information to the reader about where they'll be going when they click the link without being filler 'click me to do such-and-such' crap is a difficult skill to master.

    It's certainly one more thing to think about while you write, and of course, there's the interesting question about whether or not you go back and write in your links after you write your article, or doing in in process.

    I'm sure people have gone over this all elsewhere, but I think writing for the web raises a lot of interesting semantic issues like this.

  68. Martin Wisse said,

    July 19, 2008 @ 8:47 am

    I actually agree with Arnold that his own entry on Wikipedia is overloaded with links, but that's largely because it is such a small entry. Five links in an entry of five lines looks awful, but aren't particularly noticable in an entry of fifty lines.

    More generally, a wikipedia article should have an introductionary paragraph giving a general overview of the subject, which due to their nature almost always are incredibly link rich, which makes reading them more difficult in my experience. Especially when in the same paragraph you have another five citation links as well.

  69. James Wimberley said,

    July 19, 2008 @ 11:44 am

    Why complain? You can't fight Wiki Hall.

  70. Anonymous Cowherd said,

    July 23, 2008 @ 6:01 pm

    @Catania: "THAT's my complaint. The Wikipedia links are ONE WORD AT A TIME…" No, they're not. You can link multi-word phrases with the same syntax, e.g. [[English language]]; and you can even link parts of words, if you're so [[incline]]d. And for the really interesting cases, you can [[pipe link|pipe]] words to keep the visible text brief while still linking to the appropriate [[dictionary entry|entries]]. Most of Wikipedia is Done The Right Way; you must have stumbled on some poorly wikified backwater. Hey, it happens.

RSS feed for comments on this post