Wikipedia talk:Manual of Style/Linking/Archive 6

This is an archive of past discussions about Wikipedia:Manual of Style. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 1

←

Archive 4

→

"plain English words"

Damn edit summary field, why is it so small? Considering that all readers of the English Wikipedia can use a web browser, but probably fewer than half are native English speakers, "plain English words" such as "custard", "levee", "meadow" or "glowworm" are generally less likely to be understood than technical words such as "byte" or "operating system", and hence (provided that neither are directly relevant to the topic of the article), the former are more likely to need linking. So I've deleted that point; saying "terms whose meaning can be understood by almost all readers of the English Wikipedia" makes the intent clearer. --A. di M. (formerly Army1987) — Deeds, not words. 18:44, 4 July 2009 (UTC)

No, this is a MAJOR change and needs to be discussed here first. I strongly disagree with the change. Tony (talk) 08:59, 5 July 2009 (UTC)

in my opinion the previous format (the list of bullet points) was a lot clearer, and the point about not linking plain English words definitely needs to be kept. it's not the aim of wikilinks to serve as a dictionary-substitute for non-English speakers. unless the article on custard somehow sheds some useful insight on the article they're reading, it shouldn't be linked. Sssoul (talk) 10:02, 5 July 2009 (UTC)

(edit conflict) Feel free to revert it, but I do believe that the phrase "plain English" is way too vague to be of any use.

As for the merging of bullets, I honestly can't see the difference between "terms whose meaning would be understood by almost all readers" and "items that would be familiar to most readers". (Unless I have to interpret the latter literally, that is the items must be familiar, not the terms referring to them; in that case, I really can't see how it applies to anything. While almost all readers will be familiar with the name "New York City", very few of them will be familiar with the place it refers to; matter of fact, most of them won't even have ever been there. I don't think this is the interpretation which is meant. So if it's the term which must be familiar, the former bullet is more explicit. Also, if we must interpret it littterally, "most" means "more than 50%". But I think we don't want to discourage linking an article about something 40% of readers have never heard of.)

As for the list of common units, it seemed to be pulled out of someone's arse. The pound and the volt are way more familiar than the millisecond or the foot per second. It is way more useful to state the principle that to make a arbitrary list. --A. di M. (formerly Army1987) — Deeds, not words. 10:05, 5 July 2009 (UTC)

"Plain English" as opposed to elaborate technical terms that are little-known outside the immediate area, such as "interdigitation" (one I spotted unlinked to Wiktionary in a recent FAC). Take a look at the French WP to see common-word linking gone crazy. "Farmer" is linked in a country article I just read. And every word you can poke a stick at. It destroys the utility of wikilinking by diluting it massively, and looks messy and unprofessional. Tony (talk) 13:01, 5 July 2009 (UTC)

But words such as "farmer" are covered by "terms whose meaning would be understood by almost all readers". As for "plain English", "water" definitely is a plain English word, and "cohomology" definitely isn't, but what about "spark plug", "chlorine", "octave" or "falsetto"? Consider the sentence, "In addition to using his home-made guitar he prefers to use coins (especially a sixpence), instead of a more traditional plastic pick, on the basis that their rigidity gives him more control in playing." One interpreting the phrase "plain English words" too broadly might remove both links, but there's a surprisingly large number of non-musicians who don't know what a pick is,^* and most people outside the UK have never heard of sixpences. Writing "terms whose meaning would be understood by almost all readers" makes clear that the links should be kept.

* Or, at least, there's a surprisingly large number of non-musicians in Italy who don't know what a plettro is; but I can't see any reason to expect the situation is any different in English. --A. di M. (formerly Army1987) — Deeds, not words. 13:52, 5 July 2009 (UTC)

Yes. Here's an idea: no one has ever gathered a set of examples—not individual words or even sentences, but selected paragraphs and the articles from which they are drawn—that enable editors to acquire a good sense of where to draw the boundary on the intersection between relevance and technicality. I'd be willing to do it as an essay, and if it worked well and there was consensus here, it could possibly be turned into a sub-page or other WP page, linked from a footnote in the style guide. Do you see this as useful? Are you interested in collaborating on it? Tony (talk) 14:56, 5 July 2009 (UTC)
The idea is good: a link can be perfectly useful in a context, perfectly useless in another, and distracting or even confusing in another still. But I don't think I will collaborate on writing the essay (unless I get insomniac). --A. di M. (formerly Army1987) — Deeds, not words. 16:10, 5 July 2009 (UTC)

I think realistically speaking it will be necessary to mark some borderline expressions as optionally linkable, to prevent perpetual fighting. Otherwise I think it's an excellent idea. Thanks for the initiative. Hans Adler 20:50, 24 July 2009 (UTC)

This is part of a larger problem, and the entire section should be reviewed for relevance and appropriateness. Unfortunately, this problem began last July when much of the text was changed without discussion; the intent of the passage was unilaterally rewritten to delink "common words", reversing the previous direction. It has evolved since then in fits and starts through sporadic discussion, but suffers from the way in which it was created (having to be "pulled back" from the undiscussed rewrite). The guideline is now being used as justification for a script-based mass-delinking of countries, cities, and other terms that some editors have deemed to be "common knowledge". (This, despite the lack of consensus to do so, and the lack of consensus as to what should be linked and what should not.) Should we link everything? Certainly not - but what is happening now is that terms (such as Canada, New York and the like) are being systematically delinked regardless of context, often leaving a mess of linked and unlinked terms. --Ckatz^chat _spy 22:59, 24 July 2009 (UTC)

I think the matter is really one of statistics. If we were too liberal about the linking, book (100,000 views/month) could easily get a million more incoming links than DNA (250,000 views/month). That wouldn't be OK. It's reasonable if book has 100 times as many links as DNA, or a hundredth, or something in between. (What we have in fact is a ratio of roughly 2:3, which is pretty good.) To enforce this means basically that the odds for a single instance of the word book to be linked must be dramatically lower than those of DNA. Of course it doesn't mean that book should be orphaned. But have a look at the incoming links] to book and you will see there is still a lot of insanity there, such as:

"The term dragon tile is a Western convention introduced by Joseph Park Babcock in his 1920 book introducing mahjong to America." The incoming links from bible, Project Gutenberg, printing press, incunabulum, mass media are of course perfectly fine. Others such as poetry, Hungarian language and Book of Daniel are somewhat borderline. But links from Detroit Lions, Finland, ZX Spectrum or Iron(III) oxide are unlikely to make any sense. Hans Adler 00:13, 25 July 2009 (UTC)

Somewhere we have to draw the line between nuisance links that go to general articles of inconceivable utility to a reader ("Australia" and "Canada" in the article "Anglican Communion", the two that Ckatz insisted on relinking), and links to lesser-known geographical places. Where these items occur in isolation in the running text, there should be no problem—and it is clearly up to editors to agree on where the boundary should lie (link "Ghana" but do not "South Africa"? This is probably the case, but still depends). Almost all cases of links to plain country-names can be substituted by a link—whether piped or not—to a more specific article or section. I put to Ckatz that if he's really really upset about losing the links to those two obscure countries (Canada and Australia), then why not "Religion in Canada" and "Religion in Australia". I've had no response on that, which is strange, because he's been posting on my talk page since. I note also that there are multiple links involving those two countries elsewhere in the article—a link to the entire country articles, I'd have thought, is quite unsuitable. It's just as well we have the guideline here to steer editors towards either (1) removing redundant links, or (2) tightening them up so they go somewhere at least vaguely relevant. Tony (talk) 02:10, 25 July 2009 (UTC)

PS And I've just noticed that GoodMorningWorld has reverted Ckatz's relinking of such obscure terms as "Europe", "France" and "Spain" in the article on Andorra, here. Let us hope it remains stable so that readers are drawn towards more useful links. Europe, I ask you? Why not "Earth", then? Perhaps we need a linked "human" in case readers need to apprise themselves of what species the inhabitants of Andorra are. And "sky", linked of course, since that hangs over Andorra. Tony (talk) 02:16, 25 July 2009 (UTC)

PROPOSAL: Here's a think outside box suggestion. What if we simply not worry about whether or not SINGLE WORDS are commonly-known or not? Don't link any of them. If the reader is reading something that mentions ytterbium or homiletics, say, why not respect him/her and let them figure out that they can copy and paste those words into the search box if they're curious about them?. Thus we decide that merely saving people from copy/pasting an exact single word that has a Wiki of the exact same name (or a redirect that gets to one), is NOT reason enough to link it. And refusing to do any of this single-word linking for only this reason, not only cuts way down on total inks, but gets us out of some of the perennial argument about whether or not a word is "commonly known" enough to link to a wiki of exactly the same name (which of course is an unanswerable question, because it depends entirely on the reader, and only invites wars about background knowledge). Let's lose all that. Do we really want to write an en.wiki enclopedia for many English-speaking peoples of all backgrounds and places, and have to fight that "background knowledge war" forever?
I suggest a provisional solution of NEVER linking single words that link to (or would link to) a Wiki "as they are." That is, I suggest that any words that could be blue-linked with only the addition of brackets around them, probably should NEVER be!
Okay, so what remains? This means you would, under this policy, ONLY link phrases (or forms of words) which might be missed as having a wiki under that name. It's helpful to know, for example, that there's a whole article on the gunfight at the O.K. Corral by that name. Other helpful links will be hard to make without a pipe, and some of these pipes will look like the dreaded "easter egg link," where it looks like a common word is linked, but it's not. It is (and with this policy you can be more sure that it is) really a piped-link to the exact Wiki you didn't know you were looking for by that name, or weren't sure. For example, when you talk about the model of the atom that was the theoretical product of Niels Bohr, passing your cursor over the first piped-link tells you that there's a specific Wiki on this subject, and the second two-word name link is a reassuring reminder that you can get to the biography of the physicist using this two-word form of his name (even if THIS indeed can be exactly linked, without need for a piped link). And as before, we use piped links to avoid dab pages, so that if you're talking about a particular use of a word, the link gives you the proper Wiki without the dab.
So, to sum up: the policy on red links would not change. The provisional rule for blue-links is that: all blue links should be pipelinks, easter-egg links, and multi-word links, and further that we forget about all the single-word blue-links that COULD be linked with just brackets (if we wanted to) to go to a Wiki of the same name (or a redirect to it). And that we do this for linkable single words that are names for Wikis no matter how obscure these single-words are. That will save us a lot of links and a lot of argument, all at the same time. S B H arris 04:19, 25 July 2009 (UTC)

Disagree that plain English words such as "custard", "levee", "meadow" or "glowworm" need linking. Down that path madness lies. It is quite easy for anyone who is confused by such words to enter them into either WP's search box or into any number of dictionary sites on the internet. The suggestion that WP should provide a link to assist with the understanding of words such as "meadow" is bizarre and must be resisted. HWV258 05:09, 25 July 2009 (UTC)

Sbharris, thanks for your proposal. One comment: I'd advise against piping "Bohr model" to "model". Using the article name "Bohr model" is more specific, whereas "model" could be ... well, a lot of things. Since the inventor of this model, Niels Bohr, is embedded in that article title, and is linked prominently at the start of that article, there seems little point in sending readers on a fox-hunt by providing both links (a fork, as it were). Every link carries a slight cost in dilution, so there needs to be a reasonable value in having two rather than one. Tony (talk) 05:31, 25 July 2009 (UTC)

I suspect he was sarcastic... In case he isn't: having to hover on links to understand where the hell they go is a nuisance for people who can do that, and an accessibility issue for those who can't (see WP:ACCESS#Text, bullet 4). Piped links are OK for plurals/verbs/adjectives which can't be obtained by slapping letters at the end of the singular noun, e.g. [[atomic nucleus|atomic nuclei]], [[hadronization|hadronize]], or [[force carrier|force-carrying]], but anything more than that should be done very sparingly, only when all the ways to reword a sentence to include the linked article title (or trivial variations thereof) would be too awkward, and the link target wouldn't be too confusing or surprising. On seeing "model" in a physics article, I'd expect it goes to an article about models in physics in general; being taken to one about one particular model would surprise me.

BTW, if Andorra should not link to Europe, ~~I wonder what should. Even if I agreed on a hard-and-fast rule that no article can have more than 250 incoming links, Andorra would likely be one of these.~~ If you think that all readers know so much about Europe that none of them will ever learn something new when reading the article, you might as well nominate it for deletion. And as for "linking words is pointless because readers could look them up on dictionaries", that'd be true if all languages had the same words, but that's not the case. Italian doesn't have a word for "causeway", so if I look it up on my dictionary, it is translated as strada soprelevata, literally "superelevated road". I would still have no idea of whether a causeway is the same as a viaduct, or of which differences there possibly could be between a causeway and a viaduct.[Never mind, the fourth bullet in "What generally should be linked" already says that. --A. di M. 19:37, 29 July 2009 (UTC)]

In any event, the decision of whether to link something must be made on a case-by-case basis, because there will be situations which a guideline can predict; it can depend on considerations other than which article is linking to which, such as in which sentence the link is in. If someone asked me whether quark should link to gold and I had never read the former, I'd say, "Hell no! Gold has nothing to do with quarks (besides being made of electrons and quarks—which is true of pretty much anything made of matter), and everybody knows what gold is." But then you read a sentence such as "It had a mass much greater than had been previously expected—almost as great as a gold atom", and when considering to remove that link you realize that very few readers know how heavy a gold nucleus is. Trying to write a guideline encompassing this and even rarer cases would be a wasteful exercise, and the product would be something nobody would bother to read. --A. di M. 11:01, 29 July 2009 (UTC)

P.S. On reading Andorra, I've seen it mentions western Europe. That'd be a more specific link than "Europe", and in turn it links to it in the first sentence, as it should. But the "what if only 250 incoming links per articles were allowed" test would be passed by all the links Ckatz added and Goodmorningworld removed, except at most two or three of them. --A. di M. 11:12, 29 July 2009 (UTC)

Why not link to "Earth", and let's not forget that Andorrans are human. Somewhere, the obsessive linking has to stop, and it should be at the point where few readers would want or need to follow a link. If a reader in English doesn't know where Europe is, they should, frankly, get a life. There's nothing stopping them from typing it into the search box if they are that ignorant. There are plenty of other links in the vicinity that we do not want to dilute. This is a slippery slope to the "Build the Web" notion that linking should be maximised to maximise the Internet's cohesion; on the contrary, it must be balanced with the "sea of blue" problem. Every link carries a slight dilutionary effect, such that this must be balanced against utility. I note also that "France" and "Spain" are linked a second later as well. What, exactly, in any of these three articles, is going to increase the reader's understanding of Andorra. There's a map at the top already. Tony (talk) 12:33, 29 July 2009 (UTC)

Oh, and I forgot to say that the "orphanage" argument is very weak indeed. Link to "Europe" when the content of the article "Europe" is relevant to the reader's understanding of the topic. Possibly the articles on the other six continents might link to Europe; and many more. But the individual country articles don't seem to benefit from such a link, since the article on Europe, or Western Europe, contains nothing identifiably useful. If you believe it does, say what, please. Tony (talk) 12:36, 29 July 2009 (UTC)

Not to mention subarticles such as History of Europe. Dabomb87 (talk) 14:14, 29 July 2009 (UTC)

Even Britannica Online and Microsoft Encarta, which have a link density perhaps one tenth of that of Wikipedia, link "Europe" from "Andorra".[1][2] And they are written by people other than their readers, so there's no downside to having less than one person per year reading their "wheel" article; whereas we have many vital articles of classes C and lower, for if no-one reads an article no-one can improve it. The cohesion of the web and the sea of blue are correlated issues, but not identical: linking "Europe" the first time it occurs in Andorra allows readers to navigate through the site more easily while worsening readability much less than most of the links in Irish phonology#History of the discipline, which don't help building the web as they go to the same article. I understand the point about the sea of blue (the Italian Wikipedia is much worse), but the solution is not to pick a term such as "Europe" and unlink it from as many articles you can for the sake of unlinking it; instead, try to reduce the myriad useless links to United States dollar or euro by 90%; or, only looking at "Andorra", there are more useless links than "Europe" (e.g. to "Barcelona" piped as "Barcelona city centre" and to "Barcelona Airport" in the same sentence). As for your straw man argument, Europe links to continent which links to Earth, which in turn links to human, so omitting those links from Andorra doesn't break anything. And, for what it's worth, the content of the article "Europe" isn't "relevant to the reader's understanding" of Africa or Oceania, either. Indeed, if a link is "relevant to the reader's understanding" in the sense that you assume that the reader has to follow it in order to be able to understand a sentence, then that sentence should be rewritten in a more accessible language, per WP:NOT PAPERS; if I understand you correctly, if all articles followed WP:NOT PAPERS there would be no need for wikilinks at all. (And if you think that the article Western Europe contains nothing identifiably useful, feel free to nominate it for deletion, or to fix it.) --A. di M. 17:59, 29 July 2009 (UTC)

Why the tiny font, are you trying to ruin my eyesight? --Goodmorningworld (talk) 19:26, 29 July 2009 (UTC)

To mark up a minor point which you can skip if you are in a hurry without missing the gist of my post. (If you are not in a hurry and you want to read it, feel free to zoom in (Ctrl-+ in most browsers), or to copy and paste it elsewhere; I don't mind.) :-) A. di M. 16:08, 30 July 2009 (UTC)

← A point which, IIRC, someone made in the "Date delinking" ArbReq is that, while we argue about whether readers find some links helpful, harmful, or neither, no effort has been done to actually determine that. Someone once proposed to make a survey which would be advertised on the site banner (so to all readers, not just those who sign up and use their watchlists). Even multiple-choice questions, whit a piece of software counting the number of persons picking each answer, could be adeguate for a start. (A more utopistic idea could be implement a feature in Mediawiki counting how many times each link is followed. As a bonus, that would also solve the discussions about whether a term has a "primary meaning": if there's a disambiguation and a majority of persons who land there follow the same link, then that term is the primary meaning, the one most readers will be looking for; OTOH if there's an article and a large percentage of readers follow the hatnote link to the disambiguation page, then the topic of the article is not a primary meaning, and the disambiguation can be moved there.) Does anyone like the idea? --A. di M. 16:08, 30 July 2009 (UTC)

- I would love to see scientific research into the hit-rates of wikilinks, and if the Mediawiki feature were available, it would make a valuable PhD project. However, this is unlikely to happen any time soon. What I want to know is why more focused links cannot be found for "Europe", "Spain" or "France", and for someone to point out exactly what information in those articles is likely to be useful to most readers. This has not yet been done. I note that "French people" has been relinked further down: why this as well? Why don't we all migrate to the Italian and French WPs, which have virtually no guidelines about internal linking, and are a total mess because of it? Tony (talk) 16:29, 30 July 2009 (UTC)
  Isn't the point of sections, summary style, etc. the fact that different readers will find different information useful? --A. di M. 19:53, 30 July 2009 (UTC)

I've been a professional webmaster for large companies, and the "let's link everything" argument is very familiar to me -- I hear it once every couple weeks. Basically, pragmatically, a page only gets so many "click-throughs", maximum. After about 10 links, readers do not click links more often. This is true even if you do things such as write in large red letters "PLEASE CLICK HERE". So, generally, a page needs to pick the top 10 most important things to link. On a typical page in a knowledgebase, having 10% of the readers click-through is exceptionally high. 1-2% is more typical.

Knowing the above, excessive linking only: a) Makes articles harder to read, without any benefit, for 90%+ of readers, b) excessive linking distracts the remaining readers from the genuinely important links, c) there is Wiki editorial overhead adding and maintaining the links.

Yes, there are exceptions, such as articles about historical events, where dozens of people might legitimately be linked. However, the question still has to be asked. How many typical readers, in a history article, aren't going to guess that there's a Wiki article on Henry VIII? Or Genghis Khan? Or Winston Churchill? Or their children? And if they guess wrongly, how much time is lost in a search? 30 seconds? In some ways, the search is better, because it may provide a range of related topics in several articles -- not just the single one that is linked.

Website reporting tools analysis suggests that the utility of inline linking is vastly, vastly overrated. It's more akin to a cutsie technical trick than a tool used, or especially used effectively, by readers. Regards, Piano non troppo (talk) 16:42, 15 August 2009 (UTC)

rm hard copy argument

It is true that the intuitiveness principle helps users who read Wikipedia's articles in hard copy. However, Wikipedia is not a paper encyclopedia, and writing for the paper reader would have significantly different requirements, especially for links. (Cf. e.g. "This also means you do not have to redirect one topic to an equivalent topic of more common usage".) — Sebastian 16:57, 1 August 2009 (UTC)

While not the strongest argument, it is a valid argument. We shouldn't optimize it all for printing, but we shouldn't include information which is completely lost in print, either. (It is completely lost to readers who can't hover on links, for example ones using a cell phone, too.) I see piped links akin to the one in the Bowdler example all the time.[3] (Also, the example with Cyrillic ya would me much less bad in an actual article, which makes it much worse as an example. The reader would still be able to understand that the link takes to an article about a Cyrillic letter looking like a reversed R, and that the epiglottal trill is transcribed with a similar symbol. The Feynman example was even worse, because someone not hovering on the link could not even possibly suspect that it didn't take to the "Particle physics" article; I'm going to add it to this page. --A. di M. 16:41, 6 August 2009 (UTC)

Yeah, the Feynman example is much better! That means, we can delete the "ya" example. Re. paper: You are right, of course, that it is a valid argument, but it not necessary here. If editors would just keep in mind the simple concept of intuitiveness, much would be gained already. Adding the requirement to consider paper prints adds another level of complication. Unless this is counterbalanced by a huge improvement, it is instruction creep. I'm not even sure there are many cases where links that follow the guideline so far would not be appropriate for paper, but even if there are, they would require careful examination, since by default the policy WP:NOT trumps this guideline. — Sebastian 17:53, 6 August 2009 (UTC)

"See also" section

I also removed the recommendation to use "See also", which is hardly a way to improve intuitiveness.In fact, "See also" links are often among the least intuitive links because they do not provide context. — Sebastian 17:14, 1 August 2009 (UTC)

please restore the "see also" recommendation and let's discuss it here to see what the consensus is. thanks Sssoul (talk) 05:46, 3 August 2009 (UTC)

I disagree strongly with the removal of the "See also" suggestion. For links that are on the boundary of relevance/utility, it is often a good place to put them. Tony (talk) 09:09, 3 August 2009 (UTC)

Thank you for asking so politely; I am restoring it for now. I see that this may have some use as a temporary solution until someone else either deletes it (when the "see also section gets too big) or fleshes it out to a new subsection. But would you have some good real-life examples? I feel that, to merit being recommended in a guideline, we need to be sure that it really improves an article enough to counter the problem of reduced intuitiveness. — Sebastian 16:21, 3 August 2009 (UTC)

thanks for restoring that - as you note it's not so much a recommendation to populate "see also" sections, but rather an option that editors might consider instead of eliminating "borderline-relevant" links like (for example) a list of events that happened in the subject's year of birth, or your idea (below) about "you might also like"-type links. Sssoul (talk) 09:10, 5 August 2009 (UTC)

Funny that I contributed an example myself! I didn't actually mean we should do it like Amazon and add it to the bottom of the article, but I now see that it may be useful. I'm still not happy with the text as it stands, but for now I would rather focus on #Structure and layout. — Sebastian 15:16, 5 August 2009 (UTC)

"Piped links should never be used to introduce tendentious subtext,"

Resolved

I think this should be added back. I know of at least one instance where piped links are being used to change the context of wording and I referred to this clause. Ward20 (talk) 23:15, 2 August 2009 (UTC)

Thank you for bringing this up. I agree that this happens, and I only removed it because I believe it is possible to fight this without a specific rule for that case, and because in my experience, this sort of distinction often backfires: When someone is accused of "introducing tendentious subtext", the conversation can easily go down a spiral of personal attacks. Would you mind giving us the example? (If you want to keep it confidential, you can also write to me; I have some experience as a mediator in tendentious edit wars.) — Sebastian 23:47, 2 August 2009 (UTC)

Thanks for the offer. Alternate opinions and pointers on the matter would be appreciated. Medically unexplained symptoms (MUPS) is being piped or directly used to describe symptoms in many illnesses of unknown causation by one editor. The MUPS article is mainly undeveloped. In the medical literature some authors use MUPS to refer to many different things.[4] There appears to be no official DSM, ICD or MESH approval of the term, so its use to describe symptomatology is controversial. MUPS has also been added to Fibromyalgia, Gulf War Syndrome, Chest pain and other articles. Other editors also noted and discussed how the term was being spread throughout multiple articles.[5] MUPS was piped as easter egg links and I reverted them explaining with an edit summary.[6][7] There is another buried MUPS link here. Disclosure: there was a recent ANI and is a current RFC concerning myself and the other editor. Ward20 (talk) 07:27, 3 August 2009 (UTC)

OK with me; but could we have trial runs here first? Anyone like to suggest? From memory, the examples weren't all that convincing before. Tony (talk) 09:07, 3 August 2009 (UTC)

Waiting for Sebastian to give an example of how best to deal with this without a specific rule before suggesting an alternative. Ward20 (talk) 19:09, 4 August 2009 (UTC)

I think you touched on the easiest solution already in your message of 14:05, 4 August here: You could simply write "typified by [[unexplained symptoms]] ...". Unexplained symptoms is currently a redirect page to Medically unexplained symptoms, which is in accordance with our recommendation to use redirects. If you like, you can change that page to a disambiguation page, with an "any ..." clause in the first bullet, as I did in Sacred Concert.

"[[Medically unexplained symptoms|symptoms]] ..." clearly already violates WP:LINK#Link specificity, because obviously not every symptom is a medically unexplained symptom. I thought that was crystal clear from the sentence: "When you use a link such as [[Archery at the 2008 Summer Olympics|Archery]], the reader will expect this link to go to a general article on Archery, rather than to Archery at the 2008 Summer Olympics." If that isn't clear enough, then maybe we need to reword that sentence? — Sebastian 22:04, 4 August 2009 (UTC)

I'm moving the replies in their own section, since they go far beyond the discussion of this particular sentence. It seems to me that this particular discussion has been resolved. Please remove the resolved tag if that was in error. — Sebastian 15:07, 5 August 2009 (UTC)

Structure and layout

(This was originally a reply to the post of 22:04, 4 August 2009 above)

Actually I believe it could be better organized. Link specificity, Piped_links, and Intuitiveness, all seem to cover bits and pieces of link specificity yet they are spread out. For example, the "[[Archery at the 2008 Summer Olympics|Archery]]" example isn't really in the WP:LINK#Link specificity section.

Other things were difficult for me to follow including the many different ways the links are used and color coded. For example I would have done it this way, Always link to the topic that is specific enough in the context from which you link. Examples: Link to "[[film actor]]" instead of "[[film]] [[actor]]". In the article about Mozart, link to " [[Requiem (Mozart)|Requiem]] " instead of "[[Requiem]]". (This second example uses a piped link - see below.) Giving link examples as well as real links in the same proximity is confusing.

Some of the color coding seems inconsistent too. "For example, link to "the flag of Tokelau" instead of "the flag of Tokelau"." and "Example: In the eighteenth century, pitches were not standardized (see History of pitch standards in Western music)."

Examples of piped links are talked about before Piped_links, are introduced.

Some of the examples are so obscure (for me anyway), [[flag of Tokelau]], [[greengage|greengages]], [[truant|playing-the-hop]], and The epiglottal trill is sometimes written with a reversed "R", which looks like the Cyrillic letter [[?]])., that it detracts from understanding the example because I am unfamiliar with the content.

I hope you don't think I am nit picking but I really did have problems with this and more, but this is a start. Ward20 (talk) 05:54, 5 August 2009 (UTC)

i agree that the material wants to be organized better - i felt like the current structure was imposed prematurely, and that a number of suggestions and questions in the discussion at the time got overlooked. that's the tree i was trying to bark up in the section above called "back to that proposed restructure". among other things, the entire current "General principles" section is about internal links, so why is it not part of the "Internal links" section?? and so on. Sssoul (talk) 09:21, 5 August 2009 (UTC)

I agree, too, and I'm happy that we seem to have an intelligent, good willed team here to solve that. From the above, I see the following items that need attention:

Organization: Intuitiveness is in section "piped links" although it covers specificity. This is because, when I reorganized it, I felt that I couldn't talk about piped links before having introduced them. (But I wasn't consistent at that, either, as you point out.) Currently, they are introduced in the opening paragraph of § Piped links, which in itself seems very appropriate. If we wanted the introduction earlier, then we would have to either rip it out of its section, or move the whole big section before the principles. How do we solve this dilemma?
Link display: Yes, that needs to be unified! I prefer the "green [[nowiki]] syntax". What do others prefer?
Obscure examples: There is a tradeoff between simplicity and realism. (When I wrote the "epiglottal trill" example, I already simplified it from the real world example!) I agree that this, as well as the "greengage" and "playing-the-hop" examples, are distractingly obscure and propose we replace the trill and the greengage, and just cut the hop. (For the mechanics behind word borders in links, I think we have a page already that describes it; I can't find it off the top of my head, but we should add a link to it instead of the extra examples.) But I like the Tokelau example; it's not obscure because everybody understands that even little known places may have their own flag. And the point of it is to remind people that it's possible that even for the flag of a place many may never have heard of, we may have an article!
Internal vs General: It seems to me that practically everything that can be said about internal links also holds for links in general, and vice versa. For that reason, I would like this whole page to be about these internal/general issues, an keep issues that only apply to external links in WP:External links. I brought this up at WT:External links#Merge from WP:Linking, and there seems to be no objection to removing the external links coverage from this article. I think that should resolve Sssoul's concern, and we would be free to arrange our content by other considerations, such as the one mentioned in item 1 above. — Sebastian 15:07, 5 August 2009 (UTC)

1. Will take some thought, I have some ideas but want to read the sections over multiple times during a couple days to see if I come to the same conclusions. 2. "green [[nowiki]] syntax" is good. Standardization is more important to me. Someone might come along with a better idea later as they often do. 3. Agree. 4. Agree. Ward20 (talk) 17:44, 5 August 2009 (UTC)

Green nowiki syntax

Above, two users agreed that they preferred the "green [[nowiki]] syntax", and nobody objected. This section deals with the details of that syntax: — Sebastian 23:11, 6 August 2009 (UTC)

I like A. di M.'s new example and formatting in the Intuitiveness Piped links section but I believe the the inconstant green coloring is confusing. I tried to improve the coloring but messed up the fromatting. Does someone know a fix for the formatting?

Richard Feynman was known for ... as well as work in [[parton (particle physics)|particle physics]].

Ward20 (talk) 17:36, 6 August 2009 (UTC)

Not sure what you mean. Why are you trying to color some text in black? (It took me some time to realize that the black was not an accident, but something you actively formatted that way.) You said above said you liked green, didn't you? — Sebastian 22:02, 6 August 2009 (UTC)

I was under the impression the highlight green color was to help differentiate the "[[nowiki]] " markup text from actual Wiki links, not various parts of other explanation text. For example In the article about Mozart, link to " [[Requiem (Mozart)|Requiem]] " instead of "[[Requiem]]", rather than, In the article about Mozart, link to " [[Requiem (Mozart)|Requiem]] " instead of "[[Requiem]]". Ward20 (talk) 22:55, 6 August 2009 (UTC)

Black = Explanatory text of this guideline. Green = "quotes" (text taken as example). In your last message, "instead of" is part of the sentence "Link to A instead of B", but not part of the quoted text. Therefore, it should be black. The text "Feynman was known for ... as well as work in ..." was meant to be part of the example text, therefore green. (As I'm writing this, I realize that you may have been thrown off by the fact that the sentence is a bit inconsistent. It probably should read "... as well as for work in ...".) — Sebastian 23:08, 6 August 2009 (UTC)

OK, I was totally mistaken then. I do see what you are saying, but it seems to me there are different (text taken as example)s such as, "[[George Washington]]'s → George Washington's or [[George Washington|George Washington's]] → George Washington's" Some examples happen to be complete sentences and some not. If the convention or consensus is complete sentences are in green and others not I have no problems with it, just trying to get on the same page. Thanks for the patience. Ward20 (talk) 23:38, 6 August 2009 (UTC)

Well, that's what we have talk pages for ;-) The problem was that that whole page was completely inconsistent; it's really hard to make any rhyme or reason of the formatting by looking at the page as it was. I just went ahead and formatted the piped links section as I would like it. I'm sorry, I got a bit carried away; I didn't mean to preclude a decision about where to draw the line. If you ask me, I feel that the easiest line to draw is between explanatory text (black) and sample text (green). Are you proposing to keep non-sentence samples black because the overall impression would be more uniform if the text doesn't switch back and forth between black and green so often? The drawback of that seems to be that the rule is less clear; it would make the layout decision depend on other questions than just whether the text is a sample or not. Maybe we could just copy the two version here below and compare side by side which one looks better. — Sebastian 00:14, 7 August 2009 (UTC)

<Out, I was thinking of only nowiki links being color coded. Examples of only the links being green and purple because I had been experimenting with it:

Richard Feynman was known for ... as well as work in [[parton (particle physics)|particle physics]].

Richard Feynman was known for ... as well as work in [[particle physics]] (he proposed the [[parton (particle physics)|parton]] model).

Richard Feynman was known for ... as well as work in [[parton (particle physics)|particle physics]].

Richard Feynman was known for ... as well as work in [[particle physics]] (he proposed the [[parton (particle physics)|parton]] (model).

Richard Feynman was known for ... as well as work in [[parton (particle physics)|particle physics]].

Richard Feynman was known for ... as well as work in [[particle physics]] (he proposed the [[parton (particle physics)|parton]] (model).

You already did most of the work to make the section more consistant. I am not sure this change is better and it's a lot more work. Ward20 (talk) 01:16, 7 August 2009 (UTC)

Let's not reinvent the wheel here. Our Wikipedia:Manual of Style already employs a standard of green texts; and if it's good enough for the Manual of Style, it should be good enough for this page. The MOS uses the {{xt}} format (the one of the first two lines in your list) for all examples, including those smaller than a sentence and even individual characters - see e.g. here. Let's just go with that. — Sebastian 02:11, 7 August 2009 (UTC)

OK, I'll study the Wikipedia:Manual of Style examples and try to follow those and how you edited the Piped links section. Ward20 (talk) 02:20, 7 August 2009 (UTC)

Which is preferable?

Presently: For example, link to "Icelandic alphabet" instead of "Icelandic alphabet":

Nowiki: For example, link to " [[Icelandic alphabet]]" instead of "[[Icelandic language|Icelandic]] [[alphabet]]":

Ward20 (talk) 23:57, 9 August 2009 (UTC)

What should happen with External links section?

(Split off from previous section. This refers to {{mergeto}} in section § External links.)

i think the inclusion of both internal and external links is pretty crucial to the concept of this page, which was to consolidate information/instructions about these closely related topics. so i'm not at all sure i'm enthusiastic about evicting external links from this article - can we talk that over, please? all i'm saying is that the "general principles" as they now stand should be a subsection of "Internal links", not a separate section as if they applied to both kinds of links. Sssoul (talk) 20:34, 5 August 2009 (UTC)

Sure, let's talk about that first. I had hoped we'd all agree, but if we don't then we need to talk about this before we make any other structural changes.

I'm not sure what you mean by "consolidating information/instructions" in this case. (Let's abbreviate "information/instructions" with "info".) Info for EL is already consolidated at WP:EL. When you work with IL, then you don't need to know anything about EL. All of the info you need for IL is consolidated here. That info also happens to be relevant for EL (at least some of it; or all of it when you consider in-text EL, which are rather rare). But I don't see why we need to mention any details of EL here. All a reader of this page needs to know is that (1) EL exist, that (2) everything on this page applies to them too, and that (3) they can find all EL specific info at WT:EL. — Sebastian 20:56, 5 August 2009 (UTC)

Linking city + region

What do people prefer between

Portland, Maine ([[Portland, Maine]]) vs. Portland, Maine ([[Portland, Maine|Portland]], [[Maine]]

and does it matter if the first form would use a redirect?

Dallas, Texas ([[Dallas, Texas]]) vs. Dallas, Texas ([[Dallas]], [[Texas]])
Atlanta, Georgia ([[Atlanta, Georgia]] vs. Atlanta, Georgia ([[Atlanta]], [[Georgia (U.S. state)|Georgia]])

The context is mostly biographies, where it seems rather unlikely to me that linking the region separately adds anything. If a reader really wanted to go there, it's certainly linked at the start of the page on the city. Comments? Gimmetrow 18:59, 30 May 2009 (UTC)

Yes, I agree (and have said something similar when this question's been asked in the past). Just one link is enough in the vast majority of cases, I'd have thought.--Kotniski (talk) 20:50, 30 May 2009 (UTC)

that's how i view it as well: a single link serves the purpose Sssoul (talk) 06:50, 31 May 2009 (UTC)

Generally, try to focus the readers on a link with the minimum of blue: this makes the wikilinking system generally more effective. Personally, I'd go further by linking only the city, by piping where necessary, since the city article will have the state linked right up top (chain linking). (Dallas, Texas ([[Dallas, Texas|Dallas]], Texas). Try to focus the reader. Tony (talk) 05:10, 1 June 2009 (UTC)

I'd prefer the first style (link city+region as one wikilink), even if it's a redirect. That helps the reader read the region as a qualifier of the city, not as a separate bit of info. Rd232 ^talk 16:05, 2 June 2009 (UTC)

FWIW, I usually use {{city-state}} to get the linking as above. YMMV. --SarekOfVulcan (talk) 17:25, 15 June 2009 (UTC)

Foreign language

Instruction creep and complexity creep! Instead of

You can also indicate the language by putting a language icon after the link. This is done using Template:Languageicon by typing {{Languageicon|<language code>|<language name>}}. For example, {{Languageicon|es}} displays as: (in Spanish). Alternatively, type {{xx icon}}, where xx is the language code. For example, {{pl icon}} gives: (in Polish). See Category:Language icon templates for a list of these templates and the list of ISO 639 codes.

Shall we have:

"You can also indicate the language thus: (Spanish)"

Rich Ffarmbrough, 12:49, 3 June 2009 (UTC).

You have a very good point there, Rich. Tony (talk) 12:55, 3 June 2009 (UTC)

I'll second that.--Kotniski (talk) 12:58, 3 June 2009 (UTC)

I find the {{xx icon}} very useful: it produces a handsome result – bold text but in a lighter color (gray) – which more quickly tells users in advance what to expect before clicking on a link. I would not want to see the {{xx icon}} template deprecated by way of excising mention of it from WP:LINKING. --Goodmorningworld (talk) 09:53, 1 July 2009 (UTC)

Date unlinking bot proposal

I have started a community RFC about a proposal for a bot to unlink dates. Please see Wikipedia:Full-date unlinking bot and comment here. --Apoc2400 (talk) 10:24, 22 June 2009 (UTC)

back to that "proposed restructure" ...

it seems to me that a number of points raised in this discussion here got overlooked when the proposal "went live" and are worth revisiting. wasn't the point about not making serial links supposed to remain in the "general principles" section? and can we adopt Tony1's suggested amendments to the "link density" section, please, and/or keep discussing whether that needs to be a distinct section at all? here are his suggestions for your convenience:

~~Aim for a sensible link density. Do not link eight words in one sentence and then none in the rest of the article.~~ In general, link only the first occurrence of an item. This is a rule of thumb that has many exceptions, including the following:

If where a later occurrence of ~~a link~~ an item is ~~separated by~~ a long way from the first. ~~Avoiding duplicate links in the same section of an article is generally a safe rule of thumb.~~
If where the first link was in an infobox or a navbox, or some similar meta-content. ~~The main text of the article should link relevant terms.~~
tables, ~~entries are another exception;~~ in which each row ~~of a table~~ should be able to stand on its own.

there were some other proposals for reorganizing the text as well in that discussion, but ... well, let's start somewhere! Sssoul (talk) 09:18, 28 June 2009 (UTC)

... so does the lack of discussion mean i can go ahead and make those changes? they seemed to meet general acceptance in the earlier discussion Sssoul (talk) 09:02, 1 July 2009 (UTC)

(outdent) okay, i've made those changes. i also moved examples of what to link and not link out of the Wikipedia:Linking#Link_specificity section, since they're not actually related to link specificity. and i still think it's worth considering moving the "link density" and "link specificity" sections back into "general principles" Sssoul (talk) 06:43, 5 July 2009 (UTC)

Should "PDF" be linked?

In a reference, like the one in List of Kansas railroads#References, should "PDF" be linked, or do we assume people know what it is? --NE2 08:14, 24 July 2009 (UTC)

Use {{PDFlink}}. That way, if we ever decide to change the standard (currently to include a link), all will be changed at once. -- Quiddity (talk) 17:26, 24 July 2009 (UTC)

I think it's a pity that "PDF" has to be blued out. Does the template do that? If so, can it be changed, as many of our date templates have been over the past year? Tony (talk) 05:33, 25 July 2009 (UTC)

I concur. I asked for feedback here. --Andy Walsh (talk) 18:17, 29 July 2009 (UTC)

Inconsistency

The "Plurals and possessives" bullet seems to suggest that [[greengage]]s is more readable than [[greengage|greengages]], but [[George Washington|George Washington's]] is more readable than [[George Washington]]'s. Does "more readable text and source" means something other than the obvious thing, is it a mistake, or am I missing something? --A. di M. 20:46, 6 August 2009 (UTC)

The point here is that in "George Washington's", the apostrophe is parsed as a word delimiter, which means the "'s" is outside of the link. Could be better worded, though. — Sebastian 21:57, 6 August 2009 (UTC)

Particle physics

ADM, one or two examples from that area are fine. Pp articles are likely to be more heavily linked because they contain a higher density of very technical terms. They are unusual in that respect. It may not be necessary, but you might consider making the point that articles (and parts of articles) vary in the appropriate density of links because of their varying technicality. Tony (talk) 02:10, 10 August 2009 (UTC)

Good point. They were the first ones which occurred to me (due to my background...), but I would not object to replacing them (except the one about Feynman, which is the most spectacular example of how not to use a piped link I have ever seen in an actual article). In particular, the one about electron neutrinos is not a very good example, because it's not evident to everybody that an article about them could be written. For the example on proton mass, I had thought hard about an example of a topic which could not have its own article; WP:NOT suggests "oak trees in North Carolina", but for the second half of the example I was looking for something more obviously needing linking than oak or North Carolina. --A. di M. 20:14, 10 August 2009 (UTC)

the "what generally should be linked" section

a lot of the points in this section are really unclear. can we discuss rephrasing it? some of my questions/proposals are:

1] i don't understand this point - does "references" mean "footnotes", or ... ? and what does the second sentence mean?

articles with relevant information, through references (Example: "see Fourier series for relevant background"). Linking items in a list of examples makes them easier to reference as well.

2] i'd like to rephrase the point about technical terms to something like:

technical terms, unless they are defined in the article - but always consider providing a concise definition instead of or in addition to a link to another article. If a technical term doesn't have its own article, an interwiki link to Wiktionary may be the most appropriate.

3] the next point sounds dubious to me in a number of ways, including: what kinds of "confusing usage" are meant? what does "explicit articles" mean? and why is the point about about disambiguation buried here? it should be one of the "general principles".

explicit articles when word usage may be confusing to a non-native speaker (or users of other varieties of English). If the word would not be translated in context with an ordinary foreign-language dictionary, consider linking to an article or Wiktionary entry to help foreign language readers, especially translators. Check the link for disambiguation, and link to the specific item.

assistance improving any/all of that will be very well met - thanks Sssoul (talk) 08:00, 8 August 2009 (UTC)

I've fixed the last two as you suggested, and removed the unintelligible part of the first. Maybe someone who understands what it was supposed to mean can suggest a clearer wording, which could then be re-added. --A. di M. 12:56, 9 August 2009 (UTC)

thanks A di M - the one i numbered 3 remains incomprehensible to me, though. for that one, what i posted up there is not my suggestion - it's what's puzzling me. what kinds of "confusing usage" are meant? what does "explicit articles" mean? and why is the point about about disambiguation buried here? Sssoul (talk) 13:10, 9 August 2009 (UTC)

I think it refers to when an English word can have several meanings and it'd be very hard for non-native speakers to figure out which one is meant in a given context. One example someone once made is "He was shot in the temple", but I think there probably are better examples than that. I believe that, when possible, the solution is to use a less ambiguous meaning, but maybe sometimes there might be no valid alternative. --A. di M. 22:39, 9 August 2009 (UTC)

thanks for clarifying, A di M - i agree with you that creating links doesn't seem like the best way to address that kind of "confusing usage", so maybe the recommendation should be omitted altogether from this "what generally should be linked" section. but if it's going to stay, it certainly needs to be expressed more clearly than it is now. how about:

articles that specify which meaning of a word is intended, such as temple (although writing more clearly to begin with is usually a better solution than relying on links to clarify the meaning).

something like that? but again: i think omitting it is a better idea. Sssoul (talk) 06:55, 10 August 2009 (UTC)

I agree that clarity of linguistic context is preferable. I often point out where readers should not have to hit a link to work out what an item means. The linked article should generally not be necessary to access a basic definition. "RFID" is one example here. Yes, "temple" should be clear from the context (I've never thought of mine as religious in the least). Tony (talk) 10:46, 10 August 2009 (UTC)

A di M, hope you don't mind that i've tried to clarify that "confusing usage" point - the "explicit article" wording really isn't clear at all. the more i think about it, though, the more strongly i feel this point doesn't belong in the "what should generally be linked" section. sentences like the example given should be rewritten, not linked. Sssoul (talk) 06:51, 11 August 2009 (UTC)

The monk who was shot in the temple. The word is not going to mean a religious building, by any stretch of the imagination. This should be moved into the "What should not be linked" section", if retained at all. I think examples will come readily from goming lots of articles—that's how I come across examples for all of my tutorial pages, almost by coincidence. If "temple" is a must, I'd look at the "what links here" for "Temple". Tony (talk) 07:49, 11 August 2009 (UTC)

As currently worded, that point would encourage linking all words with several meanings, even if all-but-one of them are completely absurd in a context (e.g. I can open that can of beer); that's not what we want. I would have retained the point about non-native speakers and ordinary foreign-language dictionaries. For example, without the links a non-native speaker would have no way to understand most of the sentence and in "The Seven-Beer Snitch" his urine test contained "crack, smack, uppers, downers, outers, inners, horse tranquilizers, cow paralyzers, blue bombers, green goofers, yellow submarines, LSD Mach 3" and trace amounts of human urine (but was misread as Homer's test results), no matter how many learners' dictionaries you'd use. (And that's one of the very rare cases where I'd have links within a quotation; after all, that's a quotation of spoken language, so it'd make no sense to state that "the links were not present in the original".) Another example is the verdict of WP:RFAR/DDL, which linked "building a better mousetrap" (but this is a bad example, as I was able to correctly guess what that idiom meant; but this could not be the case with more confusing idioms). The "temple" example is a different one: even the most elementary dict will show both meanings, but the other meaning fits the sentence so well that it might not even occur to a non-native speaker about looking it up; given the high number of completely absurd choice of words in Italian translations of English texts (even by professional translators) I find everyday, which I can only make head or tail of if I guess the original English word and which other meaning could it have, it is not unlikely at all that someone might not realize that "temple" could have another meaning in that sentence. (BTW, "explicit articles" mean "articles about the one specific meaning of the word", as opposed to disambiguation pages.) I'd propose something such as: words and idioms used in a context where a non-native speaker could be unable to determine their meaning, even using an ordinary foreign-language dictionary; make sure the link goes to an article specifically about the intended meaning (or to a Wikitionary entry), and not to a disambiguation page (which would defeat the purpose of it). [Insert really good example here when you find it.] However, rewording the sentence to that it is less confusing, if possible, is strongly preferred. --A. di M. 13:52, 11 August 2009 (UTC)

... i understand what you're saying A di M, but i still feel that an invitation to guess what terms non-native speakers might have trouble understanding and to create links to help clarify their meaning is not something that belongs under "what should generally be linked".

the example with all the drug jargon is a totally different question - plenty of people whose first language is English wouldn't be able to tell you what all of those mean. that kind of jargon/subcultural slang needs to be linked for the same reason as technical terms need to be linked (which could usefully be clarified in the point about technical terms). Sssoul (talk) 14:40, 11 August 2009 (UTC)

update: sorry, i've just elected to transplant the point here until we can work it out, because it truly doesn't seem to be getting better as it goes along out there. here's the last version:

words and idioms used in a context where a non-native speaker could be unable to determine their meaning, even using an ordinary foreign-language dictionary; make sure the link goes to an article specifically about the intended meaning (or to a Wikitionary entry), and not to a disambiguation page (which would defeat the purpose of it).^{[example needed]} However, rewording the sentence to that it is less confusing, if possible, is strongly preferred.

and the last-but-one:

articles that specify which meaning of a term is used, although rewording the sentence so that it is unambiguous is preferable. For example, in A monk was shot in the temple, a non-native speaker unaware of the anatomical meaning of temple might not even suspect that in that sentence it does not refer to a religious building.

i don't feel the MoS needs to prohibit links like this, but i don't think they should be encouraged as "what generally should be linked". Sssoul (talk) 14:49, 11 August 2009 (UTC)

Sorry, ADM, I was being opaque: I really meant that "temple" shouldn't exemplify the "what should", and used the opposite ironically. May I put in a plea that we move slowly on changing this page. I feel that a lot is happening in a very short time. BTW, ADM, your spanking new signature—did you consider making the A and the M white against those dark backgrounds? Tony (talk) 14:55, 11 August 2009 (UTC)

I had gotten your point, but I couldn't find a better example than that; however, with Sssoul's wording the point is clear even without examples at all, so that's moot now. (As for the sig, the code for that would exceed 256 chars, but I think this one is fine.) -- A. di M. 18:23, 11 August 2009 (UTC)

Cross namespace links

Any objections to adding "links from article to user namespace links" as something to avoid in general. I recently came across a "see also" which linked to a list developed on the user's subpage. I removed it, citing WP:SELF but it might be nice to add that here. Any objections, thoughts or discussion? --TeaDrinker (talk) 18:36, 9 August 2009 (UTC)

I have boldly added it, since there did not seem much discussion here. --TeaDrinker (talk) 03:36, 21 August 2009 (UTC)

Doh, already in it. My mistake. --TeaDrinker (talk) 04:07, 21 August 2009 (UTC)

"Link density" et al.

The title "Overlinking and underlinking" is quite different from the replacement "Link density". The density can be high without overlinking, and low without underlinking. I suggest the pre-existing title be reinstated.
"consider linking "price" and "goods" only if these common words have technical dimensions that are specifically relevant to the topic (but since many "single dictionary word"-titled pages are disambiguation pages, make sure that the links are directed to the correct articles);". This is not a good change: if an item is worth linking, it's normally no longer a "dictionary" word, but has technical dimensions in the context.
The example of a "specific" link, to "film actor" rather than "film" "actor" is not helpful: these two items should not be jammed together in the first place, so the guideline now mixes issues. "Film actor", "film", and "actor" are highly likely to be common terms that should not be linked. In fact, they are misused in many many articles, where the guideline says already not to link professions, normally. This example needs a rethink.
"If no such page exists, then you need to link to a more general article". What, so if no article on "Religion in Australia" exists, just bung in a link to "Australia"? No. This is bad advice that countenances trivial, general linking, and should be rephrased.
"If there is any chance that the topic might become an article in the future, create a redirect page to the article as described in section § Redirects." I'm not sure I entirely like the encouragement to fill an article, especially a new one, with red links. Can it be toned down a little? "Any chance"?
"Pound sign" is known as "hash sign" outside North America.
"Let's assume for example you needed a link"—This is too informal a tone for this guideline. It's more like a transcript of a chat or talk.

This is all happening rather fast and without specific discussion. Tony (talk) 07:33, 7 August 2009 (UTC)

I agree and would have reverted the changes on the project page but found the "Revision history" too confusing. --Goodmorningworld (talk) 07:58, 7 August 2009 (UTC)

i concur - can we please back up and discuss the changes being proposed? thanks Sssoul (talk) 08:19, 7 August 2009 (UTC)

Agree – also, this is getting too technical. Remember, this is a style guideline, not a help page. Dabomb87 (talk) 13:47, 7 August 2009 (UTC)

Please feel free to revert my changes. There are obviously enough people here who deeply care about this page, so I'm not needed here. See also my talk page. — Sebastian 06:27, 8 August 2009 (UTC)

I've changed the example from "film actor" to "Icelandic alphabet". (I have actually seen "Icelandic alphabet" in an article.) As for the "more general article", I think it's supposed to address cases such as linking neutrino if you have the text "electron neutrino" (one of the three flavours of neutrinos) and there is no article specifically about electron neutrinos. If you want to make that clearer, do that (I can't find a way right now), but the point itself is valid. In the since many "single dictionary word"-titled pages are disambiguation pages, make sure that the links are directed to the correct articles part, I meant that pages such as curl are likely to be disambiguations, so before saving an article containing such a link, one should actually check it goes to the right place (e.g. curl (mathematics)). I'm going to change it to many pages whose title is spelled identically as a common dictionary word. I'm trying to address the point about "any chance". --A. di M. 16:43, 7 August 2009 (UTC)

Much better example there, ADM (can we call you that? what do you prefer?) I'm still a little uneasy about "specific enough" for a context. It seems vaguer than the previous wording. Tony (talk) 17:48, 7 August 2009 (UTC)

yeah, the Icelandic alphabet is great! i'm puzzled by some of what's now listed under "Link specificity", though - those second and third numbered options are not examples of cases where "no such page exists"; they're examples of when a page does indeed exist but using a piped link or a "see also" parenthetical works better than a plain link. Sssoul (talk) 08:19, 8 August 2009 (UTC)

maybe something like this?

Always link to a topic that is specifically relevant to the context from which you link. Examples: Link to "Icelandic alphabet" instead of "Icelandic alphabet". In the article about Mozart, link to "Requiem" instead of "Requiem". (This second example uses a piped link - see below.)

Check to see if a page for the specific topic already exists - this is often the case. For example, link to "the flag of Tokelau" instead of "the flag of Tokelau".

If no such page exists, the most appropriate link may be to a particular section of a more general article: egs/instructions. Alternatively, you can consider creating a red link, a redirect or adding a parenthetic "for more information, see Neutron".

and (to me) the part about "consider moving links to the see also section" doesn't have anything to do with "Link specificity", so i'd move that point elsewhere. Sssoul (talk) 08:42, 8 August 2009 (UTC)

Indeed, I didn't understand what points 2. and 3. were doing there, either, so I removed them.

(<<=====) This bugs me.

Always link to the article on the most specific topic appropriate to the context from which you link: it will generally contain more focused information, as well as links to more general topics. For example, link to "Icelandic alphabet" instead of "Icelandic alphabet": it will contain more detailed information about the Icelandic alphabet, as well as links to the articles "Icelandic language" and "Alphabet". Likewise, use "the flag of Tokelau" rather than "the flag of Tokelau"; in the article about Mozart, link to "[[Requiem (Mozart)\|Requiem]]" instead of "[[Requiem]]".

link to "Icelandic alphabet" instead of "Icelandic alphabet"
Likewise, use "the flag of Tokelau" rather than "the flag of Tokelau";

Huh? What's the difference?

I made a point of pasting in the above text without the bluelinks that came from Wikimarkup.
Why? Because I knew that some of you would say, "WELL D'OH, MORON, you just MOUSE OVER the Wikilinks -- they're DIFFERENT!!!" But not all readers may think of that. If they are new, they have enough trouble absorbing and processing all this information, we cannot expect that everyone will think of hovering the mouse to discover that in the first example, the entire phrase is wikilinked and in the second, every component word individually. Suggested fix: show Wikimarkup first, then actual wikilinked result.--Goodmorningworld (talk) 12:48, 10 August 2009 (UTC)

that's a good point, Goodmorningworld - you've got my vote if you want to go ahead and fix it. Sssoul (talk) 14:15, 10 August 2009 (UTC)

Okay, I went and changed Wikipedia:Linking#Link_specificity, I hope y'all like it better now. Goodmorningworld (talk) 19:21, 22 August 2009 (UTC)

very nicely done, Goodmorningworld! Sssoul (talk) 20:34, 22 August 2009 (UTC)

Links to Non-English Wikipedia articles

When should one translate (in summary if neccessary) a foreign language Wikipedia article and when should one provide a link to the foreign language? The following case study might help concentrate people's minds:

The Elfstedentocht (Eleven Cities Journey) is a classic Dutch ice skating event. One of the giants of the event was Coen de Koning (who won it twice in the 1940's). Piet Keijzer was another winner from the same era. I checked the number of hits for each of these pages during July 2009. They are:

Elfstedentocht: Dutch - 3384 hits; English 829 hits
Coen de Koning: Dutch - 195 hits; English 158 hits
Piet Keijzer: Dutch - 208 hits (of which 84 were on the first anniversary of his death); English - no Wikipedia entry.

Would is be proper in these circumstances to link Piet Keijzer in the English language version of the race to his biography in the Dutch language section, or to leave it as a "red link"? Martinvl (talk) 11:38, 18 August 2009 (UTC)

Always leave it as a redlink. --NE2 05:08, 20 August 2009 (UTC)

Wikipedia Policy WP:SINGLEEVENT advocates caution about articles concerning people who are notable only in respect of a single event and that due attention should be paid to the significance of the event concerned. The Elfstedentocht is a major event in the Netherlands, but is a minor event elsewhere - how many column inches is devoted to in in English-language newspapers compared to Dutch-language newspapers? As I see it, the following options exist:

Retain red links for ever and hide the fact that a foreign-language text is available.
Remove the links completely in the English-language versions of Wikipedia and hide the fact that a foreign-language text is available.
Create stubs in the English-language versions that automatically redirect to the foreign-language text.
Create stubs in the English-language version with a link to the foreign-language text.
Provide links to the foreign langauge text until such time as an English language article is wirtten.

I favour the last of these options.Martinvl (talk) 07:43, 20 August 2009 (UTC)

Notability is not based on the language of the sources. As for your last option, how, exactly, will you find the links to change? There's no "what links here" for interwiki links. --NE2 07:53, 20 August 2009 (UTC)

I agree that notability is not based on the language of the sources, but rather on various cultures. In some cases there is a very strong correlation between the language and the culture, for example the Dutch language Wiki caters mainly for residents of the Netherlands. However the English language Wiki caters for a number of different communities – the International Community who need a common language, the American community, the British Community, the Indian community to mention but a few.

Getting back to my original argument, if a particular entry would be notable only within a particular community and that community uses a language other than English, would you favour a stub in the English-language Wikipedia which contains the sentence “Please refer to the XXXX language version” and a link in the languages box? Martinvl (talk) 11:21, 20 August 2009 (UTC)

If those people pass the English Wikipedia notability guidelines, an article about them can be written sooner or later, and you should use redlinks until such articles are written; and when written, the English articles even if initially very stubby will have links to the Dutch ones in the obvious places. If they don't pass WP:N, there's no point in having eternal redlinks, or to create stubs. Links going directly to the Dutch articles can be useful, but make sure the reader knows where they go before following then. --___A. di M. 12:28, 20 August 2009 (UTC)

NE2: the Dutch Wikipedia is not bound by the English Wikipedia notability guidelines, nor vice versa. It's entirely possible that someone is notable according to the one but not to the other. In particular, the notability guidelines of a Wikipedia in a language strongly associated with a regional culture (such as Croatian or Japanese, and as opposed to languages such as English of Spanish) might well favour regional topics. --___A. di M. 12:28, 20 August 2009 (UTC)

One option might be a footnote referring readers to the foreign-language article. --Kotniski (talk) 09:45, 22 August 2009 (UTC)

Strong disagreement with encouragement to repeat-link

"(As a rule of thumb, do not repeat a link in the main prose of the article more than once per section.)"—Why are editors now encouraged (that is as it will be taken) to link the same item in successive sections? This seems like a massive invitation to link, link, link, link. If a reader doesn't bother to hit the link on its first occurrence, many editors would say that's too bad, they'll need to catch it as they re-read the article.

This needs to be removed. Tony (talk) 23:01, 11 August 2009 (UTC)

Because people don't always read the article from start to end... --NE2 00:14, 12 August 2009 (UTC)

Well if they do that, it's too bad. This is not a magic blue-carpet service for readers who want to read only Section 5 and expect to be linked everywhere from just that section. Articles need to stand as whole entities. Why don't we link "the" every time it occurs. Why don't we simply turn the whole article, every single word, into blue links, just to be sure that no one misses out.

High-value links are diluted by linking items in every section. It's a major change in practice, and must not be allowed to occur.Tony (talk) 00:37, 12 August 2009 (UTC)

You're not making any sense. Someone going to an article about a railroad and reading the history might want more information about a major predecessor. It would be silly to make them find the only place it's linked just because it's mentioned earlier. --NE2 00:56, 12 August 2009 (UTC)

I think we should take the middle road on this and encourage linking uncommon terms twice: once in the lead and once on the first appearance of the term in the body of the article. That way, readers who just want a quick overview of the subject can read the lead and see the link, and readers who want an in-depth understanding of the topic might skip the lead and still find the link. Dabomb87 (talk) 01:15, 12 August 2009 (UTC)

That might be acceptable, dabomb, since the lead functions quite differently to the body of the article (as does the infobox); it is already the practice of a few editors. But really, linking a term in one section and then another is wasting the value of wikilinks. There's a dilutionary cost for every one added; I don't want to puff up the crisis potential of adding a single link—you know what I mean: editors will add links all over the place in an undisciplined way unless they're guided as to the balance between cost and utility. Look at the French and Italian WPs for examples of ruined wikilinking systems. Tony (talk) 02:27, 12 August 2009 (UTC)

Wow... "wasting the value"... "puff up the crisis potential"... "ruined wikilinking systems"... either you're making a dry joke or you need to step back and look at the big picture... --NE2 02:58, 12 August 2009 (UTC)

Perhaps you didn't see where I was being ironic and where I was not. Tony (talk) 04:03, 12 August 2009 (UTC)

I concur that this needs to be removed. Please look at it the same way as an acronym. Spell it out at first mention, then use the acronym. If someone reads only one section of an article and finds themselves stymied, they will read back, same as they would do if they need more context about the subject. We don't need "in case you are just joining us" wikilinks. --Andy Walsh (talk) 04:38, 12 August 2009 (UTC)

i agree that "link once per section" is too frequent, as a rule of thumb. (besides which, having a rule of thumb within another rule of thumb was awkward.) Sssoul (talk) 05:05, 12 August 2009 (UTC)

I agree that the quoted text needs to be removed. HWV258 06:17, 12 August 2009 (UTC)

I was only trying to clarify what "a long way" meant; I was not encouraging anyone to repeat the same link in *every* section of any article. (And I don't believe that having the same link repeated twice in an article is always useless. Imagine a very technical word used in a minor parenthetical point in section 3 (with a link), and then the same term used in section 17 in a sentence where it's crucial to understand it all. Many readers will have forgotten that they had encountered that term before, so they won't know where to look for the link. Now imagine there is a redirect to section 17: a reader following it will encounter that unlinked technical term and they won't be able to make head or tail of it at all. -- A. di M. 10:15, 12 August 2009 (UTC)

The language is clear enough as stands. People can use their judgment in cases such as what you mention. --Andy Walsh (talk) 13:01, 12 August 2009 (UTC)

Agree. One problem with the text is that it will green-light the edits by those who merely seek to lift their edit count. Such editors will only have to point at the text and say: "but the guidelines allow me to add all those links". HWV258 22:57, 12 August 2009 (UTC)

That's circular reasoning: it's bad because it's bad. If linking is good then I would expect editors to add links where apppropriate and applaud such behaviour.--Michael C. Price ^talk 07:47, 20 August 2009 (UTC)

What about the editors who seek to lift their edit count by removing repeated links? --NE2 07:54, 20 August 2009 (UTC)

After thinking about this for a while, I think linking has two fundamental aspects that overlap significantly but have slightly different requirements: Navigational and explanatory. Navigational aspect: When you go to Cat, you expect to find links to domestic animal and large cat in prominent positions. Possibly in the lead, but certainly from the most relevant section. Explanatory aspect: A reader who does not know a term, or aspects of it that are assumed without explanation, they can follow a link to the relevant article.

Based on this theory I came up with this rule of thumb: (1) If it's obvious from the table of contents where a certain linkable term will be discussed, link it once from that section. In very long articles occasionally two sections far apart from each other may deserve such a link. (2) If the term first appears in an earlier section, e.g. the lead, link it from there as well.

The formulation also allows for situations where it's better to link the second occurrence in a paragraph. Hans Adler 13:22, 12 August 2009 (UTC)

i understand what A di M means about "a long way" sounding vague - but there are too many different situations for a more specific "one size fits all" solution. ("don't repeat a link more than once per section" does read like an invitation to repeat links once per section.) if we're going to add anything to the "long way" point, i propose a parenthetic to show it's vague on purpose - something like: "(how far 'a long way' is will vary; editors can make their own judgements within individual articles.)" Sssoul (talk) 14:33, 12 August 2009 (UTC)

Hans, I agree with the point about two types of links, but I'm not 100% sure I understand what the conclusion has to do with it, as you don't say that the two types of links should be handled differently. (Besides that, links in the lead section will generally be almost all navigational, and links in later sections almost all explanatory.) -- A. di M. 19:49, 12 August 2009 (UTC)

Explanation of the parenthesis above. The lead section will be read by people who have just heard about the article topic (via a search result, or following an explanatory link), so we should not assume that the reader will be willing to go read even more articles in order to understand a sentence. That is not to say that no technical terms should ever be used in a lead; but they should be used in a way that the reader will not miss the gist of a sentence if they don't follow any link. Take a look at the lead of Quark: a lay reader will likely have never heard of "color confinement" or "hadrons" before, but they will be able to understand the point of that sentence anyway. So you don't want purely explanatory links in the lead. As for navigational links, a reader just visiting the article Icelandic alphabet or Flag of Tokelau as a stepping stone to Icelandic language or Tokelau will want to find those links without scrolling down; so navigational links should all be found in the first screenful of the article.

The rule of thumb I generally use is: include each navigational link as near the top of the article as reasonably possible; repeat the most important ones in a {{seealso}} or {{main}} template before the first paragraph of the relevant section, and/or in navboxes, but not in paragraphs. As for the explanatory links, include each one the first time it occurs; and for specialist terms (ones that most people with a high school diploma have never heard of) repeat them the first time they occur in each sentence, except if they have been linked to in the lead. (This is based on the assumption that almost all readers will read the lead, and then some of then might jump through the TOC to the section they're interested in.) -- A. di M. 13:35, 13 August 2009 (UTC)

I have never understood the aversion folks have to linking, since I do not find it reduces readability; quite the contrary I find it reassuring to read a heavily linked sentence, knowing that I'm only a click away from more details if desired. It's very annoying in the middle of an article to realise that you have to go off on a long hunt for a link, and when you find it and click through you aren't returned automatically to where you where.

The notion expressed above that you have some moral duty to click on the first occurence, and if you missed it, well fuck you, is one of the most unfriendly attitudes I've seen here for awhile. It reduces rhe utility of WP. --Michael C. Price ^talk 23:21, 19 August 2009 (UTC)

I agree, I couldn't believe my eyes when I read some of the language used above. I'm appalled by the distain showed here for our readers. Limiting linking to the extent talked about above limits the usefulness for no other reason than it seems to offend some people. The appropriate level of linking will be found just as any other editorial question is settled on Wikipedia and not by some (literally) arbitrary set of rules. Editors can (and will) make judgements about linking on an article by article basis. RxS (talk) 05:05, 20 August 2009 (UTC)

MichaelCPrice, please remove your re-insertion of that awkward "rule of thumb within a rule of thumb" about linking no more than once per section. as your edit summary noted, this discussion has not concluded and there is no evidence that consensus supports the inclusion of that statement, and further discussion is appropriate. thanks Sssoul (talk) 09:43, 20 August 2009 (UTC)

I agree it was awkard, so I've reverted it.

I propose that the first sentence be changed from

In general, link only the first occurrence of an item.

to

In general, link the first occurrence of an item in a section or subsection.

The reason is straightforward. Some navigational links, by design, take readers directly to an article's section or subsection. It defeats the purpose and intent of the link to require them to go on a further link hunt.

The following exceptions are still valid and do not require updating. E.g. if the (sub)section is very long then further linking may be appropriate. --Michael C. Price ^talk 10:19, 20 August 2009 (UTC)

(outdent) thanks for the self-revert and for continuing the discussion. proposing to "link the first occurrence of an item in a section or subsection" may make sense for some particular items in articles that have very long sections, but it definitely won't make a reasonable general recommendation. i suggest keeping "first occurrence only" rule of thumb and the three exceptions listed, adding a parenthetic to the "long way" point - something like: "(how far 'a long way' is will vary; editors can make their own judgements within individual articles)". Sssoul (talk) 10:47, 20 August 2009 (UTC)

Although it may to necessary to state that "long" is a subjective matter, it doesn't address the issue of readers who are directed to specific sections (regardless of length). Unless the section issue is specifically addressed links will be removed, to the detriment of WP's readability.

As it stands the current advice about repeated linking and the ability to link to sections is inconsistent. Wikipedia should be consistent. --Michael C. Price ^talk 12:28, 20 August 2009 (UTC)

Would making that "where a later occurrence of an item is a long way from the first, or is found in a section which is directly linked to by another article" be OK? --___A. di M. 10:25, 21 August 2009 (UTC)

No, because that would require that everytime you add a link to a section you would have to check to see if the appropriate terms were all linked. And remove them all when the link was removed. Better to have them in by default.

And some readers will come to the sections directly, even without being directed there by links. --Michael C. Price ^talk 11:43, 21 August 2009 (UTC)

... is there some efficient way for an editor to ascertain whether or not a particular section is directly linked to from another article? more efficient, i mean, than a reader simply looking at the earlier parts of the article, if s/he wants more information than is in the exact section s/he's looking at? Sssoul (talk) 11:38, 21 August 2009 (UTC)

There's no way to apply "What links here" directly to sections. When you create a link to a section, you are supposed to add a comment to it (see MOS:SECTIONS). --___A. di M. 11:51, 21 August 2009 (UTC)

(outdent) maybe what i'm missing in all this is: what is wrong with considering readers capable of backing up a little to look at other sections of an article, if they want more information? it's not like linking to a particular section confines anyone to that section. Sssoul (talk) 11:58, 21 August 2009 (UTC)

It's a matter of convenience and ease of use, not capability. That's why we have links in the first place, to make finding other articles easier.

A natural extension of your argument is that we should remove all links, since any reader is capable of finding any article, without links.--Michael C. Price ^talk 12:35, 21 August 2009 (UTC)

smile: only in the sense that "a natural extension of *your* argument" is that we should link everything - which isn't a constructive way to discuss this, so let's skip all that, okay? i believe that the guidance now given (linking only first occurrences as a general principle, with exceptions noted) caters quite sufficiently to readers' ease/convenience. Sssoul (talk) 12:45, 21 August 2009 (UTC)

No, it doesn't. Why would we want to make readers scan a whole page for a link they are interested in, sometimes in sections they haven't read and are not interested in? Why would we make Wikipedia harder to use? Why would we not use the capabilities MediaWiki offers to make finding information easier to use? That's the whole point of this project. RxS (talk) 13:27, 21 August 2009 (UTC)

Exactly.

--Michael C. Price ^talk 14:53, 21 August 2009 (UTC)

BTW, just so we don't reinvent the wheel, Archive 4--Michael C. Price ^talk 04:32, 22 August 2009 (UTC)

smile: i'm glad the two of you agree, but others don't, so we need to continue the discussion here, not in the guideline itself.

the wording "in general link only first occurrences with these exceptions" doesn't prohibit anyone from repeating links - even more than once per section, if the sections are large enough to warrant that. the wording you're proposing ("in general, link once per section") would read like an invitation to repeat links in each section even if the sections are only a few sentences long - which is ~~obviously silly~~ considered overlinking. i assume you two are not arguing in favour of ~~something that silly~~ that kind of excess - but these guidelines apply to ALL articles, including short ones, and need to be formulated with that in mind. Sssoul (talk) 06:11, 22 August 2009 (UTC)

And what exactly is silly with links in sections, regardless of length?--Michael C. Price ^talk 07:47, 22 August 2009 (UTC)

Michael, are you referring to repeat links within sections? If so, they add no utility for the reader, but each comes at a slight cost. The cost has been detailed and discussed ad nauseum, here and elsewhere, and relates to the psychology of signalling and of reading—a psychology that is readily understandable by all editors. Tony (talk) 09:14, 22 August 2009 (UTC)

It is not "readily understood" by all editors as witness the archives which I linked to above, along with debate ongoing here. Such demonstrably false statements only undermine your stance's credibility.

PLease provides diffs or link to source your claim of universal support for your position, e.g. an RfC or whatever. --Michael C. Price ^talk 21:01, 22 August 2009 (UTC)

I cannot provide diffs that demonstrate "universal support", since you plainly do not support the notion of more rather than less selective linking. Tony (talk) 07:06, 23 August 2009 (UTC)

Exactly. And not just me, as the archives and this section testify. So, back to my question, "what exactly is silly with links in sections, regardless of length", and please provide evidence, not just unsupported claims of near universal support.--Michael C. Price ^talk 09:55, 23 August 2009 (UTC)

Michael, I didn't call frequent repeat-links "silly", did I? I usually prefer to be understated, and I can see where you're coming from, so I respect the internal logic of your position (while believing that the frame is not usefully conceived). Nor have I claimed "near universal support"; if that were the case, en.WP would not have a serious overlinking problem. (I've just found multiple repeat links in the article on Dolly Parton, and worse, many many items like "high school" linked—someone went wild with the square brackets without understanding the skills required to make the system work optimally.) The really valuable links—to her songs, for example, were swamped in this sea of blue. Nor would the other WPs (French and Italian especially, which have no guidelines for wikilinking) be beset by scattergun spray-paint linking that unthinkingly drowns the links you'd want to attract readers to.

Now, it really is all a matter of balance: dilution and unprofessional visual appearance on the one hand, versus the utility of links for our readers. The difficulty of establishing that balance is why editors engage in debate (sometimes vigorous); and editors who have put a lot of work into articles they want to see linked (the "orphan" issue) represent a vested interest. IT professionals and enthusiasts are another group of users who tend to perceive the potential utility of wikilinks, and the connective structure on the web, in isolation from their costs. As editors, it is easy to be beguiled by this notion of utility, because we're a little distant from the experience of (most of) the visitors to WP, who we've been told by a number of IT specialists tend not to hit links at all, or if they do, only rarely (see my talk page). Nevertheless, the en.WP has moved significantly towards that need to balance (skilled wikilinking, I'd call it) over the past few years. Tony (talk) 10:19, 23 August 2009 (UTC)

I am picking up from you that you don't see this need to balance. Is that correct? Tony (talk) 10:19, 23 August 2009 (UTC)

A balanced approach is not indicated by your statement that

"repeat links within sections .... add no utility for the reader".

--Michael C. Price ^talk 14:37, 23 August 2009 (UTC)

My view on this (as someone that doesn't read articles from top to bottom) is that when I come across a word in an article that I would like to click on, that is not linked, but which I think might be linked earlier, I find it very annoying to hunt back through the article, trying to find where it was linked. It jars my reading experience almost as much as it would to go and type it in a search box, or do Ctl-F (or whatever your search function is). And then I have to go back and try and find where I had stopped reading the article. The point of linking is to make such side-excursions easy, not difficult. As a reader, gliding past a link that maybe shouldn't be there is much easier than stopping to search or look for an earlier link (this is why I favour a conservative approach to tackling over-linking). On the other hand, there are ways around this. The method I now use is a function available in the Mozilla Firefox browser (maybe as an add-on) and maybe in other browsers as well. This function (not sure what it is called) means I can select a word with the mouse cursor (i.e. highlight it), right-click, and select a "search Wikipedia for this word" option. This is almost as convenient as clicking a link and much less disruptive than searching up and down a page, or going over to the search box. However, before anyone says this is an argument never to link, remember that useful information on relationships between articles and topics is present in the web of interconnecting links between articles (provided things aren't over-linked), so links are need for reasons other than people clicking on them. My view is that both over-linking and under-linking are things to avoid, and I don't want to see an over-reaction to linking that means we end up with an encyclopedia that is under-linked. In this particular case, for readers that don't have this add-on tool, and who might arrive at a section from a section link, and for those long articles, I would favour repeating links where it might be needed. Rigidly enforcing a "no repeats" rule is not sensible, in my view. There are things that need more urgent attention (such as clicking on existing links and making sure they are correct, and adding links that are missing). Carcharoth (talk) 14:40, 23 August 2009 (UTC)

I commented on this earlier, but perhaps it merits repeating. In my opinion, linking a term on every appearance is excessive, and linking once per section is also borderline unnecessary. However, allowing someone to link only once in the entire article may be a bit restrictive. So, perhaps we should advocate linking once in the lead, and one or two times in the body, depending on the unfamiliarity of the word (technical terms) and the length of the article. A concrete example:

In evolution, species is a relevant link that should be linked once in the lead and once in the body, but no more, as it's not overly technical or unfamiliar. On the other hand, modern evolutionary synthesis, in addition to being linked in the lead, might be linked two or even three times in the body as a relatively less-known (but important) concept. Dabomb87 (talk) 15:59, 23 August 2009 (UTC)

Carcharoth's statement above expresses my sentiments pretty closely. The "rule" should neither imply encouragement of a rigid only-link-once-per-article regimen nor any sort of link-every-occurrence mania. Much as with Varieties of English, the guidance should encourage tolerance -- link minimalists should try to appreciate that not everyone shares their aesthetic ideal and link maximalists may need to be reminded that not every possible link adds value to an article and that overlinking can detract from the usefulness of the links. older ≠ wiser 16:22, 23 August 2009 (UTC)

I've been a webmaster for major companies for some years. Every couple weeks I get a request to "make as many links to and from an article" in a knowledgebase as possible, the assumption being that more people will read the article. In practice, what Webtrends and other analysis tools show is that readers quickly become "saturated" with links on a page, even to the extent that they will ignore a link in bold red in an article "PLEASE READ THIS". In practice the golden number of article links is about ten. Any more, and the number of readers clicking ("click-throughs") doesn't increase significantly, if at all. Therefore the strategy is to pick the best ten links. (The best articles, the most important issues — whatever the criteria may be.) This affects the question of whether there should be repeat links.

There is another pragmatic issue: how many readers scroll down the screen ("below the fold"). There is debate. (Notice how AOL director [8] misconstrues her reference [9].) At any rate, some substantial number of readers do not scroll past the first screen, and most do not scroll to the bottom. This suggest two things regarding Wikilinking: 1) Links placed toward the top of an article are far more likely to be seen than ones at the bottom. The second inference, I can't provide statistics for, because the knowledgebases which which I have webmaster experience did not generally have a table of contents. However, the implication is: 2) Few readers would jump down pages in an article, without reading at least part of the opening page.

Most of the factors above in discussion against multiple linking I would agree with, but not on the basis of my statistical analysis. The statistical case alone for few links outweighs any small gain for a small number of readers. So not to gild the lily, but there are professional guidelines about hardcopy "linking". Readers expect to see the first instance of a term ... defined, footnoted, given an acronym, etc. Duplicates are confusing, because they cause the reader to think, "Wait, didn't I already see that term?" And then they have to stop reading, go back and check. I'm a skimming reader, and stumble over these links constantly. In articles with names of many foreign kings or cities or ingredients, it makes reading extremely difficult — the duplicate link leads me to think that I misunderstood something earlier.

In sum. An article in my knowledgebases that got 10% click-throughs to other articles was rare. 1-2% was typical. So multiple links are labor intensive for editors, little-used, contrary to professional hardcopy guidelines, and make articles more difficult to read for the vast majority of readers. I.e., in their current, limited technical form, they are a lose/lose/lose situation. Piano non troppo (talk) 00:32, 24 August 2009 (UTC)

fascinating - thanks, Piano non troppo. a big part of what's underlying this seems to be that some editors view articles as generally meant to be read, while others are viewing them primarily as a gateway to side-excursions.

Carcharoth, you wrote: "when I come across a word in an article that I would like to click on, that is not linked, but which I think might be linked earlier, I find it very annoying to hunt back through the article". can i ask how many times *per screenful* you want the items you're interested in linked? i still think it depends too much on the particular item/article for "one size fits all" type of solutions, but maybe considering it "per screenful" is more useful than "per section". Sssoul (talk) 05:36, 24 August 2009 (UTC)

Piano ma non troppo, many thanks for the hard-won insights "from the trenches". Most WP editors see overlinking as a problem, as is readily evident from scrolling through their commentary on the various date-delinking RfCs, discussions and WP:DATEPOLL. Goodmorningworld (talk) 08:39, 24 August 2009 (UTC)

Easter egg question

I just have a quick question: Do people think I created an easter egg with this edit? — Sebastian 22:39, 17 August 2009 (UTC)

Maybe not, but I would have kept it as it was before, per the "better safe than sorry" principle. Never underestimate readers' stupidity. --___A. di M. 09:44, 18 August 2009 (UTC)

I think it's fine.—MDCollins (talk) 12:50, 18 August 2009 (UTC)

it seems fine to me too Sssoul (talk) 13:17, 18 August 2009 (UTC)

Thanks for your replies. To A. di M.: I assume you're saying that you wold have kept it, but if you were writing it as a new text, you would leave it out. Such a view makes sense, because the editor who put it in there before probably had a reason for thinking it might be needed. But in this case, that original editor was me, too. So, I weighed both concerns: The principle of least astonishment and the ideal of legibility, and ended up deciding that the latter prevailed in this case. Anyway, this was an interesting borderline case! — Sebastian 18:16, 18 August 2009 (UTC)

Sebastian—Your edit in question is fine, I think; but the article itself is badly overlinked. And I see "Amsterdam" linked in the lead and then 15 seconds later in the first section. This is bad practice, I believe. Tony (talk) 10:04, 21 August 2009 (UTC)

Since sections are linking entry points it looks like good practice to me. --Michael C. Price ^talk 10:19, 21 August 2009 (UTC)

Tony1, I cleared up a few examples I considered overlinking. Better?----occono (talk) 01:32, 22 August 2009 (UTC)

Occono: nice work. I went through after you and caught "diary" and bad links to institutions; plus the date formats were inconsistent. Tony (talk) 05:26, 22 August 2009 (UTC)

Michael C Price: I think you overestimate the number of readers who will ever hit a link (the rate estimated at 1–2% by an ex-web master, see my talk page. I frankly don't care in the case of the extremely rare case in which someone ignores the lead and starts reading from the first section: they can type "Amsterdam" into the search box. It can be irritating to see "Amsterdam" (of limited utility, anyway) blued out twice in such a short period, and this slightly dilutes the high-value links in the vicinity. What is clear to me and many other editors is that we have to be smart about attracting readers to links. Doubling up is not a smart way to increase the hit rate, and while it might make some editors feel that they've added to the intricate interconnectedness of the project, in the cold light of day, rationing the links is a better way of making the system work optimally, I put it to you. Tony (talk) 05:08, 22 August 2009 (UTC)

The system works best when our readers can navigate easily around topics they are interested in. Aggressively rationing links hurts this goal and undermines one of the things that makes the MediaWiki platform useful for this purpose. RxS (talk) 05:40, 22 August 2009 (UTC)

Once again, I can only add my support to RxS's views. Rationing links is the same as rationing utility, since links within sections have utility, despite claims to the contrary. --Michael C. Price ^talk 14:51, 23 August 2009 (UTC)

Over-linking and under-linking

I've been following the over-linking and under-linking debate on and off for about a year, and my concern here is that changes that will affect the entire encyclopedia are being discussed by a small group. This actually applies to other manual of style discussions, but that should be raised at WT:MOS. What I fear is that what happened with date delinking will happen here. That small groups will gain local consensus for ever-more strict changes, and that as more and more people become aware of efforts to de-link or re-link (especially if anyone mentions bots or script automation being used - I don't know if that has ever been considered, or whether script automation is already being done, but that would raise tempers somewhat), that the dispute will snowball. What I propose is that major changes that affect all the articles in the encyclopedia (most MOS changes are not on this scale) should be discussed by much larger groups. And that until such large discussions are set up (there would need to be discussion beforehand on how to present such discussions) that efforts are directed towards gathering data and sources to back up assertions being made. Carcharoth (talk) 16:01, 23 August 2009 (UTC)

I share your concerns. This needs to go higher up. Perhaps an RfC and some sort of user survey.--Michael C. Price ^talk 17:11, 23 August 2009 (UTC)

What are the "changes" in question? The guidelines have been fairly stable for some time. Tony (talk) 05:43, 24 August 2009 (UTC)

Scripts that deal with linking

Is there a list anywhere of scripts that deal with linking? I've found the following so far:

I believe that Twinkle and Huggle and AWB and other such things also have functions to add and remove links. Could those be summarised here? Carcharoth (talk) 16:08, 23 August 2009 (UTC)

I used to use the Lightmouse script for unlinking common country-names, but stopped quite a while ago: it was a bit lumpy for this purpose, and proves the point that editorial judgement is required; in particular, Ckatz pointed out that there were issues with inline lists of countries (possibly with tables, I can't remember). I know of no other script that unlinks anything, but I'm totally ignorant of AWB, since I have a Mac (bot-free zone).

I've asked Nickj for clarification about a technical problem with his script that a number of editors have been complaining about. He has answered bottom of this section; he points out that his RL work is rather full-on at the moment, so his wikitime is limited. He acknowledges the need for careful judgement in accepting/rejecting the script's list of suggestions for potential links. Tony (talk) 09:09, 24 August 2009 (UTC)

Linking, delinking, changing links, and link maintenance exercises

I've noticed User:Tony1/Build your linking skills, which is a good set of exercises for spotting and removing unnecessary links and improving link syntax (and piping of links). What I think is missing there is exercises that point out when links need to be added, and how to spot and correct incorrect links (I call this link maintenance). Examples are: spotting when a redlink has incorrectly turned blue (when someone creates an article without checking what links to it); spotting when a link has turned into a disambiguation page without the incoming links being disambiguated; spotting when a link has turned red because an article was deleted (OK, that is a trivial example); spotting when a blue link is incorrect (links to the wrong article), and so on. In addition, exercises where a word that should be linked, but hasn't been linked, should also be included. Rather than have separate exercises for tackling overlinking and ones for dealing with underlinking and ones dealing with link maintenance, is it not possible to combine them and edit them collaboratively? Carcharoth (talk) 16:34, 23 August 2009 (UTC)

It's still a work in progress. Dabomb87 (talk) 16:36, 23 August 2009 (UTC)

Shall I move this discussion to the talk page of the work in progress? Carcharoth (talk) 16:40, 23 August 2009 (UTC)

Probably a good idea. Dabomb87 (talk) 16:57, 23 August 2009 (UTC)

Moved here. I'll leave this section open in case anyone wants to comment here on linking exercises in general, as opposed to the one Tony is developing. Carcharoth (talk) 20:42, 23 August 2009 (UTC)

Yes, it's a stub, really. Thanks for your suggestions, Carcharoth—I'd already identified the need to write exercises for the correction of underlinking (although I see it as a much smaller problem than overlinking). I'm unsure what material to use for the changed-red-links idea: do you have any suggestions? (I'll put this and subsequent entries at the talk page there.) Tony (talk) 23:25, 23 August 2009 (UTC)

How about "some idiot has bypassed a redirect"? --NE2 06:19, 24 August 2009 (UTC)

Oh, I think I'm missing the joke because I'm a computer klutz! I suspect the current show-and-tell unfolding exercises are not suitable for picking up errors in redirects, but let me know of any issue (or example) that would fit into Gary King's exercise template as you see it on the tutorial page. Tony (talk) 06:24, 24 August 2009 (UTC)

Well, something like [[Atchison, Topeka and Santa Fe Railway|Nevada Southern Railroad]]. Nevada Southern Railroad is a bad redirect anyway, but even if it were a good redirect that would be bad piping. --NE2 06:46, 24 August 2009 (UTC)

The same thing happens a lot with bots indiscriminately bypassing redirects - see Wikipedia talk:Double redirects#Many_double_redirects_are_good. — Sebastian 14:07, 24 August 2009 (UTC)

No, that's a completely different issue. --NE2 15:17, 24 August 2009 (UTC)

Well, then I'm missing your joke, too. — Sebastian 15:47, 24 August 2009 (UTC)

Not sure how you can miss [[Atchison, Topeka and Santa Fe Railway|Nevada Southern Railroad]], but OK... --NE2 03:16, 25 August 2009 (UTC)

I am aware that, as you said before, Nevada Southern Railroad is a redirect (to Atchison, Topeka and Santa Fe Railway), but where's the joke? — Sebastian 06:08, 25 August 2009 (UTC)

I never said it was a joke; it's an example of horrible linking. --NE2 08:35, 25 August 2009 (UTC)

"plain English words"

rm hard copy argument

"See also" section

"Piped links should never be used to introduce tendentious subtext,"

You may also like this ...

Structure and layout

Green nowiki syntax

What should happen with External links section?

Linking city + region

Foreign language

Date unlinking bot proposal

back to that "proposed restructure" ...

Should "PDF" be linked?

Inconsistency

Particle physics

the "what generally should be linked" section

Cross namespace links

"Link density" et al.

Links to Non-English Wikipedia articles

Strong disagreement with encouragement to repeat-link

Easter egg question

Over-linking and under-linking

Scripts that deal with linking

Linking, delinking, changing links, and link maintenance exercises