Wikipedia talk:Only make links that are relevant to the context/Archive 4

Issues regarding repeating links

subsection link

For longer articles: Would it be good to have a first link policy for subsections? In this case, if somebody moves the section to a separate page, the link would still be there. E.g. if you have only one link to George Washington early on the page of United States, then take a section of that page to make a subpage History of the United States, you would end up with no link to George Washington. Chris 73 04:55, 14 Nov 2003 (UTC)

I agree w/ subsection link. Even if we never move the subsection, when we use #-link, like [Univer#Solar system], the reader will be directed to that section, not the beginning. So if he started reading there (and he most likely will), he'll encounter an un-linked term, thinking there's no article. --Menchi 04:59, 14 Nov 2003 (UTC)

I'd generally vote against the subsection link. If a subsection gets moved to a new page, the context is now different. Maybe it should be linked but maybe it shouldn't. I think that should be left as a decision for the author/editors of the new page. Rossami 21:13, 18 Nov 2003 (UTC)

I agree to an extent, even for sections that aren't split out. If one mentions a term and links it early on in the article, and then starts a new section talking about the same term six or seven paragraphs down, then the initial mention in the section should definitely be linked again. This kind of redundancy is good. --Joy [shallot] 17:23, 3 Oct 2004 (UTC)

I seem to be weighing in here pretty late, but will do so anyway. This is an instance where I think people need to use a little judgment. You do not need to make a link in a new section if the last time the link appeared was "recently" in the article. If you are 5-pages down from the last time it was linked, it wouldn't hurt to link again. Chuck 18:38, 1 August 2005 (UTC)

Objection to strict guideline against repeated links

While in general I follow the advice of not repeating links to the same topic, there are many circumstances in which I have done so, because otherwise the link may be difficult to locate. For example, suppose the article has a history section with a link buried in the middle of a huge paragraph, and two pages later in a different section that same topic is central to a discussion of its cultural impact. A repeated link is clearly appropriate, particularly since some readers may not read both sections. I'm not sure how to properly codify this, but I would say something to the effect of: "Restrict repeated links to circumstances where either the first link is not prominent or the linked topic is being used in a new context distant from its previous use." 06:16, 13 January 2006 (UTC)

I agree. But can't we soften it even further? "Don't go overboard with links. You probably only need to link once to each topic." or something. I often find situations where early in the article, say on a band, you say "Jim was a friend of [Bob]". Then later on, if you follow this guideline, you end up with a sentence like "Jim, and fellow members [Mick], [Alice], Bob and [Willie]". It just seems silly not to link to Bob in the second case. Stevage 07:19, 13 January 2006 (UTC)

In this instance, ordering might be important: it wouldn't appear as odd (methinks) if the second sentence were rendered as follows: "Jim, and fellow members Bob, [Mick], [Alice], and [Willie]." E Pluribus Anthony | talk | 04:15, 10 February 2006 (UTC)

Yeah, vaguer is probably better, except that sometimes people go around ruthlessly deleting duplicate links with no regard for how much sense it makes. It might be nice to have a policy to point to - at least we should list a few examples where it may be desirable to repeat links. Deco 23:19, 18 January 2006 (UTC)

I don't know. I've seen so much abuse - such massive overlinking - that I think any softening of the rule will make cleanup even harder and less likely to occur. There may be exceptions for extremely large pages (like the first example) but if the page is really that large, I'd be seriously considering whether the page should be broken into several sub-pages.
Stevage's example of the list is a good style question. It doesn't convince me personally. I don't think the format he describes looks silly at all and see no reason to slavishly follow such a pattern. But if you really do think it looks better, you always have the style choice of creating the link in the list instead of the first usage in the article. The guideline only says don't link words "that have been linked earlier in the article". We usually link the first occurrance but we don't have to. Rossami (talk) 23:51, 18 January 2006 (UTC)

While it's generally prudent to wikilink initial terms, I think it's also prudent to rewikify germane terms that appear much later in an article – particularly a long one – or in a different section. This would enable a visitor to not have to scroll up to article text already consulted. There are many instances where dates are overwikified, for example, and serve as unnecessary distractions. The key in using the piped link is doing so judiciously – i.e., within context. E Pluribus Anthony | talk | 04:11, 10 February 2006 (UTC)

Lengthy Articles

It had been my understanding that in long articles, it was acceptable to repeat a wikilink to something linked earlier further into the article. This is to avoid making the reader scroll up and down. The reader may also have been linked into the page by section, so may not see the 'first' wikilink at all.

Can I suggest adding in language like this under Other considerations -

• Repetition of links on lengthy pages may be preferable, since the article may be linked to by section.

--Barberio 09:30, 2 April 2006 (UTC)

Oppose—The existing rule is nice and simple. People would be arguing about what defines a lengthy article. Linking by section is not very common. PizzaMargherita 09:42, 2 April 2006 (UTC)

Support—I also think that when the first link to an article is piped rather than accidental, linking the first accidental link as well makes sense. For example:

(In the introduction) This issue was most recently argued in [[X v. Y|a landmark 2002 case]].... (In the body) In 2002, the Court heard ''[[X v. Y]]''...

In my own reading, I find that sometimes I use a "depth-first" link-following style (see tree traversal), and sometimes I use a "breadth-first" style. In other words, sometimes I click on an interesting link as soon as I encounter it and later backtrack to where I was and continue reading, and other times I read an entire article before returning to links that interested me. An absolutist approach against ever relinking makes the latter style very hard—you can see the word you want to click on right there, but you have to search back and hover over links until you can find the one to click.

Of course, those of us who are participating in talk pages have no problem using the search box in such cases. But I've observed my grandfather-in-law trying to read Wikipedia; if something isn't blue, he doesn't realize he can still find an article about it. Keeping the rule "nice and simple" is a boon for us as editors. But our first priority ought to be to help readers, including neophytes, to navigate, and sometimes that will mean an extraneous link. --TreyHarris 09:59, 2 April 2006 (UTC)

The pipe issue is a seperate one, but I think you're right in that it also deserves an explicit exception. --Barberio 11:59, 2 April 2006 (UTC)

Question. Would anybody support a reversion of this edit? Because judging from his comments, this is what Barberio is seeking justification for here. PizzaMargherita 15:03, 2 April 2006 (UTC)

Oppose. PizzaMargherita said it quite well in his/her first comment. Linking by section is rare. The very real disadvantages in terms of frustrating readers and disrupting readability outweigh the marginal benefits. (In TreyHarris' terms, the frustration arises when I'm reading an article "depth-first", follow a link, return to the article and follow the link the second time, new readers (like my father-in-law) get quite frustrated and confused when they find themselves at an article they've already read.) Please remember that this page is a guideline, not an absolute rule. There are exceptions. However, I strongly believe that they should continue to be dealt with as exceptions and be evaluated on a case-by-case basis. The truth is that if a page is so lengthy that it is routinely linked to at the section level, the article should probably be broken up into separate sub-articles. Rossami (talk) 15:43, 2 April 2006 (UTC)

Unfortunatly Rossami, people like to use the guidelines as whiffle-bats to attack anyone who does decide that an exception is justified. (see above) So when we know the circumstances that would could warrant an exception, we need to explicitly state so to try and combat this behaviour. It shouldnt be needed, since we should take it as given, but human nature needs some prodding sometimes. I think you might also underestimate the amount of sectional links present, since I don't think theres an easy metric to judge that on. --Barberio 16:43, 2 April 2006 (UTC)

Another reason I'd like to add, as a reason not to have this very vague catch-all exception, is that there are several ways to make a particularly relevant link stand out, other than repeating it every other sentence. These include a "See Also" section at the bottom of the article, "See Also" openings of sections, "Main Article" openings, etc. PizzaMargherita 21:54, 2 April 2006 (UTC)

I think two competing forces have come to a bit of a head here:

1. The first relevant context should always be linked. No successive contexts should be linked.
2. In many cases, there are successive contexts that are just as relevant, and in some cases more relevant.

If Barberio is in fact seeking approval to link all those successive instances of MIT and ARPA within the same paragraph, then I certainly am not speaking favor of that. I'd have no problem whatsoever with the guideline saying that the same link should never occur twice within the same standard screenful, or even within the same section. But if the absolutist rule must stand, can someone take my example of the court case above, and explain to me why it's a good outcome that the name of the case cannot be linked?

Barberio is right on another point—we've seen time and time again that an afterthought of "every rule has exceptions" has never prevented pedants from using black-and-white rules as a club. Guidelines are supposed to be dispute resolution aids. Black-and-white rules that in reality require exceptions, but simply have a vague "every rule has exceptions" footnote, result in more disputes, not less. --TreyHarris 23:30, 2 April 2006 (UTC)

No, I don't think multiple identicle wikilinks in the same paragraph are a good thing. --Barberio 23:48, 2 April 2006 (UTC)

It occurs to me that 'sectional wikilinks don't happen a lot' is not a hugely compelling arguiment considering that this very guideline advises use of sectional wikilinks. If we want people to use sectional wikilinks, we should make sure the rest of the guideline is consistant with use of sectional wikilinks. --Barberio 23:48, 2 April 2006 (UTC)

I really think the section linking thing is a straw man. There's another much more reasonable reason to relink things, and that's simply that people skip stuff, and people forget stuff, even when reading an article start to finish. Don't link the same link twice in one sentence, but relinking the same link in two different contexts where a reader might reasonably decide to read one of them without first reading the other seems to justify a redundant link, particularly when these contexts are so spatially distant that the link is not visible. Even real encyclopedias contain multiple cross references in one article. I don't think the policy needs to spell out any of this however — instead, just make it clear that there are exceptions, which I think it does already. Deco 14:33, 7 May 2006 (UTC)

When duplicate links are ok

Should duplicate links sometimes be preferred? I think the table in the h3 section titled "Main Event Overview" on the World series of poker article is a perfect example. The lack of a link on the 1997 listing of Stu Unger, for example, looks like an omission, not like a conscious following of this page's "avoid duplicate links" style recommendation. Pokerriot 10:54, 13 August 2006 (UTC)

I think there was some considerable discussion of this, with particular regard to lists of links where one item was linked earlier. Just use common sense. Deco 11:16, 13 August 2006 (UTC)

Proposed 'Repeating Links' Section

I've added a commented out section with the following as proposed language for a 'Repeating Links' guideline. --Barberio 00:12, 3 April 2006 (UTC)

Repeating links

Generally you should avoid linking words that have been linked earlier in the article. This advice follows the standard practice of defining or explaining a term, or spelling out an acronym, on its first occurrence in a text and not subsequently.

• Repetition of links on lengthy pages may be acceptable, since the article may be linked to by section.
• Repetition of links that use synonym or other alternate wording. For example, an article on a court case may contain "This issue was most recently argued in [[X v. Y|a landmark 2002 case]]..." as well as "...again refered to the spindle sockets rule of [[X v. Y]]".

The first point I still strongly oppose. Regarding the second point (which I'm not so uncomfortable with, but still oppose to include), notwithstanding the arguments against the first point that still apply here, I would also argue that it's sometimes, probably always possible to rephrase it so that you have to link only once. In this case: This issue was most recently argued in [[X v. Y|a landmark 2002 case]] (X v. Y) [...] as in the case of X v. Y."

Also, you see, it's still a matter of degree. In your example, Tray, you point out that the first occurrence is in the Introduction, and the second one in another section, which I find a reasonable example. And yet, in this amended proposal, that clarification has magically disappeared.

Barberio is right on another point—we've seen time and time again that an afterthought of "every rule has exceptions" has never prevented pedants from using black-and-white rules as a club.—Trey, this is in stark contradiction with this edit of yours (now known as the "weasel-clause" of National varieties of English), and also with your supporting these weasel-exceptions in this proposal. If possible, it's preferable to have "nice and simple" rules, with the understood and unavoidable caveat that these are guidelines and not "black-and-white" rules.

Having these explicit and vague caveats will allow people like Barberio, who manages to "interpret", or even misquote nice and simple guidelines to prove his point, to make anything out of these "weaseled" guidelines, and use them as "whiffle-bats", possibly in completely unjustified and personal reverts such as this. PizzaMargherita 07:15, 3 April 2006 (UTC)

Please cease using Ad hominem, it is a personal attack, and this is your second warning. --Barberio 12:17, 3 April 2006 (UTC)

And yet, in this amended proposal, that clarification has magically disappeared.

I did not write the above, Barberio did. I gave a very extreme example—different link terms pointing to the same article and some physical separation—in order to see if there was an agreement that the rule against repetitive links would have problems if enforced in extremis.

Trey, this is in stark contradiction with this edit of yours (now known as the "weasel-clause" of National varieties of English), and also with your supporting these weasel-exceptions in this proposal.

I don't see the contradiction at all. My argument has been pretty consistent that style rules differ in degrees of violability, and a catch-all "every rule has exceptions" does not offer any practical guidance on how inviolate any particular rule is. When we have an entire page of varying guidance, all stated as absolutes, and then cap it off at the bottom with an appeal to WP:IAR, it reads as if each rule has the same degree of inviolability. I think it's more helpful to state reasonable exceptions, and add weasel words as necessary on the rules that are more malleable. Then, when we state a rule in the absolute (such as the recent prohibition against academic titles in name attribution, which I intentionally wrote without weasel words), it's much clearer that the rule really is one that should only be ignored in extraordinary circumstances.

If possible, it's preferable to have "nice and simple" rules, with the understood and unavoidable caveat that these are guidelines and not "black-and-white" rules.

I'm not averse to simple rules; I merely think a rule should be as simple as necessary, but no simpler. The editor who has to make an appeal to WP:IAR is already embroiled in a dispute. I would prefer to prevent the disputes before they happen by having rules that spell out their exceptions and degree of violability.

The idea that rules should be written more simply than necessary because you can ignore them when need be leads to ignoring rules becoming an everyday necessity, just to get the work of the encyclopedia done. In my opinion, that's when the extreme behavior you're concerned about becomes problematic—if people get in the habit of ignoring the rules, the rules are diluted.

In any case, I'm not particularly worried about human pedants who apply rules in extreme ways. My biggest concern is the rise of AWB and other assisted mass-editing systems. Absolute rules lend themselves to correction using such systems. I've been expecting for some time to see the edit summary, "duplicate links removed as per WP:CONTEXT using AWB". I don't know if you've had the experience of having to come behind an AWB user who hasn't paid much attention to what he or she is doing, but if you have, you know how annoying it is. Overly simple rules are often mechanistic. So it's unsurprising that people would try to use automated mechanisms to enforce such rules pedantically. --TreyHarris 08:29, 3 April 2006 (UTC)

I second TreyHarris here. --Francis Schonken 09:46, 3 April 2006 (UTC)

Going by this, and the above times this has been discussed before, I'm going to go ahead and put this in. --Barberio 16:14, 4 April 2006 (UTC)

And I have reverted those changes. Like all the previous times it's been discussed, this discussion has not yet reached consensus.

I don't think anyone is arguing that either absolute position is right for Wikipedia. That's why we have explicitly say on the page that this rule is in dynamic tension with the Build the web rule. But adding a dozen explict exceptions is not, in my opinion, the right way to express the balance that we strive to achieve. To answer TreyHarris' concern, the very fact that we haven't yet seen AWB-enabled changes tells me that in large part, the existing wording is working as designed and that readers do understand that this is a question of balance and judgment. Rossami (talk) 18:17, 4 April 2006 (UTC)

Um... Looking at the above discussions, I count eight people supporting allowing link repetition in certain cases, and two oposed. 4:1 in favour seems like consensus support to me. --Barberio 19:19, 4 April 2006 (UTC)

No, that's a supermajority. We don't make decisions here purely on the basis of supermajorities. Furthermore, it is in my opinion too small a sample size to be changing one of the longest standing guidelines we've had to Wikipedia. It may seem like a minor change to you but it does not seem minor to me. We are in no rush here. Let's give it more time and let more people contribute to the discussion. Rossami (talk) 20:24, 4 April 2006 (UTC)

Er... Actualy, the original version of the guideline was "Words that have been linked earlier in the article, unless you think that sufficient readers will want instant access to the related article in more than one place". The current stricter form was not put in place till 18 October 2005[[1]], so this is hardly a long standing accepted policy. I'm not sure that consensus was shown in support of this edit, and I don't belive talk page discussions show a support for it now.

In light of that, I'm restoring the recently reverted text, because I belive actualy restores most of the original intent of the guideline. If you disagree, I sugest starting a poll on consensus to see if your 18th October alteration is supported by consensus. --Barberio 23:27, 4 April 2006 (UTC)

"Repetition of links on lengthy pages may be acceptable, since the article may be linked to by section."—This exception is way too vague. I think (but I'm not sure) that this translates to "this exception does not spell out its own degree of violability". Or something like that. At any rate, this is a catch-all exception that would allow anybody to e.g. revert this edit.

And, as Rossami nicely put it wrt bots, this is trying to solve a problem that has not occurred yet, violating the general rule "If it works, don't fix it".

Since this post of mine will be surely interpreted as a personal attack, I promptly add the final warning to my talk page. PizzaMargherita 20:23, 4 April 2006 (UTC)

Allow me to clarify this page's history. This page has included the explicit guidance to "Avoid duplicate links on a page" since 19 May 2004. that addition was based on the established practice which was well documented in the archive page as far back as 2003 and on other pages back even further. It was originally and for a fairly long time, very simple advice which was well understood. On 5 Sep 2005, an additional clause was added attempting to soften that rule. That change was discussed and reversed in the rewrite of this page which occurred on 18 Oct. (In other words, the section that Barberio is citing as precedent was in place for about 43 days. Note that the exact wording which Barberio cites was first added on 18 Oct in an early draft of that rewrite and was not actually in place for even a full day.) I have to disagree with the implication that the rewrite significantly changed the intent of the guidance on this page from what it had been before.

You've made your case. Some people appear to agree with you. I disagree and am apparently not alone in that disagreement. Rather than continue this revert war, let's give the issue some time. I will ask again. Please allow time for other people to join this conversation. This is not an issue which requires a rush to judgment. Rossami (talk) 04:36, 5 April 2006 (UTC)

Okay, let's try another, this time heavier on when not to repeat a link:

Repeating links

Generally you should avoid linking to the same article twice within a single article. If another article is relevant enough to be linked to in two or more contexts:

• Always link the most relevant mention of the other article's subject. This is often the first occurrence of the article's name in a paragraph or section discussing the relationship to the subject of the other article.
• If the first relevant mention of the other article's subject is not the most relevant mention, but falls more than a commonly-sized screenful earlier in the article (several paragraphs or more than a few dozen lines prior), link it as well.
• Do not expect readers to hover over a piped link to determine what article it links to. Accidental (unpiped) links are preferred over piped links. If the most relevant mention of another article is piped, also link the most relevant accidental mention of an article, if any. For example, an article on a court case may contain "This issue was most recently argued in [[X v. Y|a landmark 2002 case]]..." as well as "...again referred to the spindle sockets rule of [[X v. Y]]".
• Do not link to the same article using the same link text twice within a single commonly-sized screenful (several paragraphs or more than a few dozen lines) of the article's body.
• Do not link to the same article twice within the same or adjoining paragraphs, even with different link text. If one of the references is accidental, link it and unlink the piped reference. If both are piped, link only the more relevant one.
• The "See also" section may repeat very important links from the article's body, but only if it is judged likely that a reader may use the page solely to find the other link (and hence, may not read the entire article in order to find it). For example, if an article about a province contains the text, "...the province has twelve [[List of major cities in Province|major cities]]...", you may also link [[List of major cities in Province]] in the "See also" section.

There also should be something about captions. Regardless of the advice, explain the issue to the editor with:

• Images draw the reader's attention, and may cause the reader to lose track of links nearby.

As for the advice, I'm ambivalent between:

1. Therefore, prefer a link in the caption to a link nearby in the article's body.
2. Therefore, link both in the caption and in the article's body.

I've got my flame-retardant suit on, fire away... --TreyHarris 05:32, 5 April 2006 (UTC)

That seems fine to me with a couple of changes. We should probably leave a decision on links in captions to another discussion, since it might muddy this one. And I don't think a bolded "Never" is apropriate, the only other use of "Never" on the page refers to the circular redirect. (I've edited the above box)

That aside, I think this guideline makes the issue much clearer than the 'simple' rule did. --Barberio 11:55, 5 April 2006 (UTC)

"We should probably leave a decision on links in captions to another discussion, since it might muddy this one."—Funny how the same principle didn't stop you from counting 8 people in favour of your proposed changes. (Comments on the new proposal later.) PizzaMargherita 13:29, 5 April 2006 (UTC)

Please do not continue trying to be disruptive. If you have nothing constructive to post, don't post. --Barberio 15:00, 5 April 2006 (UTC)

Trey (can I call you Trey? [yes—Trey]), I think your counter-proposal is much better. I do appreciate your efforts and your civil accepting of differing points of view (a clear indirect ad hominem, I guess).

However, by looking at your counter-proposal, with its detailed exceptions and exceptions within the exceptions, I cannot help but think that what you wrote there is a bot. In other words, it's simply calling for somebody to say, "Hey, that's a nice functional specification, I'll write an AWB after it." Which is exactly what you don't want to happen.

The counter-proposal boils down to: "Avoid linking to the same article twice within a single article and use common sense." Since the latter part is implied in any WP guideline, the present version is the one I support.

Also, I don't think the case for a different proposal is strong enough, if there at all. Could you please point out some unresolved disputes that this modified proposal would solve? The current guideline works just fine as far as I can see.

Another question: are the votes at the top of the page closed? And is that a vote applying to the whole guideline? Thanks. PizzaMargherita 07:13, 6 April 2006 (UTC)

Well, to answer your point in the abstract first, I don't actually have any problem with people using AWB or bots appropriately, even to handle style rules. My problem is when the rule is misleadingly simple, and results in somebody saying, "I can write a bot to do that" without understanding that they need to take the rule in a broader context. If, for a given issue, one actually can write a rule detailed enough for a bot to follow without error, then I say, by all means, go ahead.

That said, I've been writing code for a lot of years, and my formal training was in computational linguistics, so I've thought about these issues a lot—and that background makes me very conscious of the limitations of technology when dealing with natural language. I don't think my counterproposal is very amenable to a bot. You're right that it's a careful specification—I think we should strive for that when possible in style rules, as it cuts down on disputes. But it's not a "functional" specification, for a simple reason: it's all keyed to "relevance". A bot is not going to be able to determine what the most relevant context for a link is.

Now, AWB or a human-assisted bot could be used for the two points above starting "do not link...", if they displayed the repeated links' contexts and asked the human to select the more relevant. And I'm fine with that. Linking or unlinking is a mechanical process. It's the identification of relevance that's impossible to code. --TreyHarris 07:59, 6 April 2006 (UTC)

"it cuts down on disputes"—Which ones? PizzaMargherita 08:28, 6 April 2006 (UTC)

I think a great example was disambiguation pages. People fought and fought over those until we rewrote Wikipedia:Manual of Style (disambiguation pages) to specify a standard style, rather than just stating some principles of organization and assuming that good faith and common sense would take care of the rest. The disambig page disputes really have quieted down in the past few months since, even though we now have a much more specific style than we used to. I've noticed even just in my own individual disambig page edits that I'm much less likely to be reverted and have to discuss the change now. --TreyHarris 15:52, 6 April 2006 (UTC)

That's not what I meant. Could you please point out some real (as opposed to hypothetical), live, unresolved disputes, regarding repeated links, that this modified proposal would solve? You are proposing a solution, but to which specific problems? PizzaMargherita 16:22, 6 April 2006 (UTC)

Ah, I see. No, I think the current absolutist rule does a very good job of resolving disputes—in fact, it will do a better job than my counterproposal, just because it's so absolutist. I just think the way it resolves disputes is incorrect. Starting a dispute in order to demonstrate would be a violation of WP:POINT, so I won't do that. ;-)

I, too, often object to guidance being added on the basis that there's no current dispute, but that's when there's a proposal for new guidance on a topic currently unaddressed. My point in referring to dispute resolution earlier was that when considering any language being added to a policy or guideline (whether or not the topic is currently addressed), you need to consider possible future disputes. --TreyHarris 18:53, 6 April 2006 (UTC)

Anyway, I feel a lot for reverting

Words that have been linked earlier in the article. This advice follows the standard practice of defining or explaining a term, or spelling out an acronym, on its first occurrence in a text and not subsequently.

back to:

Avoid duplicate links on a page. Redundant links clutter up the page and make future maintenance harder. However, link the first occurrence of a term, and always link when directing to a page for more information, e.g. "Relevant background can be found in Fourier series". It is not uncommon to repeat a link that had last appeared much earlier in the article, but there's hardly ever a reason to link the same term twice in the same section.

Apart from changing the last word from "paragraph" to "section", that was the version of that paragraph before Tony1 & Rossami reworked it (apparently without consensus) on 18 October 2005. That is also a version I can perfectly live with. It also has the advantage over the Barberio/TreyhHarris/et al. versions discussed above that it is shorter, without losing clarity. --Francis Schonken 13:49, 11 April 2006 (UTC)

That seems okay with me. With the caviat that we should consider the longer, clearly stated version if there are any disputes over proper use. --Barberio 18:51, 11 April 2006 (UTC)

I still prefer the current version, but I suppose I can live with that, thanks Francis. With the following change I could even support it: "It is not uncommon" -> "It may be acceptable". Also, I would keep

This advice follows the standard practice of defining or explaining a term, or spelling out an acronym, on its first occurrence in a text and not subsequently.

as a second or third sentence. PizzaMargherita 19:51, 11 April 2006 (UTC)

Sigh. Following this discussion, I attempted to restore the agreed upon text today. This was almost imediatly reverted. While I try to assume good intent, I can't understand reverting something that *was agreed upon*. I am reluctantly forced to assume that it was only reverted because I was the person who altered it. --Barberio 10:44, 14 April 2006 (UTC)

Only three users have discussed it, and only two supported its unaltered version, so it's premature. That's why I reverted. PizzaMargherita 10:55, 14 April 2006 (UTC)

Reading PM's edit summary ("please substantiate in talk page"), I'd say Barberio's change *had* been substantiated on this talk page, so PM had no reason to revert.

So now I could back up Barberio's change by re-reverting... which I won't do (yet), instead I propose the following, partially taking account of PM's suggestions above:

1. I don't second PM's suggestion to change "It is not uncommon" -> "It may be acceptable". Reason: it somehow changes the meaning in a way I'm not prepared to support; on the other hand changing "It is not uncommon" to "It may be advisable" might be an improvement as far as I'm concerned. But I'm perfectly fine with "It is not uncommon", my proposed change would make the sentence maybe a bit more contorted, that is: in my personal appreciation;
2. Following version, including a sentence now on the guideline page (as suggested by PM) is OK for me:

   Avoid duplicate links on a page. Redundant links clutter up the page and make future maintenance harder. However, link the first occurrence of a term, and always link when directing to a page for more information, e.g. "Relevant background can be found in Fourier series". This advice follows the standard practice of defining or explaining a term, or spelling out an acronym, on its first occurrence in a text and not subsequently. Further, it is not uncommon to repeat a link that had last appeared much earlier in the article, but there's hardly ever a reason to link the same term twice in the same section.

3. (New suggestion:) Move from Wikipedia:Only make links that are relevant to the context#What should not be linked to Wikipedia:Only make links that are relevant to the context#Specific issues, in this format:

Repeating the same link

Avoid duplicate links on a page. Redundant links clutter up the page and make future maintenance harder. However, link the first occurrence of a term, and always link when directing to a page for more information, e.g. "Relevant background can be found in Fourier series". This advice follows the standard practice of defining or explaining a term, or spelling out an acronym, on its first occurrence in a text and not subsequently. Further, it is not uncommon to repeat a link that had last appeared much earlier in the article, but there's hardly ever a reason to link the same term twice in the same section.

OK?

Further, "66% consensus discussed by three users" is always better than "consensus imposed by two, opposed by two others today" (which was the 18 October version, unexplainably still being in the guideline today) --Francis Schonken 11:15, 14 April 2006 (UTC)

This 18th October was the reference I was looking for, thanks. (My reversion also took into account this comment of Barberio's.)

It doesn't look like at the time the change was controversial. It lasted for 6 months undisturbed and it seems to have performed well. So you can't really say that they "imposed" any consensus. I still think we should wait for a couple more opinions, but I won't oppose the amended change you just proposed. PizzaMargherita 11:36, 14 April 2006 (UTC)

Above I wrote I wouldn't back up Barberio's revert to the pre-18 October version yet. The link you provided to Talk:History of the Internet makes clear the issue is a bit more acute than I had assessed thus far, while you're slamming each other on the head with what this guideline is supposed to say. So I implement the 66% consensus, discussed by three "in tempore non suspecto". Note that I could even insinuate a 100% consensus by three if I read your "I suppose I can live with that" (above) correctly. Take discussions from there (that is: I advise not to proceed with any more reverts of the guideline, before the thing has been discussed and agreed upon with a sufficient consensus on this talk page). --Francis Schonken 12:08, 14 April 2006 (UTC)

I'm happy with this version. But think we should keep the option to impliment the above more explicit guidelines on repeating links if there are any future disputes. --Barberio 12:17, 14 April 2006 (UTC)