Wikipedia talk:Citing sources/Archive 41

This is an archive of past discussions. 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.

Editors often add references after sentences in a certain order for editorial reasons. For example:

Mary likes cake, but John doesn't.<ref name=Mary/><ref name=John/>

AWB contains a feature that re-orders refs to appear in chronological order after sentences: 1, 2, 3, and never 2, 3, 1. When

The re-ordering feature was added to AWB by

"Editors should exercise caution when rearranging or inserting material to ensure that text–source relationships are maintained. References need not be moved solely to maintain the chronological order of footnotes as they appear in the article, and should not be moved if doing so would break the text-source relationship."

Pinging people who have discussed this before @

@Rjwilmsi and Rich Farmbrough: too. -- Magioladitis (talk) 20:46, 12 November 2016 (UTC)

I love you, SlimVirgin! (go figure). Thank you for bringing up this perennial thorn in my side. The apparent rationale is that readers would be befuddled if they saw a series of citation numbers not in ascending sequence (gasp!!), which is asinine on its face when compared with the reasons for not doing this. I think I have raised this at AWB and the response was to the effect of "There are ways around this including use of the {{R}} template." Translated: "Arguments noted, but we really can't be bothered." It is high time to apply some reason on this issue. ―Mandruss ☎ 20:56, 12 November 2016 (UTC)

Thanks, Mandruss. The AWB feature fails to take into account how dynamic ref placement is. In a print article, or an online one that won't be edited, it's easy to make a final edit to ensure that inline refs appear as 1, 2, 3. But with ref name, live publishing and "anyone can edit", ref positions and numbers are constantly changing. SarahSV ^(talk) 21:24, 12 November 2016 (UTC)

AWB does a relatively good job of keeping the refs in order, despite the flux in articles. We cannot keep Wikipedia free of typos, no one would suggest that we don't fix as many as we can. All the best: Rich Farmbrough, 08:36, 13 November 2016 (UTC).

I see

Carl raised this with Magioladitis in May 2010 here, and M said there was a way to deactivate it. SarahSV ^(talk)

21:57, 12 November 2016 (UTC)

Per the third example at

WP:INTEGRITY, it would provide more text-source integrity to the reader (and avoid the renumbering issue) to use Mary likes cake,<ref name=Mary/> but John doesn't.<ref name=John/> GoingBatty (talk

) 22:35, 12 November 2016 (UTC)

If the only rationale is "I don't like it when it does this because I've ordered the references in such a fashion", then

WP:CITEBUNDLE). --Izno (talk

) 00:01, 13 November 2016 (UTC)

GoingBatty, there are editors who like to do that, and others who believe that articles are harder to read when referenced that way. The point is that editors working on the article should choose the best order of references. SarahSV ^(talk) 00:09, 13 November 2016 (UTC)

Three consecutive cites is certainly not CITEKILL in many cases. And, in cases of two or more consecutive cites, there are legitimate situations where I want a reader who looks at only one of the sources to look at number 31, not number 28. I assume they will look at the first one. AWB currently denies me that editorial judgment and discretion, and for no remotely good reason as I said above. Numerical sequence is a purely cosmetic issue that means nothing to any readers except—maybe—a few who should be on OCD meds. ―Mandruss ☎ 08:31, 13 November 2016 (UTC)

Editors often place refs in a particular order for an editorial reason - while I was running an AWB based bot, which made millions of edits, I had two or three queries about this feature. None of them related to an example where the re-ordering was a bad thing. I'm not sure what the value of "often" in this sentence is, but it seems to me that "almost never" would be a better approximation.

All the best: Rich Farmbrough, 08:33, 13 November 2016 (UTC).

I'll meet you halfway at "occasionally". And that beats the ascending sequence rationale as far as I'm concerned. I've yet to actually see a defense of that rationale, all I see are people telling me that I needn't have the editorial concerns that I have—much like people who criticize me for caring about MOS issues. Until I see some cogent reasoning for ascending sequence, AWB doesn't get to tell me what to care about or how to think as a Wikipedia editor. I grant you that the technique of citation ordering I describe is not widely used; so what? It might be more widely used if editors knew that AWB wouldn't come along eventually and destroy their work. ―Mandruss ☎ 08:41, 13 November 2016 (UTC)

@Rich: Your bot edits bothered many people, including me. How many times were you blocked? I never complained about the fiddling you did for a variety of reasons, but I assure you that a lack of complaints is not an indication that something is desirable. The suggestion that a bot should override an editor's careful consideration is absurd. If someone finds a good reason to rearrange a sentence so out-of-numerical-order citations are avoided, they are welcome to try. However, SarahSV is obviously talking about examples more complex than the Mary/John case. Johnuniq (talk) 08:58, 13 November 2016 (UTC)

Has this ever been RfC'd? If so, can someone drop a link without a lot of trouble? If not, I'm prepared to RfC it. The opposition arguments here are almost all

Bots enforcing pseudo-standards not being backed by community consensus against the explicit wishes of authors/main content providers are imho a general nuisance that frustrates authors depending on the concrete case somewhere between occasionally and often. Potentially it drives authors away.--Kmhkmh (talk) 11:51, 13 November 2016 (UTC)

For an extreme example of the issues mentioned by SarahSV see

United States presidential election, 2016#Other third parties and independents. A heavily edited article with changes introduced each news cycle. Updating ballot access lists for a couple of dozen candidates, with new states releasing lists of candidates almost daily is challenging. To deal with contention and advocacy, each state needed to be referenced for each candidate. To keep lists of states more readable, refs were moved to the end of each list and ordered by state referenced. With each update the order of candidates could change, frequently altering the numerical order of refs. The value of being able to easily find a particular ref among a dozen listed trumps the need to see refs in numerical order. It also facilitates cleaning up redundant and obsolete refs when they are in a predictable order. Bcharles (talk

) 15:50, 13 November 2016 (UTC)

Refs should be in order. If you have ^{[2][1][14][4]}, you're telling the reader to jump around a list of reference for no good reason. ^{[1][2][4][14]} is the natural way of presenting things. Now if some specific article warrants deviation from this for some obscure reason, then deny AWB on that specific article, but this is a legit fix in 99% of cases.

books

} 22:37, 13 November 2016 (UTC)

"Refs should be in order" - citation needed? There's no MOS statement to that effect. If the refs are presented as [2][1][4], we are telling the reader that [2] is the most important reference for that claim, [1] is the next most important, and [4] is the third. Unless the AWB operator actually reviews the references, there is no way to assess that importance. It is impossible for a bot to tell why the references might be in a particular order, so a bot should not change them. — Carl (

CBM · talk

) 22:43, 13 November 2016 (UTC)

There's no such 'importance order'. Either it supports the claim, or it doesn't. If two things support a claim, then both are equally important. If they aren't, then you are

books

} 22:46, 13 November 2016 (UTC)

Here's a preprint of a paper that was recently published in Annals of Mathematics, the premier mathematical journal. On page 2, notice the "[18], [10]". https://arxiv.org/pdf/0912.0325v4.pdf . I can find many more examples. Ordering consecutive references by importance rather than numerically is not universal, but is a well established style in the real world - and certainly an acceptable style on Wikipedia. — Carl (

CBM · talk

) 23:41, 13 November 2016 (UTC)

That's also a preprint who's first reference is [14].

books

} 01:57, 14 November 2016 (UTC)

Although I tend to agree that it is more usual in most cases to retain reference ordering and to think carefully about how articles are referenced, the danger is that bots (and bot like users) can create problems when they change stuff like this if there is the expectation that things will always be done in only one way. It's not the end of the world if things are out of order a bit - someone can always change it manually. I'd let it lie and let human being make the decisions - at least there is hope that they can be engaged in conversation. Blue Square Thing (talk) 22:48, 13 November 2016 (UTC)

I see no credible reason for a person, not familiar with the citations, to reorder them. Take it out of the guidelines and the bot, as soon as possible. — Arthur Rubin (talk) 23:56, 13 November 2016 (UTC)

• From the POV of cognitive processing, by the typical reader (i.e., a person who doesn't care about the refs and will not be clicking on any of them), ^[1][2][3] is easier/quicker to process (i.e., ignore) than ^[2][1][3].
• I suspect that Headbomb is correct when he says that 99% of these "wrong order" items are accidental/insignificant. Consider the case of <ref name"Alice"/><ref name"Bob"/>, which is in numerical order because it just doesn't matter – but then I re-arrange the top of the article, and "Bob" now happens to be ^[1] rather than ^[2]. Leaving it in the original order avoids a change to the wikitext, but the ^[2][1] order might be misinterpreted as implying that Alice's paper was more important or more relevant than Bob's, when it's purely accidental.
• On the point that Arthur makes about people unfamiliar with the citations deciding to re-arrange them, I've done that occasionally to make the wikitext easier to read. Especially in heavily sourced text with multiple citations that get used more than once, it's sometimes convenient to have all of the <ref name"Whatever"> together or consistently either at the start or the end. But some readers might believe that the order means something beyond the truth, which is that reading wikitext can be difficult. WhatamIdoing (talk) 02:48, 14 November 2016 (UTC)

• I strongly support the objections made above to any automatic re-ordering citations. There can be very good reasons for putting citations in a particular order in the text; editorial judgement should not be over-ridden in this way. Peter coxhead (talk) 08:40, 14 November 2016 (UTC)

Never accuse me of being closed-minded (treasonous, maybe). After reading the first somewhat cogent opposition arguments I have seen, I'm defecting and switching sides.
Case in point, the citations following the title above the diagram at

Thanks! I agree it's "a hack workaround to a problem that shouldn't exist", but at least it allows us to avoid the ridiculousness of automated action overriding editorial discretion. --Stfg (talk) 17:04, 14 November 2016 (UTC)

• There's consensus above that bots should not re-order references based solely on the idea that they must be in chronological order. I'm therefore going to add something along the lines suggested in my post of 20:52, 12 November. SarahSV ^(talk) 22:29, 17 November 2016 (UTC)

Hardly. There's a vocal minority that wants non-ordering, and they have a scheme to do it. Use {{

books

} 23:43, 17 November 2016 (UTC)

I haven't read the comments above thoroughly enough to know if there is a consensus in one or other direction; I'm just unlurking long enough to add my voice to the debate. I would prefer bots never re-ordered references. I'm as OCD as the next editor, but there's no value in having the refs in neatly ascending order. Perhaps out-of-order refs look odd to some people, but that's not enough reason to allow this. I don't mind if a human, not using an automation tool, reviews the refs and re-orders them for some reason; that's something I can take to the talk page if I happen to care about it. Reference order changes all the time when you work on an article, as you use a ref earlier in the article than originally. I really think it's harmless to leave them out of order, and since occasionally it's a considered decision by the editor who left them that way, it would be best to stop bots re-ordering them. Mike Christie (talk - contribs - library) 00:52, 18 November 2016 (UTC)

@Headbomb - Calling the people who disagree a "vocal minority" is inaccurate (and likely revealing of a kind of bias). The vocal minority could be equally claimed to be AWB operators who insist on re-arranging references they have never looked at with their own eyes. Repeating your opinion that references should be ordered in "99%+ of cases" does not make it more compelling. No "scheme" should be necessary to avoid the rearranging - bots should simply leave the references as they stand. — Carl (

CBM · talk

) 00:58, 18 November 2016 (UTC)

Headbomb, this was added to AWB with little consensus. You suggested it, Magioladitis agreed, JLogan objected, and Rjwilmsi added it. {{R}} only works for list-defined references. If you're dealing with references defined in the text, it doesn't help. SarahSV ^(talk) 06:05, 18 November 2016 (UTC)

The idea is that you would spend 5–10 minutes converting those refs to LDR. 1. If not already done, set up the References section for LDR. Shown at

WP:LDR. 2. Copy-and-paste the few refs involved from the body to the References section. (Note that an article can mix LDR and non-LDR usage, it is not a binary choice. I like adding a blank line before each ref for readability, but that's not required. I also like the LDRs in refname sequence for organization, but that's not required either.) 3. Add refnames there. 4. Add {{R}} in the body, replacing the refs. 5. Admire your great work (optional). ―Mandruss ☎

09:36, 18 November 2016 (UTC)

Never used that, but I was aware of it and I had the impression it was a tool to help convert the entire article to LDR. If so, it might be easier to do this manually, which also has the advantage of not having to learn a new tool. We already know how to use the wiki editor. ―Mandruss ☎ 10:03, 18 November 2016 (UTC)

I don't think a solution that requires users to convert to LDR can be said to be a solution unless you eliminate CITEVAR. Mike Christie (talk - contribs - library) 13:35, 18 November 2016 (UTC)

Please review CITEVAR, in particular the part in its first sentence where it says, "merely on the grounds of personal preference". Just in case, I'll reiterate that this does not require the entire article to convert to LDR, only the refs that need to be included in the {{R}}. ―Mandruss ☎ 15:12, 18 November 2016 (UTC)

Oh, if you were referring only to the "References segregator" tool, I misunderstood you and please disregard my reply. ―Mandruss ☎ 15:19, 18 November 2016 (UTC)

Using LDR at all should not be required to prevent bots from undoing what editors have done intentionally. The better solution to flawed bots is to fix the article, rather than trying kludges in the article to work around the bots. In general bots are held to a higher standard; see

CBM · talk

) 15:25, 18 November 2016 (UTC)

Mandruss, the language you're quoting from CITEVAR is only about changing the citation style; it doesn't say that personal preference should not apply to considerations of keeping the existing style. Further down it says the existing style should be deferred to unless a consensus for change can be reached on the talk page. (I don't think anyone has ever asked me to change the citation style on an article I've worked on, but if they did I'd probably be amenable to it.)

There are editors who would not want LDR on the articles they work on, and consensus to change would not happen; and CITEVAR would be cited to support those editors. Hence I really do think that a requirement to use LDR would fall foul of CITEVAR and can't be considered a general solution to this problem, though of course if bots keep reordering references, it's a solution for any individual article where the editor is happy to use LDR. Mike Christie (talk - contribs - library) 15:37, 18 November 2016 (UTC)

No one should be required to use LDR; see

WP:CITEVAR. Expecting people to use LDR or special templates to avoid AWB moving refs wouldn't gain consensus. SarahSV ^(talk)

22:00, 18 November 2016 (UTC)

No one is required to. It's not moving refs it's putting them in the natural order which the vast majority of published sources use. Moreover in almost every case the change is made because the numbering has changed, and the article previously matched the normal monotonic numbering system. All the best: Rich Farmbrough, 05:08, 19 November 2016 (UTC).

Why is a "give and take" required with AWB doing something that is not consensus? Bots have no business in doing stuff that isn't consensus - period!--Kmhkmh (talk) 10:58, 19 November 2016 (UTC)

I guess I haven't fully understood the context here, as I haven't been involved in a lot of the history regarding this issue. After re-reading the opening in this thread, which I failed to comprehend the first time, it seems that the proponents of reordering assert that reordering is inherently and self-evidently a Good Thing and so it does not need consensus. They therefore put the consensus burden on the opposition. That would be asinine, as very little is inherently and self-evidently a Good Thing (aside from a good medium rare steak). If I have this right, I hereby defect and switch sides AGAIN, and I encourage the AWB folks to (1) remove or temporarily disable cite reordering, and (2) seek a strong consensus for these edits. If reordering is a Good Thing, said consensus will not be that difficult to achieve. If consensus cannot be achieved, reordering is not a Good Thing, by definition. That's how Wikipedia decision-making works (and I'm both surprised and disappointed to see experienced editors who do not understand that). Failure to do that would be a matter for

WP:CONSENSUS above everything else. Having technical control over a community tool, in particular, comes with a responsibility to the community as a whole, not just those who agree with your viewpoint, and I don't see that responsibility being met here. Persistent failure to meet it should result in a ban from technical control of the tool. ―Mandruss ☎

11:41, 19 November 2016 (UTC)

My experience is that, in the vast majority of cases, refs go out of order because the article has been edited. I'm sure there have been a number of cases where the out-of-order was deliberate - we have one, viz: the US election page (which was not for editorial reasons but for ease of editing - and was probably bad editorially for other reasons).

We can insist that this (or these) examples mean that refs should never be re-ordered without talk-page discussion, or that they can never be re-ordered by a bot without discussion, or that BRD applies. It seems to me that most of the solutions suggested are simple ways of ensuring that the R part of BRD works. Except in pathological cases (like 50 refs in the election example) inserting a comment <!-- please do not reorder --> is a trivial effort for something that is supposed to be so important.

All the best: Rich Farmbrough, 14:28, 19 November 2016 (UTC).

That puts the cart before the horse. For example,

CBM · talk

) 14:44, 19 November 2016 (UTC)

@Rich Farmbrough: That's some pretty fancy reasoning, Rich, but it doesn't wash. An RfC can consider your argument and others, that's what RfCs are for. You don't get to argue that because of x, y, and z, no consensus is needed, it don't work that way. Consensus is required because there is significant and heated disagreement between experienced good-faith editors. Period. The consensus burden here lies with those who wish to reorder (it's somewhat ridiculous to say that we need consensus to do nothing with respect to cite order), and responsibility for starting the RfC lies with those who should have started it in the first place, those who wish to reorder. I suggest they start it as soon as possiible rather than wasting a lot more time in this thread, and it should be a high-visibility RfC (simply listing it will not be enough). But, if necessary, I'll start it myself soon unless I see a far better argument against such an RfC. I support the essay WP:Process is important; those who haven't given it a read might consider doing so. ―Mandruss ☎ 16:14, 19 November 2016 (UTC)

If it helps simplify this, we can frame the whole dispute in a BRD context. Whenever reordering was first added, that was the B. I'm guessing that couldn't be challenged with a revert, but it was certainly challenged, and that was the R. That required those who wished to make the "edit" to seek consensus for it. Since it couldn't be reverted, they had a responsibility to reverse the change themselves pending consensus for it. They failed to do either, instead abusing their control of the tool to evade that burden. I'm not sure what is lacking there, competence or ethics, but it was wrong. ―Mandruss ☎ 16:47, 19 November 2016 (UTC)

I don't think calling people's competence or ethics into question is useful. All the best: Rich Farmbrough, 18:27, 19 November 2016 (UTC).

It's certainly not out of line if there is evidence to support it, and I feel I made a good faith case for that. As for "useful", I guess that would depend on how receptive the targets are to constructive criticism. Some editors reflexively and incorrectly cry "AGF FAILURE!!" the second you offer any, but they are generally noobs. ―Mandruss ☎ 20:26, 19 November 2016 (UTC)

To get back to the topic,

CBM has clearly identified the core issues: should bots be allowed to make context-sensitive changes, and should editors be required to insert HTML comments to stop them. As is clear from my earlier comments, my answer to both is "no". Peter coxhead (talk

) 19:15, 19 November 2016 (UTC)

Get back to the topic? Correct process is the topic here as far as I'm concerned, and it's clear enough to me that out-of-RfC discussion of this issue is not going to reach a resolution. As I understand it, we've been trying that on and off for seven years. Therefore I have started the RfC at Wikipedia:Village pump (proposals)#RfC: AWB bot ref reordering and I invite everyone to participate there. Thank you for your service. ―Mandruss ☎ 21:16, 19 November 2016 (UTC)

Mandruss, I agree that the people who like this feature should start an RfC if they want to keep it, and that the onus is not on those who oppose it. It shouldn't have been added without a broader consensus, and according to AWB's own rules, it shouldn't have been executed over objections. (That AWB's rules are regularly flouted is the larger issue here.) But the consensus in this discussion is clear; consensus doesn't mean that every single person agrees. It's worth noting that Magioladitis has also supported removing this feature.

Rjwilmsi said he would remove this from AWB if consensus were reached here. He wrote: "So the simplest thing here would seem to me to have a wide discussion about the encouraged/approved/discouraged options for reference ordering on the MOS/WP:CITE pages, ensure that the MOS/WP:CITE etc. guidelines are updated if a consensus is reached. Then I will ensure AWB is updated if required to support the consensus reached." SarahSV ^(talk) 22:15, 19 November 2016 (UTC)

@SlimVirgin: Could be, but some editors appear to dispute that consensus. This is a common dilemma without an outside closer, do we or do we not have a consensus? Those who don't like the result of said consensus almost invariably say there is no consensus; I suppose that's just human nature. That often comes down to a question of editor personalities, which is no way to resolve a discussion. I suppose we could have requested an outside closer for this thread, but the added structure of RfC makes that job far easier for them and I think results in a higher quality close. I think we should have been there in 2009.
As far as the onus for starting the RfC, I agree, but the solution to that is just to start the damn thing rather than to demand that they do so. Same concept as in article disputes. ―Mandruss ☎ 22:24, 19 November 2016 (UTC)

Seventeen people commented; 11 support the view that bots should not reorder refs. SarahSV ^(talk) 22:47, 19 November 2016 (UTC)

Some might dispute that 64% represents a consensus for something like this. Besides, consensus is about strength of arguments and not just numbers, and of course everybody thinks they have the stronger argument. Hence, outside closer. ―Mandruss ☎ 23:01, 19 November 2016 (UTC)

When Rich objected that there was no consensus, my intention was to let the discussion continue, then request an outside closer. It would have been good to wait for a close here before opening an RfC elsewhere. (But now that you have, please don't reverse it.) SarahSV ^(talk) 23:07, 19 November 2016 (UTC)

Clearly, I have no way to know your intentions if you don't say anything about them. I am a very poor mind-reader, especially remotely via the Internet. I mentioned RfC twice in this thread before I decided to start one, and there was only one response, not from you, that failed to convince me. Communication is important. ―Mandruss ☎ 13:40, 20 November 2016 (UTC)

• Noting that the RfC was closed on 24 December with a two-thirds majority against using AWB to reorder references. SarahSV ^(talk) 20:08, 26 December 2016 (UTC)

Greetings to everyone concerned about the reliability of sources used in the Wikipedia. For the 2016 Community Wishlist Survey, I have created a proposal that addresses some aspects of this called

Cross-posted from Help talk:Citation Style 1; please discuss there. Thanks. —67.14.236.50 (talk) 19:35, 27 November 2016 (UTC)

Answered! —67.14.236.50 (talk) 12:54, 28 November 2016 (UTC)