Wikipedia talk:Double redirects

MediaWiki deficient?

If Wikipedia terminates a redirect as a precaution to infinite loops, they should come up with a better plan than that.

My plan is to improve the Wikiware to automatically detect a self-redirect to stop an infinite loop but allow a sequence to other redirects until it reaches the terminus. --SuperDude 05:54, 14 May 2005 (UTC)

Is there anyone working on removing the restrictions on double-redirects? Where might I find more information? Ewlyahoocom 22:14, 8 February 2006 (UTC)

Er, yes, that's what I thought. Loop detection is a solved problem. All you do is remember every page you've redirected through and see if you hit one again. As a back-end check you would have a maximum hop count of some reasonable value (like 20)? Easy-peazy. Why should we have to compromise the logical structure of Wikipedia (or any wiki) for the sake of a few lines of code and a few CPU cycles? Duckbill 09:40, 20 March 2006 (UTC)

Functioning redirect chains are expected default behaviour and are sometimes the correct way to set up complex redirects. The WikiMedia software is currently broken in this regard. The fact that the breakage seems to be a deliberate act on the part of the programmers doesn't make it any less frustrating (or any less broken).

If wikipedia doesn't want to allow multiple redirects as a matter of site policy then that's fine, but for those of us trying to use the software for other wikis shouldn't have our hands tied. If the programmers are worried about infinite redirect loops, then the very least, there should be a wiki-level system setting for "max number of redirects = n" , where "n=1" gives Wikipedia's behaviour, n=2 allows double redirects but blocks triples, n=3 allows up to three redirect stages (etc). I don't expect I'll need more than two on my project, but to be artificially limited to just one is silly. ErkDemon (talk) 00:28, 1 July 2011 (UTC)

I may be wrong, but I thought the developers had introduced such a variable. (The reason it's not set to a higher value on WP is that there is then no system for generating a list of overlong chains similar to Special:DoubleRedirects; and no-one seems to be in a hurry to solve this problem.)--Kotniski (talk) 09:46, 1 July 2011 (UTC)

Yaaay! Thanks Kotniski! Found it, and it seems to work ( http://www.mediawiki.org/wiki/Manual:$wgMaxRedirects , since v1.15, apparently). And my apologies to the MediaWiki programmy people. ErkDemon (talk) 19:00, 12 July 2011 (UTC)

Page move

I propose that this be moved to Wikipedia:Double redirects. To my recollection, I never even saw the term "multiple redirect" before I stumbled onto this page a bit earlier. (Well, maybe I saw it once.) Even the page itself consistently uses the term "double redirect", except for the initial defining sentence. Eric119 04:59, 24 Jun 2005 (UTC)

• I will implement this move. Eric119 16:50, 11 July 2005 (UTC)

prevention is better than cure

Whenever we move a page, we are reminded to check for double re-directs and fix them, if any. However, whenever we merge two articles, and convert one of them into a re-direct, we do not get reminded to check for double re-directs. We should probably be reminded whenever a re-direct is made so that this problem can be remedied by the initiator of the re-direct himself. --Gurubrahma 10:36, 1 November 2005 (UTC)

...in order to prevent infinite loops.

I believe the phrase "...in order to prevent infinite loops" should be removed from this description. The limitation is simply a limitation (and there are better ways to avoid infinite loops than this kind of deficiency -- I even believe following double-, triple-, or whatever-number-of-redirects could be done within a single existing SQL statement without adding any additional looping in the software). Ewlyahoocom 22:27, 21 May 2006 (UTC)

I prefer to keep the explanation about infinite loops. It's only a few words. I prefer to see an explanation of why a limitation is there, rather than just be told that there's this limitation. There may be better ways to do it; maybe the development team will change it when they have time. You could, if you want, start up a discussion perhaps at the technical section of the village pump or on Meta, or perhaps put in a bugzilla feature request, though I'm guessing the development team is already aware of the problem. If it's done with a single SQL statement there would be implicit or explicit looping within the SQL statement; perhaps the SQL interpreter automatically checks for and avoids infinite loops. I don't know the details and I'm guessing there's some reason the development team doesn't find it easy to change this, or they would have done it already. --Coppertwig 13:27, 18 February 2007 (UTC)

Fix double redirects automatically

Would it be possible for the database to fix double redirects by itself, without relying on us to find them and edit them? --Smack (talk) 18:16, 25 January 2007 (UTC)

Ask it on Bugzilla. Melsaran (formerly Salaskаn) 13:55, 4 August 2007 (UTC)

Added section "Checking for double redirects"

I just added instructions on how to check for double redirects. These instructions may seem obvious and unnecessary; but let's just say that some of us are the type of people that if we write a program that handles linked lists we need to draw little diagrams with boxes and arrows, and if we write another similar program a few weeks later we need to draw the diagrams all over again. In other words, it may be obvious to some people but confusing for others. I hope these instructions will remain here so I can re-read them next time I move a page. --Coppertwig 13:19, 18 February 2007 (UTC)

Contested move request

The following request to move a page has been added to Wikipedia:Requested moves as an uncontroversial move, but this has been contested by one or more people. Any discussion on the issue should continue here. If a full request is not lodged within five days of this request being contested, the request will be removed from WP:RM. —Stemonitis 06:47, 4 August 2007 (UTC)

• Wikipedia:Double redirects → Wikipedia:Double redirect — WP:NC#Prefer singular nouns —Melsaran (formerly Salaskаn) 04:13, 4 August 2007 (UTC)

• It is unclear to me that pages in the Wikipedia namespace must be bound by the same rules as article titles. If there is thus little point in making the move, then the disruption it might cause is reason enough not to move the page. --Stemonitis 06:47, 4 August 2007 (UTC)

• I see no reason to keep it in plural, either. We also have Wikipedia:Redirect, etc. Melsaran (formerly Salaskаn) 13:27, 4 August 2007 (UTC)

• Opposed this isn't an article, WP:NC is for articles. Even the WP:NC page name is pluralized. 132.205.44.5 23:19, 6 August 2007 (UTC)

• Yes, because there are multiple naming conventions on it. Melsaran (formerly Salaskаn) 04:24, 7 August 2007 (UTC)

Byrial method?

The article mentions that the "Byrial method" builds on the redirect (Topbanana) code. What is the Byrial method, and why is it not explained or linked? Cww 03:29, 19 September 2007 (UTC)

redirects to singulars with possibilities

I'm being told that under this policy, pattern-avoiding permutations must not redirect to pattern-avoiding permutation, which is tagged as a redirect with possibilities. If the latter becomes an article, then pattern-avoiding permutations would be left redirecting to something other than pattern-avoiding permutation, and those clicking on "what links here" at the latter page would never find out. That is absurd. At least one bot that fixes double redirects will NOT "fix" them in cases where there's a "redirect with possibilities" tag, but there's another that does (user:computer) and it's owner refuses to override it in such cases as this. Michael Hardy 20:54, 19 September 2007 (UTC)

I've already provided Michael Hardy an alternative that accomplishes what he wants without having to have broken redirects. It is trivial to add an HTML comment (or simply text as Wikipedia will ignore anything after the redirect) after the redirect that mentions the others. Example:

#redirect [[Stanley-Wilf conjecture]]

<!-- If you change this to an article, please be sure to modify [[Patern-avoiding permutation]] & [[Patern-avoiding permutations]] to redirect to here -->

It is easy to solve these concerns without having to have double redirects that confuse our readers. -- JLaTondre 21:07, 19 September 2007 (UTC)

Michael Hardy chose to place a text with links after the redirect code.[1][2] The text is not displayed and the redirect works so our readers should not be confused, but Special:Whatlinkshere/Stanley-Wilf conjecture may confuse editors (at least it confused me). It falsely looks like there are double redirects having to be fixed. PrimeHunter 01:57, 20 September 2007 (UTC)

If this is really necessary (and it may be) he should be adding something like this {{R from alternate name|Pattern-avoiding permutation}}, which would at least "look" right (even if it doesn't do anything). I would prefer if the Wiki software could be modified to at least handle just 1 extra level of redirection, which would settle this and a lot of other issues. Ewlyahoocom 02:07, 20 September 2007 (UTC)

bots are often idiots; they should be supervised by thinking humans

The redirect page titled "ML Inequality" redirects to "estimation lemma". Note that it has a capital initial "I". If "ML inequality" (with a lower-case initial "i" should ever become an article rather than a redirect that points (like "ML Inequality") to "estimation lemma" then "ML Inequality" (with a capital "I") should be edited to redirect to "ML inequality" (with a lower-case "i"). Therefore, this redirect page reads as follows:

---

#REDIRECT[[estimation lemma]]

If [[ML inequality]] ever becomes an article rather than a redirect, then this page should be changed to redirect to that.

---

It works fine as a redirect page. But this bot called Computer changed it to read as follows:

---

#REDIRECT[[estimation lemma]]

If [[estimation lemma]] ever becomes an article rather than a redirect, then this page should be changed to redirect to that.

---

Now think about that. The owner of the bot doesn't think that's a problem. And he cites THIS style page, on double redirects, to justify his bot's behavior. That's the most unreasonable view on design of Wikipedia that I've seen in a long time. If THIS page requires that behavior of that bot, I propose to alter it so that it still forbids double redirects but doesn't require that idiotic edit. Michael Hardy 23:28, 7 October 2007 (UTC)

Is this done automatically?

Special:DoubleRedirects says It is not necessary to fix these by hand. Bots will go through the entire list periodically and fix all of the double redirects. Is this true? If so, why do the move pages give stern warnings that fixing double redirects is your responsibility?--Yannick (talk) 15:52, 24 November 2007 (UTC)

Bots will fix double redirects on that list. However, that list is generated periodically which means that a double redirect can exist for sometime before it gets fixed. The move page directions are to avoid having readers being confused by double redirects in the period between the move and a bot being able to fix them. The bots are to fix ones that are missed; they shouldn't be used as a shortcut as that would be a disservice to our readers. -- JLaTondre 15:08, 26 November 2007 (UTC)

Some of these page moves can produce hundreds of double redirects. That's a lot of tedious work to do by hand, something of a disservice to our editors to put this responsibility on them. How often do those bots run?--Yannick (talk) 02:05, 28 November 2007 (UTC)

It's not an issue of how often the bots run, but rather how often the double redirects page is generated. That's generated by the Wikipedia software. I think it's done daily, but I'm not positive. I gather that it is database intensive. I doubt there are many moves that "produce hundreds". The ones that are overwhelming can obviously be left to the bots. However, the nominal case of a handful should just be taken care of by the editor. -- JLaTondre 13:20, 28 November 2007 (UTC)

MediaWiki software automatically fixes things

bugzilla:4578, "Page moves should not create double redirects", has been implemented, per Wikipedia:Wikipedia Signpost/2008-07-28/Technology report#New features and this posting at Wikitech-L. That will sharply rduce the number of double redirects. (It's possible to tell the software to let a double redirect be created, by unchecking a box before doing a move, but it seems fairly obvious that few editors will do that.) In particular, it's now unnecessary for editors, after moving a page, to bother checking for double redirects that were created by the move. -- John Broughton (♫♫) 17:32, 7 August 2008 (UTC)

Many double redirects are good

At Wikipedia:Village_pump_(proposals)/Archive_44#Double_redirects, Kotniski, Happy-melon and others provided examples for good double redirects, which all seem to fit the following formula:

• X' → X → Y

In a common situation ("type 1"), X' is an alternate spelling of X, and Y is an article that, for some reason, currently is the best we have on X. This could be because it contains something about X, among other topics.

A related situation ("type 2") also fits the above formula is in Kotniski's example:

• Mosial → Mósiał → Jan Mósiał

Here, X is a more general title than Y that might need to be disambiguated in the future.

A majority seemed convinced by that, and agreed with allowing longer redirect chains. Despite that agreement, nobody thought of changing this page then. It would be bad enough if this were only a question of wrong recommendations we give to our editors, but that could be mitigated by the fact that most users can use their common sense. The situation has become worse because we now have bots which are busily and indiscriminately "fixing" all any and all double redirects. I recently became aware of this problem (see User talk:Xqt#Xqbot is edit warring), and did a little bit of research: I counted 20 double redirects "fixed" by the bot. Out of these, I found seven - or 35% - that fit type 1.

We need to take that reality into account and do the following:

1. Change the wording of this page to clarify that not all double redirects need to be "fixed".
2. Fix the bots so that they stop "fixing" double redirects indiscriminately.

— Sebastian 00:32, 24 August 2009 (UTC)

Unless something has recently changed in how Mediawiki works (as configured for EN Wikipedia), the suggested actions seem pointless because double-redirects do not work. Unless and until the software changes, there is little point suggesting that some double redirects should not be fixed. older ≠ wiser 01:31, 24 August 2009 (UTC)

There was a recent period during which double redirects did work, and there was a discussion that seemed to produce consensus to keep that, but nothing ever happened. --NE2 02:01, 24 August 2009 (UTC)

It is very unfortunate that the software has this problem, and I admit that this situation makes my proposed changes somewhat less opportune now. (I wasn't aware that the software lapsed back.) It is not nice when a user has to click twice to get the page he or she wants. However, the problems we're creating are bigger than the problems we're fixing: For every 2 correct fixes, we have 1 where the logical connection is mindlessly destroyed. I don't think it is wise to automatically and indiscriminately destroy what people created for a reason, just to work around the momentary quirks of the software. We need to stop this destructive hack. — Sebastian 04:01, 24 August 2009 (UTC)

While I honestly believe that double redirects should work, but as long as they don't we need to live with what we have. It can't hurt to start making a list of wanted double redirects, or a potential policy/guideline for when to use them. עוד מישהו Od Mishehu 07:09, 24 August 2009 (UTC)

We should keep badgering the developers to fix this (there is a bug report at bugzilla somewhere). The thing that needs fixing, AIUI, is not the doube redirects themselves (these could easily be enabled at the flick of a switch), but the reporting of overlong chains of redirects. At the moment we have Special:DoubleRedirects to spot and fix them; if the maximum chain length were increased (to 3 or 10 or whatever value of n) then we would need an equivalent Special page to spot and fix chains of length n+1. It's that functionality that's lacking.--Kotniski (talk) 08:55, 24 August 2009 (UTC)

bugzilla:17888, as mentioned in the original discussion, and it's still open. Amalthea 11:14, 24 August 2009 (UTC)

I don't think living with what one has is a very helpful metaphor in an environment that's built on change. You can't take that literally, unless you include our intelligence and ingenuity among the things we have. Certainly we do not need to keep the wording of this page, nor do we need to keep a horde of bots running amok. The edit war which I had with one bot, which I thought was over once and for all, now got taken up by the next.[11] This is just like the science fiction nightmare of robots taking over the world, albeit only on our little world of wikipedia.org. — Sebastian 13:58, 24 August 2009 (UTC)

I'm not sure if "broken" double redirects are preferable though, speaking from a reader's point of view. Amalthea 14:57, 24 August 2009 (UTC)

Ah, that was an eye-opening remark! It seems, we're caught in a self-perpetuating meme here. People have come to call double redirects "broken link" - something every responsible web developer hates! I think people here talked each other in a panic that makes one problem appear exceedingly big, and lets them ignore the other problem completely. But look at "broken link" - that's not what we're talking about. Even the worst case, when a reader has to click one more time, is not "broken" by any stretch of the imagination. And it may not even necessary to click the second time. Just look at an example from my list above: Reader sees Kurenai Tsubasa, doesn't know what that means, and clicks on it. Up comes a page that says "→ List of Ranma ½ characters#Tsubasa Kurenai". That's already good information, that may even be all the reader wanted to know. Is avoiding such "problems" really worth destroying our logical link structure? — Sebastian 15:21, 24 August 2009 (UTC)

I'm afraid you've lost me. Are you now saying that it is a good thing to have Kurenai Tsubasa be a simple redirect to meaningful content rather than a broken redirect to Tsubasa Kurenai (which itself redirects to the same content)? That seems to contradict what you've been saying before. And to be clear, no one is talking about broken links -- I think everyone commenting here understands what a broken redirect is. older ≠ wiser 15:57, 24 August 2009 (UTC)

No, what I'm saying is that calling double redirects "broken redirects" is making mountains out of molehills. I understand that it's hard to break free of such a view, especially when, as you say "everyone ... understands" it that way. Maybe I'm a just bit more aware of what language can do to us, having grown up in a country that had a history of abusing it, and where it was a vital part of the curriculum to make us aware of the mass hysteria that can be caused by emotional buzzwords. — Sebastian 16:27, 24 August 2009 (UTC)

A broken redirect is a redirect that does not function. Where is the pernicious effects of misusing language? Your opinion that forcing readers to make extra clicks through such broken redirects is not uncontroversial, and although there is significant support for enabling double- (or multi-) redirect functionality, forcing readers to use broken redirects is a separate question which does not appear to have the same level of support. And in any case, you didn't really address my question above. Your recent example appears to contradict your earlier statements. That is rather confusing. older ≠ wiser 16:35, 24 August 2009 (UTC)

Sorry, I didn't mean to say that some evil person intentionally coined a misleading term. I understand how one can get to that term. But it is misleading, because the redirect does function, albeit not as well as we would want it to. Calling it "broken" is adding unwarranted emotional ballast. If anything is broken, it is not the redirect, it is our software. (Maybe it helps to illustrate this with an example: Imagine you're driving a car with a broken gear shift, and you have to always stay in first gear. If you took the highway, you would get into trouble, so you take a detour instead. Calling the redirect broken is like calling the highway broken, instead of the car.) And while I'm trying to open your eyes about one such emotionally loaded word, you're introducing another. The word "forcing". Again, I understand how you can feel this way, but please let's keep things in proportion. — Sebastian 16:59, 24 August 2009 (UTC)

So, now it appears that you are unnecessarily distorting the language in order to make a point. A double-redirect at present does not function as expected. The simplest sense of that is that it is broken. I'm sorry that that doesn't support your opinion, but to describe it as anything other than broken seems like Orwellian double-speak. older ≠ wiser 17:09, 24 August 2009 (UTC)

OK the redirect also breaks "related changes". Even one level of redirect does this AFAIK. Now back to the original point. If we think information is being lost all we need to do is ask the double-redirect bots nicely to populate a suitable field. E.G. {{R from double}}. Then when double (truple, whatever) redirects become available again we can unwind the redirects. Or indeed if MΘΝΠ becomes an article or a dab. Rich Farmbrough, 04:22, 30 August 2009 (UTC).

Thank you for your post and for understanding my concern. Template:R from double is a good idea; that could also contain a {{nobots}}.

I hadn't thought about the "related changes"; why do you say it breaks them? Do you mean is that if we have * X' → X → Y, and you watch related changes for Y you won't see a change in X'? This is not necessarily a problem, since X is not the same as Y. Conceivably, you may be interested in changes related to Y, without being interested in changes related to X. Admittedly though, if you're using "related changes" to watch for vandalism, then it wouldn't hurt to be alerted to changes in X'. But since that's only a redirect and unlikely to incur a lot of changes it is also much less likely to be changed, and much less of a problem if you miss a change. — Sebastian 22:19, 31 August 2009 (UTC)

Relying on bots

I've noticed that bots now appear to be more reliable and frequent (once per day) in fixing double redirects. Should I now rely on bots to fix multiple double redirects, or should I continue to correct multiple double redirects myself out of Wiki courtesy? Tinlinkin (talk) 09:08, 18 May 2010 (UTC)

I followed the posted advice that appears after I move a page, to fix only the most important double redirects and let the bot take care of the rest later. That must be a relatively new message, and I'm relieved by it. Tinlinkin (talk) 04:38, 1 June 2010 (UTC)

WikiGnomes

I think that a WikiGnome would fix double redirects (I don't know, but it seems like it would be in their nature.) If they do, it should be noted. — Preceding unsigned comment added by 74.99.167.211 (talk) 01:29, 4 October 2011 (UTC)

Noting...

30xelawalex03 (talk) 23:38, 11 March 2012 (UTC)

Special page

  A redirect....special page....

Tis could be misleading. They are pages that are special, but calling it a "special page" could be misleading. It is edited, has wikitext, etc. etc. So it is not a special page on Wikipedia.

Walex & 03. A Life together. (talk) —Preceding undated comment added 23:10, 26 February 2012 (UTC).