Policy | Technical | Proposals | Idea lab | Miscellaneous |
New ideas and proposals are discussed here. Before submitting:
|
« Older discussions, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131 |
Centralized discussion | |||
---|---|---|---|
Proposals: policy | other | Discussions | Ideas |
For a listing of ongoing discussions, see the dashboard.
Note: entries for inactive discussions, closed or not, should be moved to the archive. |
|||
Contents
- 1 Banter
- 2 Religious messages
- 3 Implementing Help:Maintenance template removal
- 4 Watson and next generation improvements to the standard Wikipedia search box
- 5 Auto-assessment
- 6 Revive WP:ABUSE, but completely revamp how it does business.
- 7 Destubbing articles
- 8 RfC: Preferred protocol for external links
- 9 The disambiguation gap has been bridged!
- 10 Election-year prophylactic on coatracking, shoehorning, and the like.
- 11 Save page -> Publish
- 12 Draft prod
Banter
I don't know how many people feel like me but I've always wanted Wikipedia to be a fun kind of place. We get work done, but let it not feel like drudge. So, I propose we make a page for banter where we can get together and do some shiz. I don't know how will other people will feel about this but hey, atleast I've spoken my mind. --QEDK (T 📖 C) 16:53, 27 March 2016 (UTC)
- Maybe Wikipedia:Village pump (banter) ? All the best: Rich Farmbrough, 22:06, 27 March 2016 (UTC).
- We had that many years ago. It was called Wikipedia:Esperanza. It was killed with pitchforks and torches for upsetting the local villagers. --Jayron32 02:09, 28 March 2016 (UTC)
- It seems to have dissolved into some useless bureaucracy that caused it to be brought down. I went through it and why not just make a coffee lounge kind of place? (If enough people support this proposal, I'll frame some rules and make one) --QEDK (T 📖 C) 14:26, 28 March 2016 (UTC)
- I think it could be fun to have a relaxed kind of place on here where it wouldn't be all about AfD fights, vandalism and other unpleasant aspects of the encyclopedia. I would support it. White Arabian Filly Neigh 21:53, 29 March 2016 (UTC)
- I've seen other online entities that had a "work" part and a "banter" part. The banter parts were so lightly used that a little banter remained in the "work" part, because if it were put in the banter part hardly anyone would see it. Jc3s5h (talk) 12:51, 30 March 2016 (UTC)
- I think this would be a good thing to have. We need a place to see our fellow editors as human beings and make links. In fact, it would help me no end in my NPP and newbie welcoming tasks if I knew some editors working in various areas. However, get ready for the shouts of wp:nothere :) Happy Squirrel (talk) 01:16, 31 March 2016 (UTC)
- I've seen other online entities that had a "work" part and a "banter" part. The banter parts were so lightly used that a little banter remained in the "work" part, because if it were put in the banter part hardly anyone would see it. Jc3s5h (talk) 12:51, 30 March 2016 (UTC)
- I think it could be fun to have a relaxed kind of place on here where it wouldn't be all about AfD fights, vandalism and other unpleasant aspects of the encyclopedia. I would support it. White Arabian Filly Neigh 21:53, 29 March 2016 (UTC)
- It seems to have dissolved into some useless bureaucracy that caused it to be brought down. I went through it and why not just make a coffee lounge kind of place? (If enough people support this proposal, I'll frame some rules and make one) --QEDK (T 📖 C) 14:26, 28 March 2016 (UTC)
- Wikipedia is not a social network. Such a page cannot be reconciled with our policies. RGloucester — ☎ 01:58, 31 March 2016 (UTC)
-
- Given that the goal of this banter page would be to increase collaboration by allowing less structured interraction, the end goal does focus on Wikipedia. I would tend to see this more as similar to meetups being organised on-wiki. From reading the policy you linked, I rather got the impression it was forbidding using Wikipedia as a place to conduct one's off wiki social life, not as forbidding social interractions with people we are contributing to Wikipedia with. Happy Squirrel (talk) 02:11, 31 March 2016 (UTC)
- It's actually a general content policy, I see no reason why not till now. But 5 voices are barely it. --QEDK (T ☕ C) 10:31, 4 April 2016 (UTC)
-
- What does a suggestion like QEDK's have to do with a social network ? Can we please add a WP:NOTAFUCKINGSTERILEOPERATINGROOMEITHER ???? —TheDJ (talk • contribs) 17:38, 28 April 2016 (UTC)
- Indeed, any serious endeavour requires a serious approach. To do anything else is to invite mockery, and furthermore, to distract editors from the mission of this project. As soon as editors start turning this project into a social game where one's goal is to acquire social relations or social status, that will result in a nothing more than a social club, and a social club is not compatible with the serious task set before us. RGloucester — ☎ 17:43, 28 April 2016 (UTC)
- It's scary how some people think every social thing is a game not to be taken serious and dismissed up front with a wiki shortcut. It's like erecting a 6 meter high wall every single time someone mentions "mexico" or "palenstine". It's too easy, oversimplified and misguided response. To pretend we have no social aspects is to bury your head in the sand. We have social aspects and the 4th pillar even affirms this (even though we seem to ignore it most of the time). Userboxes, decorated user pages, FA and GA badges, barnstars and decorated signatures also all have social aspects, that we are have accepted. To talk to someone on his talk page instead of on a forum is a social choice. To pretend that we are not social is thus a fallacy, and replying to every single thing, that potentially has a social aspect to it, with a NOTSOCIAL link is even more so a fallacy. The world is not that simple. We are not Facebook indeed, we are not Twitter, we are Wikipedia, a collaboratively written encyclopedia by people who thus form a social network. Just not the Internet age one. —TheDJ (talk • contribs) 18:10, 28 April 2016 (UTC)
- Indeed, any serious endeavour requires a serious approach. To do anything else is to invite mockery, and furthermore, to distract editors from the mission of this project. As soon as editors start turning this project into a social game where one's goal is to acquire social relations or social status, that will result in a nothing more than a social club, and a social club is not compatible with the serious task set before us. RGloucester — ☎ 17:43, 28 April 2016 (UTC)
- What does a suggestion like QEDK's have to do with a social network ? Can we please add a WP:NOTAFUCKINGSTERILEOPERATINGROOMEITHER ???? —TheDJ (talk • contribs) 17:38, 28 April 2016 (UTC)
-
- It's actually a general content policy, I see no reason why not till now. But 5 voices are barely it. --QEDK (T ☕ C) 10:31, 4 April 2016 (UTC)
- Given that the goal of this banter page would be to increase collaboration by allowing less structured interraction, the end goal does focus on Wikipedia. I would tend to see this more as similar to meetups being organised on-wiki. From reading the policy you linked, I rather got the impression it was forbidding using Wikipedia as a place to conduct one's off wiki social life, not as forbidding social interractions with people we are contributing to Wikipedia with. Happy Squirrel (talk) 02:11, 31 March 2016 (UTC)
-
-
-
-
-
-
- A social club is exactly what has built the encyclopedia. All of those discussions through which policies are made, admins are elected, naughty contributors are banned, edit wars are resolved, newbies are helped, and collaboration on articles occurs, are all elements of the social aspect of Wikipedia and are key to the functioning of this project. Many editors will also tell you that the only reason they stayed involved was because of the opportunity for interaction with their peers, through talk pages and IRC. Humans are social creatures, and we need to have that potential for interaction for this site to continue. However, because of this prevailing desire to not be a social network (and by this I think people mean not be Facebook), this side of the project has been seriously neglected and has not evolved since 2001. That's a problem. General comment, no preference for or against the proposal as I wouldn't use it. Ajraddatz (talk) 18:28, 28 April 2016 (UTC)
-
-
-
-
-
- Support as log as it is just a room or two in which a few admins can act as moderators... nothing else is needed. Fritzmann2002 18:31, 7 April 2016 (UTC)
I like the idea, but the possibility of vandals doing whatever they want is just too great. ThePlatypusofDoom (talk) 21:00, 7 April 2016 (UTC)
I'm totally behind this! It would need some moderation of course, but I'm sure that can be sorted out. Commissaress (talk) 21:25, 7 April 2016 (UTC)
- All but died on the vine in January 2015, here, albeit with a slightly different context/angle. Beware the scarlett letter P(erennial). ―Mandruss ☎ 21:50, 7 April 2016 (UTC)
Successful WikiProjects usually have some chatter going. You might find a large WikiProject in an area that interests you, and join in. Alternatively, some of this happens on a few users' talk pages. WhatamIdoing (talk) 05:29, 13 April 2016 (UTC)
- I can see how meeting and greeting fellow Wikipedians could be useful, especially for newer editors, and along those lines would see a place for it in a back room of the teahouse.
- For more established editors, an extension to userbox induced categorization facilitating editors' search for and interactions with Wikipedians with specific attributes could be handy, where rather than simply finding a list to individually pester, one could open open discussions with all of them on a dedicated
bantersymposium page. - Overall, I can see the encouragement to let one's guard down and blow off steam turning rapidly into an arena where everyone's guard is forced up, and folk get burned. It may seem an officious place at times, but considering how busy it is, the rules have kept the peace and order in a remarkably efficient way for years. fredgandt 02:55, 16 April 2016 (UTC)
I personally oppose this suggestion - Wikipedia is supposed to be a serious encyclopaedia, not a joke book. Vorbee (talk) 14:57, 23 April 2016 (UTC)
- Well, it did last a few hours without getting MfD-ed. Maybe for once, the community should be less stuck up and try to give a go at new things. If the name is a problem it can be changed but I see no reason to oppose the concept (yet). --QEDK (T ☕ C) 05:31, 28 April 2016 (UTC)
- Since I am part of this community, and it was just referred to as stuck up, I take umbrage. Perhaps this is an indication of why editors should maintain a professional distance on community talk pages. fredgandt 17:25, 28 April 2016 (UTC)
- That comment made me wince, too. I'm not sure what I think of the Banter page, yet (obviously should've been an RfC first, but beyond that I don't know). It's really disappointing, however, that discussions about new ways to foster community so often lead to supporters of the idea attacking the community for not supporting their particular method of fostering community. — Rhododendrites talk \\ 19:38, 29 April 2016 (UTC)
Religious messages
No consensus shown to change the status quo through this overly-broad proposal. It is noted that users are given a freer hand on their pages than those in the article space. Self Identification of any kind is not limited by Wikipedia policy. Limits are already in place to control over-enthusiastic proselytizing. Non-Administrative closure – GenQuest "Talk to Me" 11:52, 30 April 2016 (UTC)
- The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
I propose that people should not be allowed to use their Wikipedia user pages, signatures etc. to promote their religious beliefs. 81.152.70.247 (talk) 00:28, 9 April 2016 (UTC)
- Wikipedia:What Wikipedia is not#Wikipedia is not a soapbox - Says no advocacy.
- Wikipedia:User pages#What may I not have in my user pages? - Agrees.
- Wikipedia:Userboxes#Content restrictions - Agrees, but
- Wikipedia:Userboxes#Life, personal status and situation - lists religious userboxes
- User:UBX/Userboxes/Religion - like these.
- Wikipedia:Userboxes#Life, personal status and situation - lists religious userboxes
- I'd imagine this has been discussed many times before, and it's all very open to interpretation. I personally couldn't care less; each to their own fredgandt 04:59, 9 April 2016 (UTC)
-
- Personally, I have no problem with a user identifying his religious beliefs (or the lack thereof) on his user page with an infobox or otherwise. There is a difference between mere identification (which can, indeed, be useful in understanding where a user is coming from and it's the kind of revealed information about a person which helps to build community here) and advocacy or promotion, which I agree is inappropriate, though I think a good deal of tolerance should be exercised in regard to user pages. A couple of infoboxes, even if kind of fervant, should be tolerated; long rambling essays, proselytizations, or screeds should not. Raising it in a signature, however, might well be pushing the line, since a signature will be repeated again and again. On the other hand, I think that existing policies and guidelines, including those listed by Fred Gandt, above, are more than sufficient to deal with any situation which might come up and I'd be opposed to writing any new specific rules about the issue. If you have doubt about a particular user's usage, first take it up with them, then if you get no satisfaction take it to an administrator or to ANI. Regards, TransporterMan (TALK) 05:44, 9 April 2016 (UTC)
- There is really only something to pick a fight over, and it would not surprise me to find more people promoting their atheist religious beliefs, and that they would have a lot hard time reining that in without feeling extremely suppressed. Mangoe (talk) 12:28, 9 April 2016 (UTC)
- There are no "atheist religious beliefs". HTH. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:04, 11 April 2016 (UTC)
- In addition to the hurdles Fred has already pointed out WP:CSD#U5 applies "where the owner has made few or no edits outside of user pages". Users need to be substantial contributors before they start writing about their imaginary friends or anything else "not closely related to Wikipedia's goals". Bazj (talk) 12:50, 9 April 2016 (UTC)
- Don't care if someone espouses their beliefs, relgious or non-religious, on their user page. And actually it is a good way to vet bias in articles. Jaldous1 (talk) 16:06, 10 April 2016 (UTC)
- Something Bazj points out (purposefully or not) is where policy and good manners will decide inappropriate posts of pretty much any kind.
- "This user is [insert personal choice here]" is fine within reason, whereas "This user thinks [insert other's personal choices here] are deluded" or even "This user [insert thing that typifies a personal choice here] in order to save your retched souls" is definitely NOT okay.
- In other words, stating simply that one is inclined a certain way is (within reason) fine, but stating that not being that way inclined is bad, or that being inclined another way is bad, or even stating that being a particular way is especially good, begins to smack of advocacy or derision, which is not acceptable. fredgandt 16:23, 10 April 2016 (UTC)
- If a user starts to add major Jewish (or Muslim, or Catholic, or atheist) POV issues in articles, it would be very useful for the project if this user also happens to admit to being Jewish (or Muslim, or Catholic, or atheist) prominently on his/her userpage. עוד מישהו Od Mishehu 05:41, 11 April 2016 (UTC)
- If a user starts to add any non-neutral POV we already have policies to deal with that. WP is about facts with sources not about opinions or biases, religious or otherwise. Saying we need to ask a user to divulge their religious views to ensure good editing is saying WP's key policies of WP:V, WP:RS, WP:NOR, and WP:AGF do not work. The day that becomes true I will quit editing and even quit reading WP. Koala Tea Of Mercy (KTOM's Articulations & Invigilations) 06:28, 11 April 2016 (UTC)
- The OP's question was about promotion which would clearly be POV if it were in article space. In their own user space, not so clear. As Od Mishehu pointed out, disclosure is good. Even better to edit in such a way that nobody ever suspects you even have an opinion, and that the question of POV never arises.
- If the disclosure crosses the line into proselytising or bearing witness, it's clearly "not closely related to Wikipedia's goals", and clearly flags up a topic on which the editor may be easily trolled (as Fred_Gandt called me on above). As the late, great Dave Allen used to sign off, "May your god go with you", Bazj (talk) 07:18, 11 April 2016 (UTC)
- Obviously, there are red lines even in this context; certainly, though, any content which would fit into a normal-sized userbox in normal-sized text would be too short to be proselytising. עוד מישהו Od Mishehu 14:40, 11 April 2016 (UTC)
- If a user starts to add major Jewish (or Muslim, or Catholic, or atheist) POV issues in articles, it would be very useful for the project if this user also happens to admit to being Jewish (or Muslim, or Catholic, or atheist) prominently on his/her userpage. עוד מישהו Od Mishehu 05:41, 11 April 2016 (UTC)
- Unless sharing sharing personal preferences of any type that aren't strictly related to encyclopedia editing is prohibited, which would basically force everyone to edit anonymously, expressing personal preferences of any type could fall under the first three bulletin points the IP mentions (if we interpret them in the manner the IP is).—Godsy(TALKCONT) 04:24, 12 April 2016 (UTC)
-
- We only ask that editors be professional, not perfect. Set aside your personal biases to the best of your ability and only add content that can be appropriately sourced. That is all. Koala Tea Of Mercy (KTOM's Articulations & Invigilations) 05:00, 12 April 2016 (UTC)
- @Koala Tea Of Mercy: My comment is mainly in response the notion users shouldn't be allowed to put userboxes (or just state it in general) stating their religion on their userpage (and signatures to a certain extent). Nothing to do with editing articles. To sum it up: Almost all preferences are allowed to be expressed (pedophilia is a notable exception). They are either all in or all out. We can't discriminate upon the expression of religious views by assuming expressing that particular preference is an attempt to proselytize, advocate, or an indication of more bias than any other association or veiw expression. Statements and expressions like rainbow flags, peace signs, Flying Spaghetti Monsters, and even nationality (notably those Canadians with their 🍁 signatures ) and sex would all have to be disallowed.—Godsy(TALKCONT) 06:45, 12 April 2016 (UTC)
- We only ask that editors be professional, not perfect. Set aside your personal biases to the best of your ability and only add content that can be appropriately sourced. That is all. Koala Tea Of Mercy (KTOM's Articulations & Invigilations) 05:00, 12 April 2016 (UTC)
- This proposal is rather broad. As for signatures - I don't really have issue with someone adding a character to their signature unless it is specifically offensive; I'm opposed to adding POV messages to signatures though - they are meant to be identification points only. As for user pages - I'm fine with people self identifying their beliefs (e.g This user is a christian) because it also helps identify potential COI's -- but I'm opposed to using userpages for "promoting" of personal beliefs in general. — xaosflux Talk 18:34, 12 April 2016 (UTC)
- Potentially outdated, but related: http://www.wikichecker.com/religion/ Ckoerner (talk) 18:34, 18 April 2016 (UTC)
Is not this already covered at Wikipedia: What Wikipedia is not under the section which states that Wikipedia is not a soapbox? This section does note that on some articles - such as those to do with politics (one could also argue on religion) it might be tempting to promote one's own viewpoint, but clarifies how Wikipedia at least aims to be neutral - ergo, this has already been covered.Vorbee (talk) 20:59, 22 April 2016 (UTC)
- I think this is a slightly different situation, as it relates to user pages and signatures, rather than article content. Cordless Larry (talk) 18:29, 24 April 2016 (UTC)
- WP:NOTCENSORED. That said, I've seen users whose only edits were to make pretty userpages about themselves and their personal beliefs and preferences (religious and otherwise) and those need to go per WP:NOTWEBHOST. PCHS-NJROTC (Messages) Jesus Christ loves you! 03:53, 25 April 2016 (UTC)
- I think at the end of the day, it all comes down to whether the user is a net positive. If the user encourages other editors to read the Bible on the userpage (even all over their userpage in an obnoxious way) but creates dozens of Good Articles, are we going to demand they take it down and drive them off the site? No. The negative of religious expression on the user page is clearly outweighed by the good the editor is doing. This is not to say content creators (or other productive editors) can't be the subject of scrutiny, but we have to pick and choose our battles. Obnoxious religious expression is obnoxious, but it's a very small negative for the encyclopedia as a whole. It's not worth driving away good editors. ~ RobTalk 04:05, 25 April 2016 (UTC)
- What why is everyone slapping this puddle of chunky sals-- Oh, that's not salsa. I'm seeing a pretty clear consensus that this is already covered by WP:NOTSOAPBOX and other pages, but that those pages do not apply to a user simply stating that they hold (or reject) certain beliefs (whether they be religious, political, philosophical, aesthetic, or whatever), particularly if those beliefs are not leading to any sort of disruption. The only real discussion I'm seeing here is nitpicking, and I'm aware that someone could find something to nitpick over how I expressed the consensus, but that was the general direction I see that it's already gone. Time to close and move on? Ian.thomson (talk) 04:19, 25 April 2016 (UTC)
- Baseless. Identifying yourself as something doesn't mean you're spreading it and/or have a COI with the subject. --QEDK (T ☕ C) 05:33, 28 April 2016 (UTC)
- The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
Implementing Help:Maintenance template removal
- Note: the genesis for this started with Wikipedia:Village pump (miscellaneous)/Archive 51#Are notice tags demotivating?--Fuhghettaboutit (talk) 01:57, 10 April 2016 (UTC)
Hey all. For years we regularly get questions at various fora about maintenance template removal. (In fact, a few questions related to this issue were posted today (1, 2.) The questions we often receive show that:
- i) it is not obvious to many that maintenance templates are placed and removed manually. Many people assume there’s some mysterious automated threshold or A.I. program that governs removal;
- ii) some people seem put off from helping out in the first place—refrain from tackling the problem the template flags—because the process of template removal is neither obvious nor transparent;
- iii) because there’s nothing in the display of most templates to indicate to the reader that they can dive in to help, and no link to a page explaining the process (as proposed here), they assume it’s some job for an elite class of user (and so we lose the desperate help we need from the masses); and
- iv) people often do not grasp what needs to be done to address the issue flagged by the template, despite the links in them.
So, I’ve drafted Help:Maintenance template removal. The intent is that we add to the display of a variety of maintenance templates the following text, linked to this help page:
I’ve spent a lot of time at this help page not just describing the ins and outs of the removal process, but a section addressing some some of our more important and high-use templates — explaining the issues involved, relevant guidelines and policies, and what needs to be done before removal.
If consensus is gained to add such a proposed link to maintenance templates, whether as currently drafted or as suggested after discussion, I will need some help from the more tech-savvy as to the proper way to incorporate the link into the templates. Right now my manner of adding it, as seen in the {{unreferenced}} example I used at the help page, does not play well with that template’s date parameter (“April 2016”), which should appear after the main text, not after the proposed link.--Fuhghettaboutit (talk) 01:57, 10 April 2016 (UTC)
Discussion (maintenance tag removal)
- Misunderstandings about the purpose of templates and the process of removing them seem to be fairly common among new editors. I would Support adding a notice like this to maintenance templates. It is possible that this could lead to some templates being removed prematurely, but I think making the process easier to understand is worth it. Howicus (Did I mess up?) 02:11, 10 April 2016 (UTC)
- Support. One thought: new editors who this is aimed at may not understand that the message they are seeing comes from a template. So might it be more meaningful to refer to "how and when to remove this message"? --Gronk Oz (talk) 02:36, 10 April 2016 (UTC)
- Hey Gronk Oz. How about "how and when to remove this template message"? I kind of like the idea that we're being precise and teaching what the "thing" is (a template), but I like your idea too. Plus, this mirrors the name of WP:TM.--Fuhghettaboutit (talk) 02:49, 10 April 2016 (UTC)
- Support A great idea. I would not object to minor wording changes. Cullen328 Let's discuss it 06:44, 10 April 2016 (UTC)
- Support Excellent idea. Templates are about 3rd dan for most new editors, and some more accessibly presented How-To would clearly be welcome.-- Elmidae (talk) 13:33, 10 April 2016 (UTC)
- I would definitely support this. This issue comes up over and over at the Teahouse. DES (talk) 15:14, 10 April 2016 (UTC)
- Support anything that helps new uses is welcome. I assume that it only apples to the box type templates and not the inline versions. Keith D (talk) 15:54, 10 April 2016 (UTC)
-
- Hey Keith D. At present, yes, this is geared toward and would only apply to the box type. The templates for which I provided specific instructions at the help page are where I thought these would first be added.--Fuhghettaboutit (talk) 16:58, 10 April 2016 (UTC)
- Support. Needed and necessary. GenQuest "Talk to Me" 16:48, 10 April 2016 (UTC)
- Support per everybody. ―Mandruss ☎ 17:16, 10 April 2016 (UTC)
- Good idea. If anything, it's a way to help people realise they can edit Wikipedia. BethNaught (talk) 17:23, 10 April 2016 (UTC)
- Support. Excellent idea. --ColinFine (talk) 17:35, 10 April 2016 (UTC)
- Support - great idea. Thanks for working on this. Ajraddatz (talk) 17:42, 10 April 2016 (UTC)
- Support - this issue comes up a lot at the Teahouse, and this seems a good way to try to help new users understand the issue. Cordless Larry (talk) 17:44, 10 April 2016 (UTC)
- Support - and a hearty pat on the back for the effort. Comment - There's currently a RfC at {{Multiple issues}} (talk) about showing links to related talk page sections, for discussion of the issues raised, in the concise versions of maintenance templates. I would suggest that this proposed change and any that may arise from that discussion be borne in mind together. fredgandt 17:51, 10 April 2016 (UTC)
- Support — I think the tendency for most people who aren't experienced with Wikipedia's mechanisms is to view maintenance tags simply as red flags over particular articles' reliability, and not as actionable issues that can be resolved and then checked off. This change is a good idea that would perhaps put them to better use. —Nizolan (talk) 01:57, 11 April 2016 (UTC)
- Comment, good help page, maybe start with 1..3 cleanup templates to test/show your intended link feature. –Be..anyone (talk) 22:25, 11 April 2016 (UTC)
- Support-- I read the Help article and it was like a breath of fresh air. Thank you so much User:Fuhghettaboutit for this long overdue addition. Fritzmann2002 18:11, 13 April 2016 (UTC)
- Support - excellent work. JohnCD (talk) 19:10, 13 April 2016 (UTC)
- Support - Well thought out solution. Gmcbjames (talk) 00:12, 14 April 2016 (UTC)
- Support - A good solution to a real problem. The help page is good and very accessible. Happy Squirrel (talk) 04:54, 14 April 2016 (UTC)
- Support all efforts and any implementation approach to do this. We have a major problem with new editors believing that some anointed admin needs to decide even the most obvious of cases, so they add refs but leave {{unref}} at the top of the page for months afterwards. Seriously: it's occasionally awkward if we screw up a widely used template, but the fact is that this is a wiki, and if we don't like the first place/color/size/style/whatever, then we can change it later. Let's do this. WhatamIdoing (talk) 18:07, 14 April 2016 (UTC)
- Support - great idea, and the draft looks good. Long overdue! --SubSeven (talk) 06:00, 15 April 2016 (UTC)
- Support Staggeringly good idea. WaggersTALK 09:52, 15 April 2016 (UTC)
- Support, to pile on a bit. Excellent idea. APerson (talk!) 22:50, 16 April 2016 (UTC)
- Support a great idea. --Tom (LT) (talk) 00:40, 17 April 2016 (UTC)
- Support This is very good. My quibble is that the help page concentrates on new editors removing a template after they have addressed the issues, though in my experience this is not the only time someone should be looking to remove a template. I have seen templates languishing on articles years after the issue has been dealt with. Templates should be removed after the issues have been addressed, regardless of who addressed the issue or when. Sometimes the issue might have been resolved in pieces by several editors over a period of time, and nobody took final responsibility for the template. Sometimes a template is placed in error (or anger) so there is no amendment work to be done. I think it would be helpful if the guidance page addressed everyone, not just newbies. The advice on Template:POV is useful, and some of that could be incorporated. I might leave some comments on the help page talkpage. SilkTork ✔Tea time 08:51, 18 April 2016 (UTC)
-
- Note: SilkTork already proposed at the talk page and then made excellent additions to address the matters raised above.--Fuhghettaboutit (talk) 15:46, 23 April 2016 (UTC)
- Support - no brainer really, why has it taken so long? Include various suggestions above too. Aoziwe (talk) 10:50, 23 April 2016 (UTC)
- Support as discussed above. Doc James (talk · contribs · email) 09:29, 24 April 2016 (UTC)
- Hold on please Fuhghettaboutit. I would support a link to a page describing how and when to remove a particular maintenance template, but how useful is a generic page like Help:Maintenance template removal? You could at least link to a section or anchor of that page, which deals with the particular template. Ideally there would be separate guidance for each maintenance template. — Martin (MSGJ · talk) 12:03, 26 April 2016 (UTC)
-
- Martin, I feel such a disconnect between what you’re saying and this proposal’s purpose, the bases I gave for the why of it all, and how ungeneric and tailored the help page is for that purpose. “How useful is this” non-"generic" page? Extremely? Surpassingly? Pick your synonym. The constant notion that some elite class will fix the issue is dispelled. People will get that they can contribute and be bold. They will stop thinking templates are automatically removed (which we know is a persistent and common misunderstanding). With the clarity of removal explained, they will be more likely to act in the first place, maybe even enticing some bright people to first edit Wikipedia when seeing that, yes, they really can dive in to fix the problem. And the page is geared to teaching the information necessary to deal with any maintenance template, regardless of whether it’s one of the nine for which specific guidance is provided (the nine were chosen because they are so common). People can read. If they’re there as to one of those templates, and they’re the type of person who is likely to actually help, then they're much more likely in my experience to be the type of person to actually read the page – which is not overwhelmingly long – and to see the specific template guidance section. And for those who arrive there from a template which is addressed in the specific template guidance section, they should still be reading the content higher on the page, because it’s information is more important. Though I did provide anchors for the specific template section, I think it would actually be less helpful to have the link go to that section, thereby making it less likely people will read the page.--Fuhghettaboutit (talk) 01:55, 27 April 2016 (UTC)
- Support. Great idea, may encourage editing and helps reveal more of Wikipedia under the hood to readers. The help pagr is good. Fences&Windows 17:28, 29 April 2016 (UTC)
- Support per everybody. Comment Templates, especially PoV/unreffed/Peacock/etc. are often removed by the PoV editor who is convinced there isn't an issue, would it be worth strengthening the 'when not to remove' advice so that editors are specifically instructed to NOT remove if an/other editor/s objects. I note it says 'seek concensus', but could this be strengthened? Pincrete (talk) 22:22, 5 May 2016 (UTC)
Implementation discussion (maintenance tag removal)
- This section was originally posted to Wikipedia:Village pump (technical)#Template help and relocated here.
<introductory text removed as out of context here> What I need is help with the technical aspect of doing this – a standardized way to pass that link through to the templates. It should appear after all other content, after a line break. The way I've done it up to now (as seen in the examples I placed at the linked help page) is to just add the message after a break (<br />) to the fix= parameter these template all have. But the result is that the date parameter of the templates appears after the link:
- • Learn how and when to remove this template message. (April 2016)
So, what needs to be done in order to leave the date parameter where it belongs, after the main template content, and have the link appear after all of that content? Is there a way to keep it in the fix parameter but make the date not appear after it? If not, do we need a new parameter in Ambox to accomodate the link (which if I understand correctly is just a conduit for Module:Message box, so is there a change needed there?)? Thanks--Fuhghettaboutit (talk) 00:34, 14 April 2016 (UTC)
@Fuhghettaboutit: Would this be any better (I know it's not exactly what you want, but appears to be an improvement)?
Mdann52 (talk) 12:13, 14 April 2016 (UTC)
- Thank for responding Mdann52. I really think it should be set off in the manner proposed. It's not part of the main text, and interrupts its flow.--Fuhghettaboutit (talk) 12:17, 14 April 2016 (UTC)
- Please add
{{multiple issues}}
near the top of your test list, it should not trigger multiple links to the same help page. A WikiProject Templates exists or existed, maybe you find more support there. –Be..anyone 💩 12:58, 14 April 2016 (UTC)
- Please add
- The change would almost certainly have to go into Module:Message_box, probably an additional parameter that could be specified in template:ambox or whatever. Rwessel (talk) 05:48, 15 April 2016 (UTC)
- Thanks to everyone thus far for participating. Since at this point it looks pretty good for implementation, I'm pinging the creator and main editor of Module:Message box.--Fuhghettaboutit (talk) 02:32, 17 April 2016 (UTC)
- The module was updated by Mr. Stradivarius to add the link when | removalnotice = yes is placed. Multiple issues was tested so that it would only place the link but once. I've added it to the nine templates specifically addressed at the help page, and will wait a fair amount of time before adding it anywhere else. Thanks all.--Fuhghettaboutit (talk) 11:56, 26 April 2016 (UTC)
- Comment from an NPP and hopeful template editor: I feel like the proposed modification was poorly implemented from a design standpoint. I feel like it is unbalanced and expands templates more than necessary. {{More footnotes}} is now five lines long when blp=yes. I hate to complain and then not have a solution, but I personally feel like it needs to be integrated into the body better. -©2016 Compassionate727(Talk)(Contributions) 17:29, 26 April 2016 (UTC)
- I came here to voice the same thing; the line-fed left-aligned bulleted list of only one list element is unbalanced and adds an unnecessary amount of whitespace (grayspace?) to the tag. How about putting the link in parenthesis after the tagging date parenthesis, and make it italic and lowercase to match the date? – voidxor 20:43, 26 April 2016 (UTC)
- I think it's better on its own line, though having it in is what's important; having it appear after the date paramater would not be my choice but would still be fine. Certainly though, there is significant whitespace before it that would be best removed; snug it up as much as possible (and also if possible, maybe the space after can be reduced as well?) Unfortunately, I do not have the technical ability to help. Any tech people out there?--Fuhghettaboutit (talk) 01:22, 27 April 2016 (UTC)
- I asked and received assistance from Mr. Stradivarius. It is now on the same line as the date.--Fuhghettaboutit (talk) 12:19, 27 April 2016 (UTC)
- I think it's better on its own line, though having it in is what's important; having it appear after the date paramater would not be my choice but would still be fine. Certainly though, there is significant whitespace before it that would be best removed; snug it up as much as possible (and also if possible, maybe the space after can be reduced as well?) Unfortunately, I do not have the technical ability to help. Any tech people out there?--Fuhghettaboutit (talk) 01:22, 27 April 2016 (UTC)
- I came here to voice the same thing; the line-fed left-aligned bulleted list of only one list element is unbalanced and adds an unnecessary amount of whitespace (grayspace?) to the tag. How about putting the link in parenthesis after the tagging date parenthesis, and make it italic and lowercase to match the date? – voidxor 20:43, 26 April 2016 (UTC)
- Comment from an NPP and hopeful template editor: I feel like the proposed modification was poorly implemented from a design standpoint. I feel like it is unbalanced and expands templates more than necessary. {{More footnotes}} is now five lines long when blp=yes. I hate to complain and then not have a solution, but I personally feel like it needs to be integrated into the body better. -©2016 Compassionate727(Talk)(Contributions) 17:29, 26 April 2016 (UTC)
- The module was updated by Mr. Stradivarius to add the link when | removalnotice = yes is placed. Multiple issues was tested so that it would only place the link but once. I've added it to the nine templates specifically addressed at the help page, and will wait a fair amount of time before adding it anywhere else. Thanks all.--Fuhghettaboutit (talk) 11:56, 26 April 2016 (UTC)
- Thanks to everyone thus far for participating. Since at this point it looks pretty good for implementation, I'm pinging the creator and main editor of Module:Message box.--Fuhghettaboutit (talk) 02:32, 17 April 2016 (UTC)
Related proposals at Meta
Editors interested in Maintenance tag issues may also want to read and contribute to two proposals in Meta:
- Maintenance_tags_follow-up
- A better interface re maintenance tags --S Philbrick(Talk) 00:22, 17 April 2016 (UTC)
Watson and next generation improvements to the standard Wikipedia search box
Although natural language use in search boxes is still a long way off, the recent press is full of announcements of IBM making massive investments into the use of WATSON software in the search boxes of the medical industry and various specialized business sectors. Is Wikipedia evaluating or talking about the possible use of WATSON software for improvements and enhancements to the standard Wikipedia search box currently in use at the top of the page. Fountains-of-Paris (talk) 15:28, 11 April 2016 (UTC)
- You are suggesting using advanced artificial intelligence to a software development team that, despite millions spent of search, created a search box where searching for "a = b" returns "R.A.B.", "A-B", "(a,b)-tree", "A. B. Guthrie, Jr.", and "A/B testing". If, for whatever reason (I strongly suspect bad management, not technical incompetence) they cannot manage to provide basic functionality such as searching for a literal string, for FSM's sake don't ask them to partner with IBM to try to provide A.I.! --Guy Macon (talk) 01:37, 12 April 2016 (UTC)
- You mean, just like google does and any other modern search engine. This is how search engines work these days, they ignore things like punctuation. If you want a 100% literal match, you need to use insource regexp searches. —TheDJ (talk • contribs) 16:07, 12 April 2016 (UTC)
- A point you probably have not considered is that WATSON is not just software, it is also highly proprietary hardware customized for high-speed AI algorithms. The software and the hardware are an inter-dependent system and one is useless without the other. Koala Tea Of Mercy (KTOM's Articulations & Invigilations) 05:08, 12 April 2016 (UTC)
-
- I believe that you are incorrect. Watson runs on SUSE Linux Enterprise Server with Apache Hadoop. Such systems are readily available, but the IBM hardware is presumably much faster. The real problem is that all of that expensive hardware answers one question by one user at a time. It would take a lot more computing power to handle all of the searches Wikipedia gets. --Guy Macon (talk) 05:50, 12 April 2016 (UTC)
-
-
- You are correct that WATSON uses SUSE Linux Enterprise Server and Apache Hadoop but these two software packages are only a fraction of the whole WATSON system. The core software is called DeepQA and that runs only on the proprietary POWER7 processors.
- From our own article on WATSON (see Watson_(computer)#Current_and_future_applications) allow me to point out this: "IBM also intends to market the DeepQA software to large corporations, with a price in the millions of dollars, reflecting the $1 million needed to acquire a server that meets the minimum system requirement to operate Watson."
- For a quick but detailed overview of the scope of the software and hardware used by the WATSON system take a glance at this pdf of a presentation made by IBM at one of the SHARE conferences. Koala Tea Of Mercy (KTOM's Articulations & Invigilations) 06:34, 12 April 2016 (UTC)
-
-
-
-
- You mean the PDF that says "All of the hardware is available for sale at your friendly international purveyor of business machines"? Your claim that Watson is "highly proprietary hardware customized for high-speed AI algorithms" is factually incorrect. Call IBM, pay them three million dollars, and they will deliver a shiny new cluster of ninety IBM Power 750 servers and all the networking hardware needed to run Watson -- none of it custom. Then of course you need to buy the Watson software, which looks like it will run you a couple of million more. Or you can accept a slower response time and use fewer servers. --Guy Macon (talk) 13:47, 12 April 2016 (UTC)
-
-
- The whole point is moot I think, since it is a proprietary system if I'm not mistaken, and we only use Free and open-source software for the core functionality of the website. —TheDJ (talk • contribs) 15:58, 12 April 2016 (UTC)
- TensorFlow is open source. Praemonitus (talk) 21:31, 14 April 2016 (UTC)
- @Praemonitus: If its free source then does that put it back on the map of what Wikipedia can take up as part of the evaluations of the next increment in the enhanced version of the Wikipedia search box? Fountains-of-Paris (talk) 14:45, 15 April 2016 (UTC)
- TensorFlow is open source. Praemonitus (talk) 21:31, 14 April 2016 (UTC)
The "distance" of capability between Watson and the current Wikipedia search box
@Guy Macon, TheDJ, and Koala Tea Of Mercy:The distance between Watson and the current Wikipedia search box was discussed here [1]. Basically the current conventional search box does a little bit more than a simple search look-up which simply fails if the search item is not found. This was slightly improved a month ago to a 3-character look-back algorithm to correct letters only at the very end of the search item. For example "David Bowtie" will be corrected to "David Bowie" because the correction is only done at the end; if you type in "Fostoevsky" you get a failed search error message and no correction to "Dostoevsky" is even attempted. There are super fast algorithms for single letter correction, but Wikipedia is still short of accomplishing even that level. If Watson in the search box is not even on the radar screen for the Wikipedia search box, then what should be the expectation for the next generation of the standard Wikipedia search box improvement? Fountains-of-Paris (talk) 14:54, 13 April 2016 (UTC)
- You have to keep in mind that many AI problems are known to be NP-complete, i.e. we don't think we have efficient algorithms for them. This is part of why even Google pours tons of research into its search engine many years after it was first invented.--Jasper Deng (talk) 21:35, 14 April 2016 (UTC)
- @Jasper Deng: Yes I think that's correct about some very hard problems. If I put the two extremes of search box effectiveness at Wikipedia between full natural language interface at one extreme and the simple find/fail approach the matching the search item placed in the Wikipedia search box, then my feeling is that Watson added capabilities would be at the half way point between the two extremes. What do you think should be the current expectations for an improved search box? Can the Watson level of Wikipedia search box performance be practical and implemented if the Watson source TensorFlow is open source? Fountains-of-Paris (talk) 14:45, 15 April 2016 (UTC)
- @Fountains-of-Paris: First off, please stop conflating Watson with AI in general (TensorFlow is not part of Watson, for example).
- Even with TensorFlow, this would still require substantial development (and server resources) on the foundation's side. @Jdforrester: may have comments on how feasible it is; I am not a developer for the foundation and can't speak for them.--Jasper Deng (talk) 20:54, 15 April 2016 (UTC)
- @Jasper Deng: There is no preference on my part for using either Google's Tensorflow or IBM's WATSON. I first listed WATSON up above because of the large success and publicity of its "Jeopardy" quiz show. The current Wikipedia search box is currently non-AI in any appreciable way. It seems to follow the find-or-fail approach of simple look-ups of the search term. It does not even seem to incorporate the spelling correction software which Wikipedia currently uses for standard editing of its articles generally. Either Tensorflow or WATSON is fine if it gets the discussion started. At present when I type in "Fostoevsky" all Wikipedia gives is a fail error message, in place of a simple correction to "Dostoevsky" and a successful link to the page. Either Tensorflow or WATSON is fine for starting this discussion. Wikipedia should implement its very efficient word processor spelling correction software into the standard Wikipedia search box at the top of the page as quickly as possible since the software already exists in Wikipedia. Fountains-of-Paris (talk) 14:23, 16 April 2016 (UTC)
- @Jasper Deng: Yes I think that's correct about some very hard problems. If I put the two extremes of search box effectiveness at Wikipedia between full natural language interface at one extreme and the simple find/fail approach the matching the search item placed in the Wikipedia search box, then my feeling is that Watson added capabilities would be at the half way point between the two extremes. What do you think should be the current expectations for an improved search box? Can the Watson level of Wikipedia search box performance be practical and implemented if the Watson source TensorFlow is open source? Fountains-of-Paris (talk) 14:45, 15 April 2016 (UTC)
- @Fountains-of-Paris: You appear to be wrong about the "three character lookback algorithm". For example, if you type "Dxstoevsky" or "Dstoevsky" it will suggest Dostoevsky without trouble, and it even does well with something like "Axtidisestablishmentarianism" or "Pnumonoultramicroscopicsilicovolcanoconiosis". I've read (but can't find the link again, although T125671 looks relevant) that your problem with "Fostoevsky" that you keep harping on is that it currently requires the first letter to be correct so as to limit the search space. Anomie⚔ 13:47, 17 April 2016 (UTC)
- @Anomie: This is the discussion which includes the link I think you just referred to here [2]. When I copy your "Dxstoevsky" into the search box and search I then get a failed search message with no options listed. Same for your "Dstoevsky" misspelling, which again offers only to start a new article if you want to write one. However, when I type in something as bad as "Fostoevsky" then a Google search does give the corrected search results with useful options, as it should be expected to do. If the Google version of the search box is that much better than Wikipedia's version then possibly they can be approached by someone from Wikipedia or WMF to see if they could provide their search box to Wikipedia so that Wikipedia can dazzle users with a new "super" search box (that is, Google's search engine installed here at Wikipedia). In the general press there are reports that Google is friendly with and a sponsor of Wikipedia, so perhaps they are open to such a possible approach. Fountains-of-Paris (talk) 14:45, 18 April 2016 (UTC)
- Hello, I was pointed over this way from the discovery-l mailing list. While I can't claim to know it all when it comes to how our search works, I did want to help provide a little more supporting information to the discussion. I’m writing this under the assumption that readers aren’t super-savvy with every technical components. For more savvy readers, this might be stuff you already know.
- @Anomie: This is the discussion which includes the link I think you just referred to here [2]. When I copy your "Dxstoevsky" into the search box and search I then get a failed search message with no options listed. Same for your "Dstoevsky" misspelling, which again offers only to start a new article if you want to write one. However, when I type in something as bad as "Fostoevsky" then a Google search does give the corrected search results with useful options, as it should be expected to do. If the Google version of the search box is that much better than Wikipedia's version then possibly they can be approached by someone from Wikipedia or WMF to see if they could provide their search box to Wikipedia so that Wikipedia can dazzle users with a new "super" search box (that is, Google's search engine installed here at Wikipedia). In the general press there are reports that Google is friendly with and a sponsor of Wikipedia, so perhaps they are open to such a possible approach. Fountains-of-Paris (talk) 14:45, 18 April 2016 (UTC)
-
-
- So in a way Elastic and SOLR are children of Lucene, which powered wiki search up to about 2014.[2]
-
-
-
- Elastic, like it’s other family members, is open-source, with a pretty robust ecosystem of plugins. It’s also a well-used solution, with many big web services using it for their search. You can check out their site for a fancy list of clients and what not. Using Elastic lets engineers (of which we have 4 dedicated to search) focus on integration with our wiki-specific needs. We also get the advantage of using an active project and benitifing from it’s developments. [3]
-
-
-
- Right now, we're using a pretty good chunk of what Elastic offers us. The completion suggester update and some work on popularity scoring are recent examples of work on increasing the power of search.[4] [5] We’re also looking at leveraging a language-detection library called TextCat (=^.^=) to provide search results in a desired language (given the input).
-
-
-
- What makes Watson different is its focus on machine learning. This is, if you will accept my apologizes on the over-simplification, an extension of traditional search. While our current search ‘learns’ (like with popularity scoring, when new pages are added to the index, etc.) machine learning would allow search to use metadata to make associations that we humans can easily make, but computers can’t.
-
-
-
- Of which, ORES is a great example of where the foundation is looking into machine learning (and where the Discovery team is listening to experiment and learn).
-
-
-
- I guess this is all to say, “Search is complicated. We’ve got a solid stack to work on top of. We’re always eager for ideas and learning.” :)
-
-
-
- I hope this provides a little insight on what we’re using, why, and how far we are down the path of improving search. I’m happy to help answer related questions and/or bug those with more knowledge.
-
-
-
- Personal note: I worked in the data management team of a large healthcare provider before joining the foundation. We had a demo or two of Watson. It demos great. It's really expensive. :) CKoerner (WMF) (talk) 16:44, 18 April 2016 (UTC)
-
References
- ^ https://www.ibm.com/smarterplanet/us/en/ibmwatson/developercloud/doc/retrieve-rank/configure.shtml
- ^ https://blog.wikimedia.org/2014/01/06/wikimedia-moving-to-elasticsearch/
- ^ Like this: http://opensourceconnections.com/blog/2015/10/16/bm25-the-next-generation-of-lucene-relevation/
- ^ https://blog.wikimedia.org/2016/03/17/completion-suggester-find-what-you-need/
- ^ https://lists.wikimedia.org/pipermail/discovery/2016-March/000901.html
-
- @Fountains-of-Paris: FWIW we initially tested the completion suggester allowing first character substitutions, but after load testing against our production search cluster had to scale it back to not checking changes in the first character. The per-request time spent in the search engine goes from 2-3ms for what we use now to 10-12ms when also allowing first character substitutions. We perform a few thousand of these queries every second and the limitation is strictly a performance issue. We unfortunatly do not have the option to throw a few million servers at the problem like google does. There is a request to expand the search cluster's processing capacity by ~66% in the FY16-17 budget so we may be able to revisit this after expanding the search cluster, but it depends on how much of that extra capacity ends up being available and how much is used by natural growth in usage. EBernhardson (WMF) (talk) 17:46, 22 April 2016 (UTC)
- @EBernhardson (WMF): The issues you strongly represent here are also under discussion in the subsection directly below this one and are all important issues and benchmarks. The balance of how much of this assessment is shifted by hardware capacities being balanced by the strength of the search engine are of significance. In the subsection directly below the highlights were on the strength of the search engine software itself and its capacities, with possible discussion of enhancing with current Wikipedia search box using progress in Watson-like software applications, or, by emulating/implementing Google search engine capacities in the next generation Wikipedia search box. Some of the other editors and analysts at Wikipedia and WMF have started to indicate the practical limitations involved in such future development which I'll try to response to in direct order. If you have any data on the hardware capacity trade-off relationship to the relative strength of the search engine being used, then it would be interesting to see it or have it linked here. I do remember Sergei Brin in his often quoted interview from the 1990s telling people that the key transition for his development of the Google search engine was the emerging hardware capacity in his time back then "to download the full internet" in an accessible format. If you have some trade-off data on the hardware to software relationship then it would nice to have it linked here. Cheers. Fountains-of-Paris (talk) 15:45, 23 April 2016 (UTC)
- @Fountains-of-Paris: FWIW we initially tested the completion suggester allowing first character substitutions, but after load testing against our production search cluster had to scale it back to not checking changes in the first character. The per-request time spent in the search engine goes from 2-3ms for what we use now to 10-12ms when also allowing first character substitutions. We perform a few thousand of these queries every second and the limitation is strictly a performance issue. We unfortunatly do not have the option to throw a few million servers at the problem like google does. There is a request to expand the search cluster's processing capacity by ~66% in the FY16-17 budget so we may be able to revisit this after expanding the search cluster, but it depends on how much of that extra capacity ends up being available and how much is used by natural growth in usage. EBernhardson (WMF) (talk) 17:46, 22 April 2016 (UTC)
Would a Google-Wikipedia search engine be a better option.
@CKoerner (WMF): CKoerner (WMF) has clarified that it is unlikely that a Watson-like Q & A in the standard Wikipedia search box can be expected (see section above) for financial reasons and otherwise. Although less flashy than the Watson-like Q & A features, are there any options to pursue by way of getting the Google search engine installed to enhance the current Wikipedia search box software? If Wikipedia is budgeting upwards of half-a-million dollars per year on improving the current Wikipedia search box (based of ELASTIC and possibly ORES), then does offering a partnership to Google make sense if they would be willing? Google has the best search engine available today and it would seem to make some sense to try to partner with them. Does a Google-Wikipedia search engine partnership make sense financially? Fountains-of-Paris (talk) 21:09, 18 April 2016 (UTC)
- There's already a Google search available in .js, which I use and find rather good. DuncanHill (talk) 22:10, 18 April 2016 (UTC)
- @DuncanHill: Would installing the Google search engine software into the current Wikipedia search box (that is, replacing/updating the current Wikipedia search engine in the upper right corner of the standard Wikipedia page) be a big improvement with no costs added? Fountains-of-Paris (talk) 16:45, 19 April 2016 (UTC)
- I really couldn't give any comment on whether that would be a "no cost" improvement. As it is, I use the Wikipedia search box when I know, or am fairly sure of, the article title. I use the Google box when I don't know a likely title, or when I am searching for combinations of words etc. DuncanHill (talk) 16:50, 19 April 2016 (UTC)
- Due to the foundation's principle of user privacy, third party services are usually out of the question for things like this.--Jasper Deng (talk) 22:03, 19 April 2016 (UTC)
- @DuncanHill and Jasper Deng: This is the public announcement made by Wikipedia WMF about its generous relation to Google here: [3]. At present all of this is public and in the public domain. Google has made a grant in the seven digit range and a Wikipedia-Google partnership on using their search box engine, the best around, as an upgrade of the current Wikipedia search box engine would appear as being promising. Should the Google search engine be considered as a possibility to upgrade the current standard Wikipedia search box engine currently based on ELASTIC (possibly with or without ORES)? Would this also get the Watson-like AI research initiatives started on a stronger and enhanced footing? Fountains-of-Paris (talk) 14:37, 20 April 2016 (UTC)
- @Fountains-of-Paris: Please re-read my statement. Third-party grants are very common and usually highly welcomed. But the use of third-party software is a different manner.--Jasper Deng (talk) 15:26, 20 April 2016 (UTC)
- If you are aware that Google would be unsympathetic in advance regarding sharing their knowledge base and applications library then please let us know. Tensorflow and many other software apps are open source and publically available for anyone to use from Google. The highly welcome grants from Google to Wikipedia suggests that they might be more open to closer ties with Wikipedia which would be constructive. Should Wikipedia be interested or even open to the idea of enhancing the current standard Wikipedia search engine if Google would be willing to share their applications technology and their top level search engine technology? Fountains-of-Paris (talk) 16:08, 20 April 2016 (UTC)
- It just doesn't work the way you want it to. The foundation pretty strongly prefers to have its code developed from scratch unless it's free and open-source. Therefore, it is theoretically possible for the foundation to use a TensorFlow-based solution, but as I explained above, it is not deemed to be technically feasible. --Jasper Deng (talk) 17:37, 20 April 2016 (UTC)
- At the moment we were just told by Koerner that maintaining the Wikipedia search engine is a large effort involving the expense of at least 4 dedicated programmers at the WMF foundation. Estimate that at half-a-million dollars per year with incidentals added, and it sounds like a real bite into the annual expenses. At present I do no know if Google is even using TensorFlow in their own version of their search engine, or if the AI part is still in research and development for them. If you have some added insight here then great. The Google search engine is the best one out there and if Wikipedia can make it work here as well then so much the better for enhancing the standard Wikipedia search box. As I said before, the Wikipedia search box currently fails on "Fostoevsky" but the Google search box gets me straight to "Dostoevsky" no questions asked even when the same mistype of "Fostoevsky" is keyed-in. It sounds like a point in favor of trying to use the Google version of a better search engine at Wikipedia. Fountains-of-Paris (talk) 18:14, 20 April 2016 (UTC)
- It just doesn't work the way you want it to. The foundation pretty strongly prefers to have its code developed from scratch unless it's free and open-source. Therefore, it is theoretically possible for the foundation to use a TensorFlow-based solution, but as I explained above, it is not deemed to be technically feasible. --Jasper Deng (talk) 17:37, 20 April 2016 (UTC)
- If you are aware that Google would be unsympathetic in advance regarding sharing their knowledge base and applications library then please let us know. Tensorflow and many other software apps are open source and publically available for anyone to use from Google. The highly welcome grants from Google to Wikipedia suggests that they might be more open to closer ties with Wikipedia which would be constructive. Should Wikipedia be interested or even open to the idea of enhancing the current standard Wikipedia search engine if Google would be willing to share their applications technology and their top level search engine technology? Fountains-of-Paris (talk) 16:08, 20 April 2016 (UTC)
- @Fountains-of-Paris: Please re-read my statement. Third-party grants are very common and usually highly welcomed. But the use of third-party software is a different manner.--Jasper Deng (talk) 15:26, 20 April 2016 (UTC)
- @DuncanHill and Jasper Deng: This is the public announcement made by Wikipedia WMF about its generous relation to Google here: [3]. At present all of this is public and in the public domain. Google has made a grant in the seven digit range and a Wikipedia-Google partnership on using their search box engine, the best around, as an upgrade of the current Wikipedia search box engine would appear as being promising. Should the Google search engine be considered as a possibility to upgrade the current standard Wikipedia search box engine currently based on ELASTIC (possibly with or without ORES)? Would this also get the Watson-like AI research initiatives started on a stronger and enhanced footing? Fountains-of-Paris (talk) 14:37, 20 April 2016 (UTC)
- Due to the foundation's principle of user privacy, third party services are usually out of the question for things like this.--Jasper Deng (talk) 22:03, 19 April 2016 (UTC)
- I really couldn't give any comment on whether that would be a "no cost" improvement. As it is, I use the Wikipedia search box when I know, or am fairly sure of, the article title. I use the Google box when I don't know a likely title, or when I am searching for combinations of words etc. DuncanHill (talk) 16:50, 19 April 2016 (UTC)
- @DuncanHill: Would installing the Google search engine software into the current Wikipedia search box (that is, replacing/updating the current Wikipedia search engine in the upper right corner of the standard Wikipedia page) be a big improvement with no costs added? Fountains-of-Paris (talk) 16:45, 19 April 2016 (UTC)
Again, no matter how good of an idea, the foundation is unlikely to do anything involving third-party non-free code, for privacy reasons.--Jasper Deng (talk) 20:08, 20 April 2016 (UTC)
I don't have the time right now to write a longer rationale here, but suffice to say that using more Google-based infrastructure (or any kind of third-party infrastructure) is not under consideration and likely won't be any time soon. Here are some of the reasons:
- Firstly, negotiating with Google and maintaining that relationship would take up significant staff time (which costs money), and then we'd likely need as many staff members as we have now to maintain the bridge between our systems anyway. It seems unlikely we'd save any money by doing this.
- Secondly, we'd lose a significant amount of control over where our search logs go, potentially compromising the privacy and safety of our users. This would be bad because search logs can contain highly personally identifying information and therefore are presently very closely guarded.
- Thirdly, by using Elasticsearch, an open source search backend, we're able to contribute back to the open source search community by upstreaming improvements to Elasticsearch.
- Fourthly, we have three dedicated search engineers, which only represents around 1% of our total staff. The Foundation's investment seems perfectly proportionate to me.
- Finally, as demonstrated by the recent launch of the completion suggester, which vastly reduced the number of queries that give zero results and increased user satisfaction with search, we are able to make our own progress on this without relying on third parties.
Hopefully that helps clarify the situation. --Dan Garry, Wikimedia Foundation (talk) 23:53, 21 April 2016 (UTC)
-
- @Jasper Deng and Deskana (WMF): Thanks for both your comments. I have already made a preliminary response in the previous section regarding some of the trade-off questions concerning the strength of the search engine being used as compared to the strength of the hardware supporting the search engine technology being used. My reference was to the Sergei Brin comments in the 1990s where he indicated that the strength of the Google search engine was based on the then emerging technological ability to fully "download the internet" in an accessible format. This is the statement that needs to be tailored to the next level of inquiry for the next generation search engine design and implementation at Wikipedia. This would help determine the extent of your comments above on the practical limitations of using the model of the Google search engine for any future capacity objectives for the next generation Wikipedia search box engine. Could one you mention or indicate if the current capacities at Wikipedia and WMF are sufficient to do the equivalent of the Sergei Brin comment for Google in the context of Wikipedia; that is, can you "download the entirely of Wikipedia" as a basis for enhancing search box access performance in terms of preprocessing the direct database access addressing to articles, both properly spelled and improperly spelled, and then dynamically updating it on a daily or hourly (on-going) basis for all five million Wikipedia articles. I know the Wikipedia version is open source and I am asking the system design version of this question for purposes of the discussion here if one of you has a ready answer. Cheers. Fountains-of-Paris (talk) 16:10, 23 April 2016 (UTC)
- We have an open source Q and A machine here [4] Doc James (talk · contribs · email) 09:24, 24 April 2016 (UTC)
- In Q and A machines generally, both the Siri and Jarvis approaches to AI have had a mixed review. The link you provide for a similar Q and A machine takes over 30 seconds to respond to its own sample question ("Who wrote Ender's Game?") on a high speed internet line. That would be well over the under 1 second response time Wikipedia normally expects from its standard search box performance. Should the next generation to the standard Wikipedia search box move towards the Brin model of the Google search engine which is the industry leader and which is under one second in response time? Fountains-of-Paris (talk) 15:34, 25 April 2016 (UTC)
- We have an open source Q and A machine here [4] Doc James (talk · contribs · email) 09:24, 24 April 2016 (UTC)
- @Jasper Deng and Deskana (WMF): Thanks for both your comments. I have already made a preliminary response in the previous section regarding some of the trade-off questions concerning the strength of the search engine being used as compared to the strength of the hardware supporting the search engine technology being used. My reference was to the Sergei Brin comments in the 1990s where he indicated that the strength of the Google search engine was based on the then emerging technological ability to fully "download the internet" in an accessible format. This is the statement that needs to be tailored to the next level of inquiry for the next generation search engine design and implementation at Wikipedia. This would help determine the extent of your comments above on the practical limitations of using the model of the Google search engine for any future capacity objectives for the next generation Wikipedia search box engine. Could one you mention or indicate if the current capacities at Wikipedia and WMF are sufficient to do the equivalent of the Sergei Brin comment for Google in the context of Wikipedia; that is, can you "download the entirely of Wikipedia" as a basis for enhancing search box access performance in terms of preprocessing the direct database access addressing to articles, both properly spelled and improperly spelled, and then dynamically updating it on a daily or hourly (on-going) basis for all five million Wikipedia articles. I know the Wikipedia version is open source and I am asking the system design version of this question for purposes of the discussion here if one of you has a ready answer. Cheers. Fountains-of-Paris (talk) 16:10, 23 April 2016 (UTC)
@Fountains-of-Paris: Thanks for the question. To be honest, I'm unclear what you're really asking me here. You're using a lot of buzzwords and jargon in somewhat meaningless ways. Your response to me also has almost nothing to do with what I wrote, so I think we might be talking past each other. Anyway, I will try to answer your question. It's a red herring to ask if we can "download the entirety of Wikipedia" when, obviously, we already have access to the servers upon which the content Wikipedia is stored, and search indices built from that content. But, yes, there are dumps of the content that exist and can be downloaded; you can read more about that at Wikipedia:Database download. That has very little to do with search, however. --Dan Garry, Wikimedia Foundation (talk) 18:15, 27 April 2016 (UTC)
- @Deskana (WMF): Yes that reply by Brin on the Charlie Rose interview show I think was him trying to "dumb-down" the language for general discernment in a broad audience. The way Brin describes the Google search engine more accurately is when he speaks of it being based upon and operating on the approach of providing a list of backlinks ranked by importance. His further comparison is that other search engines (like Lucene with or without Elastic) are inferior because they do not contain crawling and parsing, and other reasons. The question for the next generation Wikipedia search box is whether it should continue on a development plan using approaches (Lucene, Elastic) which are inferior to the Google-model search engine, or, if it is better to take the Google-model search engine approach, the industry's best search engine, as the model for future development of an improved and enhanced Wikipedia search box in the next generation. By the standards of Google, Wikipedia's 5 million articles are a tiny sample of the backlists ranked by importance which they use for the world wide web. The question is: Should Wikipedia begin to change over to developing a new Google-like search engine, which would be modelled on the best one out there, custom written and adapted for the 5 million plus articles at Wikipedia for the benefit of the next generation of Wikipedia readers and users? Fountains-of-Paris (talk) 16:10, 28 April 2016 (UTC)
- @Fountains-of-Paris: I'm reading that, based on an interview where Google's co-founder said that he thinks Google's approach is superior to Elasticsearch, you think my team should should switch all of our infrastructure away from Elasticsearch, and instead build something totally from scratch. If so, no, that's not going to happen, for the reasons I explained above. I'm sorry, but this will be my last reply here; I don't have the time to devote to a back-and-forth on this. --Dan Garry, Wikimedia Foundation (talk) 16:35, 28 April 2016 (UTC)
Statistics for page counts and breakdown of the source of the page counts
Page count statistics are stored and available for all Wikipedia articles for the cumulative number of page hits per day on one of the tabs readily accessible on the edit history page. Does anyone know if these page count statistics are broken down anywhere as to the source of where the page request is coming from? What part of the the cumulative page count at Wikipedia comes from searching and linking from the Yahoo engine, as opposed to the Google engine, as opposed to the Wikipedia search box engine, how many are from desktop, how many are from mobile, etc? Which percentage of page counts for individual Wikipedia articles come from which sources whether from inside Wikipedia or from outside Wikipedia? Fountains-of-Paris (talk) 15:34, 4 May 2016 (UTC)
Auto-assessment
There exist nearly 700,000 articles currently in the category tree of Category:Unassessed-Class articles. Based on a sample I've looked at, it seems somewhere between 1/6 and 1/8 of these articles have been assessed, but some of the WikiProject templates on the talk page never had the {{{class}}}
parameter filled in! This sort of task is perfect for a bot. I'm proposing that a bot run through all unassessed articles and fill in the {{{class}}}
parameter using the already-filled-in {{{class}}}
parameter from other WikiProject templates on the same page. If there's not a non-empty {{{class}}}
parameter on the talk page, the article would be skipped. Some example edits: [5] [6] [7] [8]
This is a surprisingly simple task that seems unlikely to be controversial and potentially helps out hundreds of WikiProjects. I've already created the bot, but I'm seeking consensus that such a task is appropriate to run. Thoughts? ~ RobTalk 05:56, 21 April 2016 (UTC)
Discussion (Auto-assessment)
In theory, different WikiProjects are entitled to different set of criteria (in practice, the basic classes Stub–B are judged against the same criteria by all). Some other complications might arise:
- What if existing classes disagree? In your third example, edit WikiProject Women's History gave the article Start, but WikiProject Discrimination ranked it as Stub-class. In your example edit, the result is to favor the lower ranking. Is this the intended behavior, why? Typically articles progress with time, and the lower rankings are based on older revisions (as was indeed the case with this article).
- How will you treat non-standard classes (more of them here)? Say, if WikiProject Military history gives the article B+, but the WikiProjects with empty class parameters do not recognize that class neither in their assessment scheme nor in the way their template is coded.
- Unassessed articles encourage editors to manually assess the article based on its current state. The bot, on the other hand, will inherently give articles assessments based on their past, and not current, status.
– Finnusertop (talk ⋅ contribs) 06:52, 21 April 2016 (UTC)
-
I currently have it set to use the lower class, but I could just as easily skip it.I've now set it to skip articles that have multiple standard classes set. If the article has a standard and non-standard class set (i.e. B and B+), then it will evaluate to the standard class (since it's very unlikely other templates will be using the non-standard one).- Any class other than FA, FL, A-C, GA, Start, and Stub are ignored.
- I'm of the opinion some info is better than none. In reality, if an editor comes across an article with the situation I described, they'd probably plug in the old assessment value. Like always, editors can come by and update the classes. ~ RobTalk 11:31, 21 April 2016 (UTC)
-
-
- Good answers, Rob. Regarding the third point, it would be great if the edit summary of the bot said that manual assessments are preferred. Many people watch articles and talk pages along them, and this bot could simultaneously serve as a reminder to conduct a fresh manual assessment. – Finnusertop (talk ⋅ contribs) 19:34, 21 April 2016 (UTC)
- @Finnusertop: I can definitely do that. ~ RobTalk 20:32, 21 April 2016 (UTC)
- @Finnusertop: The
|auto=
parameter was provided on many WikiProject banners for just that purpose, see WP:AUTOASSESS; if copying from another banner the bot would set|auto=inherit
. Many WikiProject banners have had the feature removed in the last few years; some (like{{WikiProject Women}}
) never had it; some (like{{WikiProject Women's sport}}
} still have it. --Redrose64 (talk) 20:58, 21 April 2016 (UTC)
- Good answers, Rob. Regarding the third point, it would be great if the edit summary of the bot said that manual assessments are preferred. Many people watch articles and talk pages along them, and this bot could simultaneously serve as a reminder to conduct a fresh manual assessment. – Finnusertop (talk ⋅ contribs) 19:34, 21 April 2016 (UTC)
-
I think that every WikiPpoject that want auto-assessment has to state this individually. Otherwise, I do not see why we have class defined by each project separatelly. -- Magioladitis (talk) 16:52, 21 April 2016 (UTC)
- We could also do it that way; place a mass-message on all WikiProject talk pages and allow either opt-in or opt-out. I'd greatly prefer opt-out, because of how many inactive WikiProjects there are on-site. ~ RobTalk 16:55, 21 April 2016 (UTC)
- In case of inactive projects what's the reason to assess pages? Assessment is here to help wikiproject members. -- Magioladitis (talk) 17:13, 21 April 2016 (UTC)
There will also be example of different classes. -- Magioladitis (talk) 18:51, 21 April 2016 (UTC)
- Already mentioned by Finnusertop at 06:52, 21 April 2016. --Redrose64 (talk) 19:12, 21 April 2016 (UTC)
@Xeno: if they want to comment on this. I used their script to autoassess in the past. -- Magioladitis (talk) 21:47, 21 April 2016 (UTC)
- I think Mag has a point in that unassessed articles can help focus the efforts and eyes of the members of said WikiProject. Probably best to go for opt-in so that only the projects interested in mass auto-assessment have their WP templates adjusted. –xenotalk 00:50, 22 April 2016 (UTC)
- I've tested my script on a sample, and it seems to work well. I'd like to avoid focusing on the technical issues here, since that's best handled in the BRFA process. Before we start talking details of the technical implementation, we need consensus for the task itself. ~ RobTalk 23:13, 21 April 2016 (UTC)
- You can check User_talk:Xenobot_Mk_V/requests/archive (and older archives) for projects that wanted auto-assessment work in the past. –xenotalk 00:54, 22 April 2016 (UTC)
I am involved in assessment for WP-Australia and support this task. --99of9 (talk) 23:35, 21 April 2016 (UTC)
- Seems like everyone agrees this would be beneficial as an opt-in task, so let's go with that. @Magioladitis and Xeno: How would you recommend notifying all projects of this possibility? I'm unaware of any mass-messaging service that will hit all WikiProjects currently, but I could do a one-time botrun that places a brief message on all project talk pages. I compiled a list of 1,194 WikiProjects/task forces from Category:Active WikiProjects, Category:Semi-active WikiProjects, and Category:Inactive WikiProjects, manually trimmed of the pages that are miscategorized. I'm not sure if it would be appropriate to use a semi-automated program to place a notice on all their pages or if a BRFA would be required for that many edits. Given the nature of the edits, I'd basically be a human bot if I was using a semi-auto program, just clicking away. This is on the line of what WP:BOTASSIST might apply to. ~ RobTalk 21:08, 22 April 2016 (UTC)
- What's the point of having a separate "class=" parameter if all the WikiProjects are supposed to share the same class assessment? On the French Wikipedia {{WikiProjectBannerShell}} is used to generate a unified "quality" for the article as a whole. However, many WikiProjects do not share the same class structure. A problem is that "page type" is mixed into "quality class". First, PageType should be separated from quality class. A list is a page type, it doesn't indicate the quality of the page. Article/List/Redirect/Template/Project/Portal/etc should be set in a "pagetype" parameter instead of the class. This can be shared between wikiprojects with no conflicts in assessments, the only conflicts would be some wikiprojects recognize various pagetypes and others just use NA. Quality classes are also not shared across wikiprojects, with some having A-class and some not, etc. If we can have a unified page quality, then individual wikiproject quality assessments should be solely about the quality of the infornation in the project's purview. Also, having WPBS accept quality settings (and merging {{Vital article}} into WPBS) will allow setting quality assessments for articles without memberships in Wikiprojects (they are unbannered, and perhaps no wikiproject covers the topic area). -- 70.51.46.195 (talk) 05:04, 24 April 2016 (UTC)
- I don't necessarily disagree with a new system of assessment, but I'm working within the bounds of the current system. I doubt there's going to be significant support for changing the system since it's worked well enough so far, but you're welcome to try proposing it in an appropriate venue. As a side note, I'm going to remove "A" from the classes that this bot will auto-assess, because some projects don't use the A class. ~ RobTalk 05:16, 24 April 2016 (UTC)
- Many projects do have their own assessment system. The military history project for one is heavily complicated for their own purposes. I'd say first making a message on each project's talk page and see who bites. Collect all the supporting projects and then you can automate those together. For example if the NFL project supports it and the biographies one as well, you can also make sure that the biographies on NFL pages includes the sports parameter rather than doing it in rounds. Else, propose a bot that does this as a running project. A number of projects don't even use the extended class system which in particular doesn't expose them to draftspace pages (draftspace is a bigger area now than ever) which are basically the new pre-stub pages. -- Ricky81682 (talk) 02:31, 30 April 2016 (UTC)
- When you say "many projects", can you give examples other than MILHIST (which is different in several ways, not just assessment criteria) and Help Project (which doesn't classify pages in mainspace but those in Help: space). Other than those two, I don't know of any that have a
|class=
parameter but which don't respect the basics of Wikipedia:Version 1.0 Editorial Team/Assessment#Quality scale for pages in mainspace, perhaps with a small number of omissions (A-class being commonly omitted) or additions (Disambig, Redirect). For pages outside mainspace, it's a different matter, but other than Help project, those aren't subjective judgments - a non-redirected template might be template-class on one project and NA-class on another, but there is no way that it can be anything other than those two. --Redrose64 (talk) 08:32, 30 April 2016 (UTC)
- When you say "many projects", can you give examples other than MILHIST (which is different in several ways, not just assessment criteria) and Help Project (which doesn't classify pages in mainspace but those in Help: space). Other than those two, I don't know of any that have a
- Many projects do have their own assessment system. The military history project for one is heavily complicated for their own purposes. I'd say first making a message on each project's talk page and see who bites. Collect all the supporting projects and then you can automate those together. For example if the NFL project supports it and the biographies one as well, you can also make sure that the biographies on NFL pages includes the sports parameter rather than doing it in rounds. Else, propose a bot that does this as a running project. A number of projects don't even use the extended class system which in particular doesn't expose them to draftspace pages (draftspace is a bigger area now than ever) which are basically the new pre-stub pages. -- Ricky81682 (talk) 02:31, 30 April 2016 (UTC)
- I don't necessarily disagree with a new system of assessment, but I'm working within the bounds of the current system. I doubt there's going to be significant support for changing the system since it's worked well enough so far, but you're welcome to try proposing it in an appropriate venue. As a side note, I'm going to remove "A" from the classes that this bot will auto-assess, because some projects don't use the A class. ~ RobTalk 05:16, 24 April 2016 (UTC)
Revive WP:ABUSE, but completely revamp how it does business.
In recent years, the old Wikipedia:Abuse response project has gone inactive, and there doesn't seem to be anything going for notifying ISPs, schools, employers, etc anymore. This is despite the existence, and continued use of, templates like {{WHOIS}} and {{Shared IP}} which mention the possibility of abuse reports being sent (and with some templates still even linking to the inactive abuse response project). My proposal is that we either revive WP:ABUSE or create an entirely new project to create template messages which admins and vandal fighters can use to send to ISPs, contact information can be documented for ISPs, schools, etc (we already have this, which I would incorporate into the project), and users could document success stories. This is 'not a proposal to bring back the concept of people submitting cases to the abuse response project, but rather a proposal that we either revive the old project, or create an entirely new one, for the purpose of making it easier for administrators and vandal fighters to contact ISPs and organizations themselves. Thoughts? PCHS-NJROTC (Messages) Jesus Christ loves you! 01:08, 25 April 2016 (UTC)
- Screaming into the void before didn't produce satisfactory results, why do you think that will change? More to the point: The project died mostly because ISPs and other organizations basically ignored any and all requests for remediation, so why do you think that will be better if we revive the project? --Jayron32 01:20, 25 April 2016 (UTC)
- Before we try to restart getting ISPs and schools to take action about their vandals, I think that we need to try to answer a question that is difficult to know the answer to. Why are ISPs and schools so unwilling to deal with vandals (and other Internet scum) who use their address blocks in ways that are detrimental to Wikipedia? In the case of ISPs, I can see several reasons. First, nothing in an ISP's Terms of Service with its customer prohibits the customer from violating Terms of Service with an unaffiliated web site, such as Wikipedia. Second, there isn't a financial incentive for the ISP to throw off bill-paying users. Maybe a Wikipedia editor who works for an ISP can provide insight. As to schools, I don't have as good an answer. However, I don't think that we will be able to motivate ISPs to work with us until we identify business reasons for them to work with us. Robert McClenon (talk) 01:42, 25 April 2016 (UTC)
-
-
- People think ISPs don't respond to abuse reports. That simply isn't true; I've submitted hundreds of them, some related to WP, others not. If you search the internet, you can find people chattering in forums and other websites about being contacted about abuse issues (mostly virus activity and copyright infringement; probably a good 95% of the abuse reports they receive relate to that, and they get tons of them). As for the AUP issue, a lit of AUPs prohibit activity that hinders other subscribers' use of the service. Especially when we start talking rangeblocks, that certainly applies here. The reality is, WP:ABUSE was a failure for dealing with ISPs because its way of doing things was incompatible with the ISPs' way of doing things. WP:ABUSE required 5 blocks on an IP before they would act. On my experience monitoring my home router's firewall log, virus infected Time Warner Cable IP addresses will stop trying my port 445 after I've sent exactly three reports to anuse@rr.com, and I never receive a personal response unless I submit it wrong (or if the person at RR Abuse Control is retarded and thinks I submitted it wrong). ISPs will almost surely send a warning to the subscriber as the first action they take, if they take action. My guess is someone who has repeatedly ignored our warnings and been blocked 5 times will similarily wipe their ass on a warning from their ISP, unless it's a kid and the warning goes to the parents. With WP:ABUSE, contactors would often send one report and close the case as "actioned" afterward, and it didn't help that IPs with such a history of abuse were bound to get tagged with an extended block as WP:ABUSE is trying to work the case. In contrast, I've sent many abuse reports relating to Wikipedia and Conservapedia, both within and outside of WP:ABUSE and had good results. Jack Stevens with Embarq was good about calling subscribers on the phone about vandalism issues. I've had Comcast say over the phone they would take action. AT&T took action against User:Mmbabies after I contacted the Better Business Bureau (and he went away for roughly a year). I received a personal response via email from Cox Communications thanking me and asking me to contact them again if the problems were to persist. I saw persistant pom-pom type vandalism (clearly the same person) from a Department of Homeland Security IP registered through Sprint cease when I contacted Sprint. Karajou at Conservapedia shared an email with me where he received a personal response from Bell Canada. JANET in the UK always responds well to abuse reports. I've had good experiences dealing with schools and universites as well. Dealing with schools is a bit of a different animal. If you just contact what is listed at ARIN, you are wasting your time as that is probably out of date. I normally go to the school website, and if I don't call them on the phone (which is actually what is best), I send an email to either the technology department or a school administrator (like a principal). If anyone is interested, I could email you copies of some of the interactions I've had over the years. PCHS-NJROTC (Messages) Jesus Christ loves you! 02:54, 25 April 2016 (UTC)
- Something else I forgot to add is that a couple of times I saw threads at RationalWiki's Saloon Bar about prople getting their internet service suspended for vandalizing Conservapedia. The one that sticks out in my mind I highly suspect was using AT&T Mobility, based on the timing of the reports. It took two reports to achieve that. PCHS-NJROTC (Messages) Jesus Christ loves you! 02:57, 25 April 2016 (UTC)
- @Jeske, throughtout my experience, I can remember one high school which said they wouldn't do anything unless the WMF contacted them. That's okay, in that instance it was literally a cheerleader posting about how she was the captain of the cheer squad, so I enailed her coach about her making the team look bad. I know, a little harsh, when I was 18 I was merciless about it though and her coach said she would talk to her about it. I don't think I would go to that extreme if the same thing hapoened today though. PCHS-NJROTC (Messages) Jesus Christ loves you! 03:12, 25 April 2016 (UTC)
-
- For those of you doubting the effectiveness of abuse reports. PCHS-NJROTC (Messages) Jesus Christ loves you! 01:23, 28 April 2016 (UTC)
- Given the increased problems with harassment on and off-wiki nowadays, I see this as ripe for massive abuse. The only party that I think would have legitimate support (if it did do something) would be ARBCOM since only they would have the actual long-term evidence (and can examine on and off-wiki) and the long-term perspective on the worst of the worst characters. Otherwise, I can't imagine how this will work unless people are prepared for the absolute and real risk of being blocked themselves for OUTING and harassment violations if they get the wrong person. -- Ricky81682 (talk) 02:22, 30 April 2016 (UTC)
- Huh? How does abuse reporting have anything to do with WP:OUTING? One contacts the abuse department for an IP address and forwards the logs, it's not like we're naming suspects (and I long since quit contacting cheer coaches, etc about issues, just IT departments and abuse departments). PCHS-NJROTC (Messages)Growing tired of the bullshit day by day. 06:03, 30 April 2016 (UTC)
- Given the increased problems with harassment on and off-wiki nowadays, I see this as ripe for massive abuse. The only party that I think would have legitimate support (if it did do something) would be ARBCOM since only they would have the actual long-term evidence (and can examine on and off-wiki) and the long-term perspective on the worst of the worst characters. Otherwise, I can't imagine how this will work unless people are prepared for the absolute and real risk of being blocked themselves for OUTING and harassment violations if they get the wrong person. -- Ricky81682 (talk) 02:22, 30 April 2016 (UTC)
I actually think this would be a good idea, as I got a question about it, and wasn't really sure what it even was...If we say no to WP:ABUSE here, then I think it makes sense to remove the abuse section from the templates that include it. TJH2018talk 21:20, 1 May 2016 (UTC)
- I don't see any reason not to do this. No one is making anyone file abuse reports under my proposal, and for those against sending them, any "who denny" desiring to send one can do it without even being a member here because we have the logs out in the open. In fact, I've heard of organizations doing it when vandals post something disparaging about their organzation. I'm just looking to make it a little easier for those who do want to submit them. Besides, it sending abuse reports might at some point save us the headache of dealing with WP:Long term abuse by stopping someone in their tracks early. PCHS-NJROTC (Messages)Growing tired of the bullshit day by day. 05:26, 2 May 2016 (UTC)
Destubbing articles
I was thinking that it might be healthy for wikipedia if we proposed some destubbing contests for a number of countries, including those of the UK to improve general standards and make them more consistent. Get people destubbing articles and try to make it fun. I'm testing out an idea by area of Wales at Wikipedia:WikiProject Wales/Awaken the Dragon/Stub Obliteration/The Preserved County Challenge this week. So if anybody would like to get involved with helping destub a few articles and see how things go this would be warmly appreciated. You don't have to be a contestant in the overall contest, but anybody who can independently help destub a few articles is very welcome. Potentially something like this could eventually eliminate a good number of our stubs for all countries and topics.♦ Dr. Blofeld 12:33, 25 April 2016 (UTC)
- Such a competition has been organized a few times before: Wikipedia:Stub Contest. – Finnusertop (talk ⋅ contribs) 16:40, 25 April 2016 (UTC)
- We've had success at WP:VG with simple cleanup drives, at least one of which I think was triaging and acting on our stubs (either deciding they're perma-stubs and merging as appropriate or actively getting some up to a higher quality level). --Izno (talk) 17:34, 25 April 2016 (UTC)
- We should probably do something like this every so often - the stubs are currently around 38.5% of all articles around here (1,981,113 articles out of 5,144,017). עוד מישהו Od Mishehu 16:12, 26 April 2016 (UTC)
- A scan (it took a while, you can check my results at this link) says that there are 318,714 tagged stubs which are in Category:Living people; this would be about 41.6% of the 765,091 articles in that category. I strongly suspect, although I didn't actually check it out, that many of these are simply AA7 cases that got missed by our New Page Patrollers. עוד מישהו Od Mishehu 20:30, 26 April 2016 (UTC)
- I know there's a good deal of stubs on athletes. WP:NSPORTS is fairly inclusive, so theoretically any athlete that has played a single professional game usually qualifies. There are 7,000 stubs for WP:NFL, 14,000 for WP:TENNIS, 8,000 at WP:BASKETBALL, etc. When all's said and done, wouldn't be surprised to find 100,000 perma-stubs from athletes. They likely do meet GNG, but sources are hard to come by to expand them because sports news decays very quickly. It's hard to find articles on an athlete that played a single season and was a middling player from the 1950s. ~ RobTalk 12:24, 27 April 2016 (UTC)
- It's hard to fine online sources for that. The solution may be to find people to volunteer to check out a book or something from a library or something and flesh out a hundred or so stubs based on a source. I find it's easier to work off a topic than it is for a single article sometimes. Perhaps those projects should make a contest against each other on a percentage basis or something. -- Ricky81682 (talk) 02:18, 30 April 2016 (UTC)
- I know there's a good deal of stubs on athletes. WP:NSPORTS is fairly inclusive, so theoretically any athlete that has played a single professional game usually qualifies. There are 7,000 stubs for WP:NFL, 14,000 for WP:TENNIS, 8,000 at WP:BASKETBALL, etc. When all's said and done, wouldn't be surprised to find 100,000 perma-stubs from athletes. They likely do meet GNG, but sources are hard to come by to expand them because sports news decays very quickly. It's hard to find articles on an athlete that played a single season and was a middling player from the 1950s. ~ RobTalk 12:24, 27 April 2016 (UTC)
- A scan (it took a while, you can check my results at this link) says that there are 318,714 tagged stubs which are in Category:Living people; this would be about 41.6% of the 765,091 articles in that category. I strongly suspect, although I didn't actually check it out, that many of these are simply AA7 cases that got missed by our New Page Patrollers. עוד מישהו Od Mishehu 20:30, 26 April 2016 (UTC)
- We should probably do something like this every so often - the stubs are currently around 38.5% of all articles around here (1,981,113 articles out of 5,144,017). עוד מישהו Od Mishehu 16:12, 26 April 2016 (UTC)
- We've had success at WP:VG with simple cleanup drives, at least one of which I think was triaging and acting on our stubs (either deciding they're perma-stubs and merging as appropriate or actively getting some up to a higher quality level). --Izno (talk) 17:34, 25 April 2016 (UTC)
RfC: Preferred protocol for external links
As discussed at Wikipedia:Village pump (technical)/Archive 145#Preferred protocol for external links templates, there seems to be no consensus as to whether, if an external site uses both https and http, our links to it should use the former, the latter, or be in the protocol-agnostic form //example.com/foobar
. This is particularly significant for templated links, such as those in {{TED speaker}}, but also applies to hand-crafted links.
I therefore ask should we:
- a: use
https://
- b: use
http://
- c: use
//
I will post neutrally-worded links to this discussion, in {{Centralized discussion}} and at WT:EL. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:26, 8 April 2016 (UTC)
Restored from archive. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:27, 27 April 2016 (UTC)
Prefer 'a'
Prefer to use https://
- I'll play. Support this option per prior RfC, which established clear consensus for HTTPS for links to Google and Internet Archive—until someone shows how these cases are substantially different from those and therefore need a new debate and separate consensus. ―Mandruss ☎ 18:50, 8 April 2016 (UTC)
- Kaldari (talk) 19:00, 9 April 2016 (UTC)
- Per Mandruss. APerson (talk!) 00:36, 15 April 2016 (UTC)
Prefer 'b'
Prefer to use http://
- Less things that can go wrong. –Be..anyone 💩 22:59, 12 April 2016 (UTC)
- Most websites will upgrade the connection to https:// if the browser supports it, like Wikipedia does, and if we link to http:// websites that do not support https:// with https:// links, then those links will appear to be not working. Tom29739 [talk] 17:36, 13 April 2016 (UTC).
- This RfC has nothing to do with websites that do not support HTTPS. Please read the intro. However, http://news.google.com does in fact upgrade to HTTPS, as you say. That would appear to change the issue somewhat. I wasn't involved in previous discussions, so I don't know whether that was considered there. ―Mandruss ☎ 17:48, 13 April 2016 (UTC)
- Bender235? ―Mandruss ☎ 18:00, 13 April 2016 (UTC)
- Update: The upgrade was apparently being done by my browser (Firefox), not by Google, because I had previously visited news.google.com in that browser using HTTPS. When I tried it in Microsoft Edge, which had never visited that site, it did not upgrade. ―Mandruss ☎ 18:30, 13 April 2016 (UTC)
Prefer 'c'
Prefer to use //
- No reason to force the protocol on people; the encryption adds overhead and some people (e.g. on metered bandwidth) don't want it. Second preference is for 'a'; if we were to force one, it should be the secure one, per common sense. — SMcCandlish ☺ ☏ ¢ ≽ʌⱷ҅ᴥⱷʌ≼ 08:07, 12 April 2016 (UTC)
- @SMcCandlish: One of us is misinformed. As I understand it, this option merely carries forward the existing protocol, which for Wikipedia is now always HTTPS. Thus, 'a' and 'c' are identical in effect, for links from this site. There is no option that does not "force" one protocol or the other; 'a' and 'c' "force" HTTPS and 'b' "forces" HTTP. ―Mandruss ☎ 18:32, 12 April 2016 (UTC)
- If you tell it "//", will it not try https://, and failing that, fall back to http://? I thought that was the point. I might well be misinformed. If it's just an alias for "https://", then yes, let's use that instead. — SMcCandlish ☺ ☏ ¢ ≽ʌⱷ҅ᴥⱷʌ≼ 20:01, 12 April 2016 (UTC)
- @SMcCandlish: One of us is misinformed. As I understand it, this option merely carries forward the existing protocol, which for Wikipedia is now always HTTPS. Thus, 'a' and 'c' are identical in effect, for links from this site. There is no option that does not "force" one protocol or the other; 'a' and 'c' "force" HTTPS and 'b' "forces" HTTP. ―Mandruss ☎ 18:32, 12 April 2016 (UTC)
- As in my experience such a conditional URL will fallback to the lowest common denominator.
{{u|Checkingfax}} {Talk}
22:55, 3 May 2016 (UTC)
Discussion of link protocols
- Speedy close. Did you ignore or simply not see the below from the Archive 145 discussion:
@Pigsonthewing: The January 2014 link [Izno: inserted link] is that general consensus; the individual site consensuses have been specifically to modify all uses of those protocols for those websites to HTTPS via (semi-)automatic method (so as to explicitly suggest that the edits being made do not run afoul of WP:COSMETICBOT). You can probably make an argument after-the-fact, if questioned, that those RFCs show that COSMETICBOT is not infringed on w.r.t. TED links, but it is usually better to show such a priori.
- How is that related to the RfC above? It seems pretty reasonable to ask how an external link should be formed. I saw the VPT discussion and it was a bunch of mumbo jumbo in response to a simple question—what is preferred? Johnuniq (talk) 12:16, 8 April 2016 (UTC)
- I neither ignored nor missed it. Did you ignore or miss me explaining why the claimed consensus is not apparent? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:42, 8 April 2016 (UTC)
- Oops, I followed a link without noticing this is VPT. That is not a good place for an RfC. Please put it somewhere else because this is a high-turnover page focused on technical stuff, not interminable discussions. Johnuniq (talk) 12:19, 8 April 2016 (UTC)
- Be my guest. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:42, 8 April 2016 (UTC)
- Oppose 'c' - Less recognizable as a link to the layman.—Godsy(TALKCONT) 04:14, 12 April 2016 (UTC)
- @Godsy: This seems easily recognisable as a link to me: Layman. How does it not, to you? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:16, 13 April 2016 (UTC)
- Strong Oppose all, there is no community consensus to make this change for all external links. A straw poll on the Village Pump is not sufficient to justify bots rewriting a majority of articles. Bots will be blocked if they implement changes from this poll. The consensus from Archive 127 shows that link modifications are only approved for Google and Internet Archive links [9]. Nakon 06:29, 5 May 2016 (UTC)
Invalid Question
Please explain exactly why you think this is invalid (wrong place, consensus already established, begs the question, etc)
- Invalid Question The question says "there seems to be no consensus" but the strong consensus as established in multiple discussions is to use HTTPS whenever and wherever possible (including external links) and to use HTTP only when HTTPS fails to bring up the desired page. (This is also posted in the wrong place, IMO, but that can be easily fixed with a move and a link to the new location.) --Guy Macon (talk) 15:10, 8 April 2016 (UTC)
-
-
- This pretty much sums it up. Google and Archive.org support and encourage HTTPS connections, but so do Yahoo.com and a couple of other sites. For instance, HTTPS works for a lot of university websites, such as U of Chicago or Brown U, therefore we should link to the secure connection preferably. Only if HTTPS is not supported, we should leave HTTP. --bender235 (talk) 20:14, 13 April 2016 (UTC)
-
The disambiguation gap has been bridged!
The Wikipedia:Disambiguation pages with links/May 2016 competition is underway, but this month is special. Normally, we have two sets of lists from which points can be scored for the competition - the main list containing the top 1,000 most linked disambiguation pages, and the bonus list containing disambiguation pages with four or fewer incoming links. Historically, there has been a gap between these, with the main list having pages with at list six or seven or eight incoming links. However, thanks to the heroic efforts of our disambiguators, we have for the first time ever managed to bridge the gap. That means that as of the start of May 2016, every single disambiguation link in Wikipedia counts for the contest. Go forth and disambiguate! bd2412 T 02:40, 1 May 2016 (UTC)
Election-year prophylactic on coatracking, shoehorning, and the like.
The current discussion at Wikipedia:Articles for deletion/Clinton donors in the Panama Papers is a reminder that even if the BLP articles on the candidates will be closely monitored, an enterprising campaign partisan can still easily attack their candidate of choice through the creation of a new article. What's next, List of shady people seen hanging around Donald Trump properties? List of Bernie Sanders donors who distribute foods containing dihydrogen monoxide? I believe that we need some measure in place to screen the creation of new articles separate from the candidate bios and purporting to describe things relating to the candidate. This might include some means of flagging new articles containing Clinton, Trump, Cruz, Sanders, Kasich, Hillary, Bernie, or Donald in the title to double-check before releasing into mainspace, and some means of holding articles specifically relating to the candidate for at least some glancing review. bd2412 T 23:52, 1 May 2016 (UTC)
- That article was deleted. Looks like existing structures are working just fine to stop the problem. Nothing looks broken to me. --Jayron32 01:28, 2 May 2016 (UTC)
- Yes, it was - after several days, and the participation of 20-something editors in a discussion about whether it was appropriate to have such an article. The problem might not have been discovered as quickly, but for the boldness of the editor who decided to add a link to this article from the subject's heavily watched main article. We could have saved that time and effort. bd2412 T 11:23, 2 May 2016 (UTC)
Save page -> Publish
- The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
Apparently the default text on the button label is about to change from "Save page" to "Publish". I don't think this has been discussed locally yet, so what do people think? Some of the reasons for changing this are articulated on the phabricator link. — Martin (MSGJ · talk) 21:16, 2 May 2016 (UTC)
- Already under discussion at Wikipedia:Village pump (technical)#Change of "Save page" button to "Publish". --Redrose64 (talk) 23:17, 2 May 2016 (UTC)
- Seems reasonable to me for the reasons laid out there - people don't always realise that 'Save' means 'Show to the world right now'. Sam Walton (talk) 23:21, 2 May 2016 (UTC)
- Per WP:MULTI please can we avoid split discussions? --Redrose64 (talk) 23:43, 2 May 2016 (UTC)
- The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
Draft prod
Non-AFC drafts in draftspace have no G13 method for automatic deletion. Or for any review at all. As such, there are hundreds of stale drafts that haven't been edited in a long time. Attempts to add G13 or other criteria have been repeatedly and expressly rejected so instead these pages take up a lot of time at WP:MFD where deletion is somewhat routine. Limiting an idea lab suggestion, I'd like to propose that we allow for a new draft proposed deletion system with the following rules:
- Draftspace drafts that are not tagged with AFC that haven't been edited in six months can be considered.
- If the deletion is proposed, it would go into a dated category similar to Category:Proposed deletion for at least one month at which time any one can remove the proposal for any reason whatsoever. This is similar to the 5 month notification of Category:AfC G13 eligible soon submissions with a month before G13.
- After a month with no deletion, an admin can then decide if it's worth deleting.
I also tag draft with WikiProject so I imagine we can have WP:ALERTS for these draft prods similar to Prods so projects can see if there's anything of interest. Any thoughts? It's much lengthier than an MFD discussion but it's also essentially a single veto and not a consensus. Deletion can still be done at MFD if the prod fails and it still hasn't been improved. -- Ricky81682 (talk) 20:15, 5 May 2016 (UTC)
- Didn't we discuss a similar proposal quite recently? I support something like this, and would prefer G13 to be abolished to be replaced with this draft prod. — Martin (MSGJ · talk) 20:43, 5 May 2016 (UTC)
- I don't think it went anywhere as I recall. I'd like to get it formalized for draftspace at least. -- Ricky81682 (talk) 21:41, 5 May 2016 (UTC)
Oppose. There is no reason to delete good drafts just because people want to drive users away from the site. — Preceding unsigned comment added by 166.171.121.222 (talk) 21:02, 5 May 2016 (UTC)Strike banned user — JJMC89 (T·C) 02:15, 6 May 2016 (UTC)- While we get a lot of G13 requests at WP:REFUND a good 20% or so of them are re-requests for a draft that has seen multiple G13s. Of these, about 60% of them don't get restored because the person requesting re-restoration won't answer the followup question for requests of that nature. ("If we were to undelete this again, what efforts would you take to make it an acceptable article?") —Jeremy v^_^v Bori! 21:31, 5 May 2016 (UTC)