Misplaced Pages

Latest revision as of 22:47, 7 January 2025

• Table of contents
• First discussion
• End of page
• New post

• WP:VPI
• WP:VPIL

• Discussions of technical issues belong at Village pump (technical).
• Discussions of policy belong at Village pump (policy).
• If you're ready to make a concrete proposal and determine whether it has consensus, go to the Village pump (proposals). Proposals worked out here can be brought there.

Before commenting, note:

• This page is not for consensus polling. Stalwart "Oppose" and "Support" comments generally have no place here. Instead, discuss ideas and suggest variations on them.
• Wondering whether someone already had this idea? Search the archives below, and look through Misplaced Pages:Perennial proposals.

Discussions are automatically archived after remaining inactive for two weeks.

• Refining the administrator elections process
• AI-generated images depicting living people
• Blocks for promotional activity outside of mainspace
• Voluntary RfAs after resignation
• Proposed rewrite of WP:BITE
• LLM/chatbot comments in discussions

• view
• edit
• history
• watch
• archive
• talk
• purge

Toward helping readers understand what Wiki is/isn’t

I’ve often noticed confusion on the part of both general readers and editors about what Misplaced Pages articles are AND aren’t. Truth be told, I suspect all of us editors probably had it not only before becoming editors but also well into our Wiki work.

So I got thinking that perhaps a cute (but not overly so!) little information box that would fly in or otherwise attract attention upon accessing a new article could help halt some common misunderstandings or lack of awareness of general readers. Because I think most editors here at the Pump would be aware of many such examples, I hope you’ll forgive my not providing e.g.’s.

(Of course if such an info box were put in place, there’d also need to be a way for readers not to see it again if they so wish.)

I started to check elsewhere at the Pump to see if a similar idea had ever been submitted before, but I couldn’t figure out a relevant search term. And I didn’t want to suggest an outright proposal if anything similar had in fact ever been proposed. So IDEA LAB just seemed a good place to start the ball rolling. Looking forward to seeing where it leads. Augnablik (talk) 10:58, 17 November 2024 (UTC)

I'm a strong supporter of providing more information about how Misplaced Pages works for readers, especially if it helps them get more comfortable with the idea of editing. Readers are editors and editors are readers—this line should be intentionally blurred. I don't know if a pop up or anything similar to that is the right way to go, but I do think there's something worth considering here. One thing I've floated before was an information panel featured prominently on the main page that briefly explains how every reader is an editor and gives some basic resources. Thebiguglyalien (talk) 17:49, 17 November 2024 (UTC)

The problem with putting stuff on the main page is that many (probably most) readers get to Misplaced Pages articles from a search engine, rather than via the main page. Phil Bridger (talk) 17:57, 17 November 2024 (UTC)

Another issue is a large number of these users tend to be on mobile devices, which have known bugs with regards to things like this. —Jéské Couriano v^_^v 20:45, 17 November 2024 (UTC)

The main page gets 4 to 5 million page views each day. And even so, I would guess that people who go out of their way to read the main page are better candidates to become frequent editors than people who treat Misplaced Pages like it's part of Google. Thebiguglyalien (talk) 15:12, 18 November 2024 (UTC)

I wasn't thinking of the main page. What I had in mind was that whenever someone requests to go to an article — irrespective of how he or she entered Misplaced Pages — the information box would fly in or otherwise appear. Augnablik (talk) 17:30, 18 November 2024 (UTC)

I know you weren't thinking of the main page. My reply was to Thebiguglyalien. Phil Bridger (talk) 20:23, 18 November 2024 (UTC)

So I see now. Sorry. Augnablik (talk) 09:44, 20 November 2024 (UTC)

What sort of confusion are you seeking to dispel? Looking over WP:NOT, basically everything on there strikes me as "well, DUH!". I honestly can't understand why most of it has had to be spelled out. --User:Khajidha (talk) (contributions) 13:04, 18 November 2024 (UTC)

@Khajidha, i don't see the box as ONLY to dispel confusion but ALSO to point out some strengths of Misplaced Pages that probably readers wouldn't have been aware of.

A few things that came to my mind: although Misplaced Pages is now one of the world's most consulted information sources, articles should be considered works in progress because ... however, there are stringent requirements for articles to be published, including the use of strong sources to back up information and seasoned editors to eagle-eye them; writing that is objective and transparent about any connection between writers and subjects of articles ... and (this last could be controversial but I think it would be helpful for readers in academia) although not all universities and academic circles accept Wiki articles as references, they can serve as excellent pointers toward other sources.

if the idea of presenting an information box including the above (and more) is adopted, a project team could work on exactly what it would say and look like. Augnablik (talk) 18:58, 18 November 2024 (UTC)

I think that considerably overstates reality (the requirements are not stringent, sources do not have to be strong, many things are not checked by anyone, much less by seasoned editors, hiding COIs is moderately common...).

BTW, there has been some professional research on helping people understand Misplaced Pages in the past, and the net result is that when people understand Misplaced Pages's process, they trust it less. This might be a case of Careful What You Wish For. WhatamIdoing (talk) 19:20, 18 November 2024 (UTC)

Ooops. Well, if stringent requirements, etc., overstate reality, then official Wiki guidance and many Teahouse discussions are needlessly scaring many a fledgling editor! 😱 Augnablik (talk) 19:40, 18 November 2024 (UTC)

All of these points also fall into the "well, DUH!" category. I did, however, want to respond to your statement that "not all universities and academic circles accept Wiki articles as references". I would be very surprised if any university or serious academic project would accept Misplaced Pages as a reference. Tertiary sources like encyclopedias have always been considered inappropriate at that level, as far as I know. --User:Khajidha (talk) (contributions) 19:38, 18 November 2024 (UTC)

Point taken about encyclopedias being generally unacceptable in academic writing.

But as we’re having this discussion in an idea lab, this is the perfect place to toss the ball back to you, Khajidha, and ask how you would describe Misplaced Pages for new readers so they know how it can be advantageous and how it can’t?

As I see it, that sort of information is a real need for those who consult Misplaced Pages — just as customers appreciate quick summaries or reviews of products they’re considering purchasing — to get a better handle on “what’s in it for me.” Augnablik (talk) 20:05, 18 November 2024 (UTC)

I think the logo at the top left already does a pretty good job: "Misplaced Pages: The 💕". Especially if you look at the expanded form we use elsewhere: "Welcome to Misplaced Pages, the 💕 that anyone can edit."--User:Khajidha (talk) (contributions) 12:39, 19 November 2024 (UTC)

@Khajidha, a mere tag saying "The 💕" seems to me just a start in the right direction. The addition of "that anyone can edit" adds a little more specificity, although you didn't mention anything about writing as well as editing. Still, I think these tags are too vague as far as what readers need more insight about.

I'm working on a list of things I'd like to bring to readers' attention, but I'd like to put it away tonight and finish tomorrow. At that point, I'll humbly request you to "de-DUH" your evaluation of my idea. Augnablik (talk) 17:01, 20 November 2024 (UTC)

Seems to me the problem is that people don't understand what an encyclopedia is. That's a "them" problem, not an "us" problem. And what exactly do these readers think editing the encyclopedia would be that doesn't incude writing it? User:Khajidha (talk) (contributions) 17:45, 20 November 2024 (UTC)

Misplaced Pages is very different from the historical concept of encyclopedia. The open editing expands the pool of editors, at the expense of accuracy. -- Shmuel (Seymour J.) Metz Username:Chatul (talk)

Misplaced Pages may have put traditional general encyclopedias out of business, or at least made them change their business model drastically, but it does not define what an encyclopedia is. One example is that Misplaced Pages relies largely on secondary sources, but traditional encyclopedias, at least for the most important articles, employed subject matter experts who wrote largely on the basis of primary sources. It is our job to explain the difference. Phil Bridger (talk) 20:03, 20 November 2024 (UTC)

After a little longer gap between than what I thought it would take to create a list of things I believe all readers need to be aware of from the git-go about what Misplaced Pages is and isn't, due to some challenges in other departments of life, here's what I came up with. It would be in sections, similar to what you see below, each surrounded by a clip art loop, perhaps golden brown, and perhaps a few other pieces of clip art to set it off visually.I wish I knew how to separate paragraphs with line spacing ... I know this looks a little squished.

_____________________________________

New to reading Misplaced Pages articles? Here are some helpful things for you to be aware of about Misplaced Pages. They'll help you get more clearer ideas of how you can use the articles to best advantage.

If you'd like to go into more depth about all this, and more, just go to the article in Misplaced Pages about itself by typing WIKIPEDIA in the Misplaced Pages search field.

Misplaced Pages is a different kind of encyclopedia.

—   Its articles can be written and edited by anyone.

—   They’re supposed to be based completely on reliable outside sources.

—   They can be updated at any time, thus allowing for quick corrections or additions if needed.

—   Misplaced Pages is free.

That’s the main difference between Misplaced Pages and traditional encyclopedias.

BUT:

All encyclopedias serve as starting points where readers can find out about information — especially the main thinking about particular subjects — then follow up as they wish.

Students and researchers: keep in mind that schools and professional research journals don’t accept encyclopedias as references for written papers, but do encourage using them to get some ideas with which to go forward.

Misplaced Pages has become popular for good reason.

—   Misplaced Pages is the world’s largest-ever encyclopedia.

—   It’s consistently ranked among the ten websites people visit most.

—   Because it’s all online, it’s easy to access.

—   Because it’s highly interactive, it’s easy to move around from topic to topic.

Quality standards for writing articles are in place and in action behind the scenes.

—  Misplaced Pages has high standards for choosing the subjects of articles.

—   Misplaced Pages also has high standards for writing articles, especially freedom from bias.

—   Certain editors are assigned to ensure that articles follow Misplaced Pages standards.

— Although differences of opinions naturally arise about whether a particular article does so, there are sets of procedures to work them out and arbiters to step in as needed. Augnablik (talk) 10:18, 25 November 2024 (UTC)

The <br /> tag should take care of line spacing. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 13:49, 25 November 2024 (UTC)

Is this possible to do in Visual Editor instead (I hope)? Augnablik (talk) 13:52, 25 November 2024 (UTC)

Why would you put information about "reading Misplaced Pages articles" in an editing environment?

Also, several things you've written are just wrong. Misplaced Pages is not considered a "highly interactive" website. "Certain editors" are not "assigned to ensure" anything. Misplaced Pages does not have "high standards for writing articles", and quite a lot of readers and editors think we're seriously failing in the "freedom from bias" problem. We might do okay-ish on some subjects (e.g., US political elections) but we do fairly poorly on other subjects (e.g., acknowledging the existence of any POV that isn't widely discussed in English-language sources). WhatamIdoing (talk) 20:14, 28 November 2024 (UTC)

Actually, I think a more magnetic format for this tool I'm hoping can one day be used on Misplaced Pages would be a short series of animated "fly-ins" rather than a static series of points with a loop around each set thereof. Augnablik (talk) 13:51, 25 November 2024 (UTC)

@Augnablik, personally, I think your idea would be great and would help bring new editors to the project, especially with these messages, which seem more focused on article maintenance (more important nowadays imo) than article creation.

JuxtaposedJacob (talk) | :) | he/him | 02:32, 5 December 2024 (UTC)

as unfortunate as it is, people are generally not that smart. Considering the number of people I've had to explain the concept of editing wikipedia to, I'd be shocked if most people know how wikipedia works and what it isn't Mgjertson (talk) 08:44, 17 December 2024 (UTC)

It’s exactly because it does seem to take a lot for some people to get the idea that I‘m convinced something can be done about that when readers first come to Misplaced Pages. Something catchy and animated, in contrast to “chapter and verse.”

Or so many other groups around the world have found. Augnablik (talk) 11:15, 17 December 2024 (UTC)

Idea Labmates …

Because I had such high hopes of being on the trail of something practical to help prevent some of the main misunderstandings with which readers come to Misplaced Pages — and at the same time to foster awareness of how to use it to better advantage — I wonder if a little spark could get the discussion going again. Or does the idea not seem worth pursuing further? Augnablik (talk) 11:05, 30 November 2024 (UTC)

I guess not.

At least for now.

📦 Archive time. Augnablik (talk) 02:53, 3 December 2024 (UTC)

I hope you won't be disheartened by this experience, and if you have any other good ideas will share them with us. There are two stages to getting an idea implemented in a volunteEr organisation:

1. Getting others to accept that it is a good idea.
2. Persuading someone to implement it.

You have got past stage 1 with me, and maybe others, but I'm afraid that, even if I knew how to implement it, it wouldn't be near the top of my list of priorities. Phil Bridger (talk) 09:17, 3 December 2024 (UTC)

Thank you, Phil. No, not disheartened … I think of it as an idea whose time has not yet come. I’m in full agreement about the two stages of idea implementation, plus a couple more in between to lead from one to the other.

When we in the creative fields recognize that continuum and get our egos out of the way, great things begin to happen. Mine is hopefully drying out on the line.😅 Augnablik (talk) 09:41, 3 December 2024 (UTC)

A starters guide with most common things you need to know and problems you will come up against would be good Sharnadd (talk) 11:37, 30 December 2024 (UTC)

Do please keep going, @Sharnadd. Augnablik (talk) 13:31, 30 December 2024 (UTC)

Thank you I will Sharnadd (talk) 15:46, 30 December 2024 (UTC)

"Sensitive content" labels (only for media that is nonessential or unexpected for an article's subject)

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.

You see, many Misplaced Pages articles contain images or other media that are related to the article's subject, but that readers might not want to see, and have no way of avoiding if they are reading the article without prior knowledge of its contents.

For instance, the article Human includes an image which contains nudity. This image is helpful to illustrate the article's subject, but many people who read this seemingly innocuous article would not expect to see such an image, and may have a problem with it.

Of course, if someone decides to read the article Penis and sees an image of a penis, they really can't complain, since the image would just be an (arguably, essential) illustration of the article's subject, and its presence can easily be known by the reader ahead-of-time.

My solution to this is to have editors look for media or sections of an article which could be seen as having a different level of maturity compared to the rest of the article's content, then ensuring that the reader must take additional action in order to see this content, so that readers of a seemingly innocuous article would not have to see content that could be considered "shocking" or "inappropriate" when compared to the rest of the article's content, unless they specifically choose to do so.

I posted this idea here so other people could tell me what they think of it, and hopefully offer some suggestions or improvements. -A Fluffy Kitteh | FluffyKittehz User Profile Page 15:56, 10 December 2024 (UTC)

As with just about every other proposal related to "sensitive" or "shocking" content it fails to account for the absolutely massive cultural, political, philosophical and other differences in what is meant by those and similar terms. On the human article, at least File:Lucy Skeleton.jpg, File:Anterior view of human female and male, with labels 2.png, File:Tubal Pregnancy with embryo.jpg, File:Baby playing with yellow paint. Work by Dutch artist Peter Klashorst entitled "Experimental".jpg, File:Pataxo001.jpg, File:HappyPensioneer.jpg, File:An old age.JPG, File:Human.svg and quite possibly others are likely to be seen as "shocking" or "sensitive" by some people - and this is not counting those who regard all depictions of living and/or deceased people as problematic. Who gets to decide what content gets labelled and what doesn't? Thryduulf (talk) 16:18, 10 December 2024 (UTC)

Who gets to decide? Editors, by consensus, just like everything else.

But more pointfully, @FluffyKittehz, our usual advice is not to do this, and (importantly) to be thoughtful about image placement. For example, decide whether a nude photo is better than a nude line drawing. Decide whether the nude image really needs to be right at the top, or whether it could be a bit lower down, in a more specific section. For example, the nude photos in Human are in Human#Anatomy and physiology, which is less surprising, seen by fewer users (because most people don't scroll down) and more understandable (even people who dislike it can understand that it's relevant to the subject of anatomy).

BTW, the people in that particular nude photo are paid professional models. They were specifically hired, about a dozen or so years ago, to make non-photoshopped photos in the non-sexualized Standard anatomical position (used by medical textbooks for hundreds of years). I have heard that it was really difficult for the modeling agency to find anyone who would take the job. WhatamIdoing (talk) 03:53, 11 December 2024 (UTC)

First, if you, dear reader, have a tendency to mouse over bluelinks much as I do, I'd suggest not doing so without first reading what I'm linking to.

There are certainly some pages where NOTCENSORED is taken more than a tad too far. My opinion is that if there exists a diagram that would do a comparable job in depicting an objectionable subject, the diagram is to be preferred to the photograph. We sometimes do a pretty good job of using diagrams, just look (or don't, your choice) at where Seedfeeder's illustrations are used.

Also, I think a diagram (even if inferior) is preferable in the lede, so as not to shock readers who open (or even mouse over) the page. The images human are alright in comparison. We're perhaps the only esteemed publication which has images reasonably portrayable as pornographic, and I don't think it's a good look. JayCubby 23:12, 17 December 2024 (UTC)

if there exists a diagram that would do a comparable job in depicting an objectionable subject, the diagram is to be preferred to the photograph. Which subjects are "objectionable"? Who gets to decide? What if there is disagreement about whether a diagram does a "comparable" job? What about those who think a diagram is equally (or even more) objectionable to a photograph? Thryduulf (talk) 01:03, 18 December 2024 (UTC)

@Thryduulf By 'objectionable', I mean subjects that are considered to be objectionable on a fairly brad scope. There are very few places (let's say the Western world for sake of argument, but this would probably hold true across the world) where a photograph of an erect human penis or a woman pleasuring herself with an electric toothbrush wouldn't be taboo if put on a billboard. There are few (but certainly more than the above) public places where it's acceptable to parade around in one's birthday suit. That I think we can agree on. I'm not giving a concrete definition, because norms do vary across cultures, but there is a baseline of what most people agree on.

The reason we have media at all in articles, including for human penis or Female ejaculation, is to describe the subject matter. In some circumstances, the subject matter might be best not illustrated with a photograph (some aspects of anatomy, sexuality, society), or would be adversely affected by not having a photograph or video (.

On the diagram bit, I think that diagrams are almost always less offensive than images, certainly so in the case of simply objectionable subject matters. JayCubby 14:53, 18 December 2024 (UTC)

1) what would be taboo on a billboard is not relevant to an encyclopedia. You mention "public places". This isn't a public place. We are not throwing these images out to the public with no warning. They are used to illustrate articles on the subject depicted. And, before you mention "bystanders" seeing what you are looking at: a) they need to not be so rude as to do that and b) if you worry about it so much, don't look at Misplaced Pages in public

2) "the subject matter might be best not illustrated with a photograph" I would be interested in what things you think could be best illustrated by not showing them. Because I can't really think of any. --User:Khajidha (talk) (contributions) 15:36, 18 December 2024 (UTC)

Re #1: I used a billboard as a more extreme example. I'd argue that we are throwing those images out to the public without warning. Were I to look at what other books or websites (not just encyclopedias) addressed to the general public informing people on the topic, I'd be hard-pressed to find instances where photographs are put as we do. Readers don't expect Misplaced Pages to be any different.

2. It was late when I wrote the above, I posted the unfinished bit earlier today. What I mean is there are cases where a diagram is sufficient and a photograph wouldn't add anything but shock value. JayCubby 17:06, 18 December 2024 (UTC)

Other books in general and other websites in general are also not relevant. We are an encyclopedia. And we aim to be the most comprehensive one ever. And, no, we are not throwing things out to the public. We are allowing the public to access our work. You come here for information on a topic. We provide it. Including relevant images. --User:Khajidha (talk) (contributions) 17:45, 18 December 2024 (UTC)

(edit conflict) objectionable on a fairly brad scope so that means we should regard everything that is objectionable to any large culture, such as Christians, Muslims, Hindus, Jews, Americans, Indians, Chinese, Nigerians, etc (there is no single "western" culture)? Or do you mean only those cultures you are personaly familiar with? or perhaps agree with? Personally I find File:Redneck Revolt Armed Demonstration.jpg far more objectionable than an erect human penis.

I think that diagrams are almost always less offensive than images You are entitled to your opinion, but how representative is it? Why does your opinion matter more than e.g. my opinion or an Islamic cleric's opinion, or a pornographer's opinion? simply objectionable subject matters what does this mean in objective terms? Simply objectionable to whom? Thryduulf (talk) 15:41, 18 December 2024 (UTC)

On the first point, I mean there are things that Christians, Muslims, Hindus, Jews, Americans, Indians, Chinese, and Nigerians would agree to be objectionable. As I said, there's a baseline. I didn't suggest censoring everything anybody is offended by.

On the second, see above for the audience. Can you state instances of where diagrams are in fact more offensive than photographs of the same subject? JayCubby 17:10, 18 December 2024 (UTC)

Obviously there isn't a baseline. Otherwise we wouldn't be having this discussion. You have not mentioned even a single thing that I would object to being illustrated in a comprehensive encyclopedia.--User:Khajidha (talk) (contributions) 17:48, 18 December 2024 (UTC)

There is a baseline taboo against depictions of sexual abuse of children, and we kick people who disagree with this baseline off the project. —Kusma (talk) 19:36, 18 December 2024 (UTC)

Thank you for finally finding an example. I still doubt that there is much more that could be agreed on.--User:Khajidha (talk) (contributions) 19:44, 18 December 2024 (UTC)

The primary reason we do not display images depicting sexual abuse of children is that nobody has uploaded any freely licensed images of this subject that we can legally host. If a free image depicting this exists (not impossible) that we can legally host (currently extremely unlikely) and is uploaded then we will include it in any articles where it is encyclopaedically relevant and due (whether there are any such articles is unknowable without seeing the image).

Off the top of my head, maybe an annotated diagram about a homemade bomb would be more offensive than a photograph of a bomb? There are certainly no shortage of examples where, to at least some people, diagrams are equally offensive as photographs.

I didn't suggest censoring everything anybody is offended by. then you need to state how you are choosing which things to censor. Whose opinions matter? How many people being offended by something is enough? Or does it matter who it is? Thryduulf (talk) 19:53, 18 December 2024 (UTC)

Jfc, that is not the primary reason. Even if we had a freely-licensed image, and WMF Legal was like "sure, go ahead," we would not go ahead. Levivich (talk) 20:01, 18 December 2024 (UTC)

It's obviously hypothetical given that such an image does not currently exist (and I can't think of an image that would be both encyclopaedically relevant* and legal), but if it did you would need to explain why NOTCENSORED didn't apply. Any arguments that an image were not DUE would have to be based on things other than "I don't like this image" or "I don't like the subject of this image".
*Some years ago I remember images of FBI child pornography raids and/or of specific people convicted of child pornography were proposed to illustrate the Child pornography article, but rejected for not being clearly related enough/on BLP grounds. Thryduulf (talk) 20:48, 18 December 2024 (UTC)

• WP pretty explicitly doesn't care if someone finds content offensive. Penises and vaginas are things that exist. Anatomically correct images of penises and vaginas are educationally useful. Anatomy isn't pornography. GMG 16:43, 18 December 2024 (UTC)

I do wonder if we should be considering sources when discussing this topic. Including a graphic image in an article, when sources do not typically include such an image, could be viewed as undue weight or a type of original research. It’s normal for anatomy textbooks to contain pictures of anatomy, so it should be normal for our anatomy articles to include that type of picture too. Barnards.tar.gz (talk) 21:08, 18 December 2024 (UTC)

Yes, it's appropriate to follow the sources' lead in choosing images.

We also have guidelines against the WP:GRATUITOUS inclusion of Misplaced Pages:Offensive material – and the near-total absence of disputes, for many years, about when and whether that guideline relevant pretty much disproves the "but nobody can possibly decide what's offensive" whingeing above – and we require that illustrations be WP:PERTINENT, and MOS:LEADIMAGE says that "Lead images should be of least shock value; an alternative image that accurately represents the topic without shock value should always be preferred". We comply with foundation:Resolution:Controversial content, which requires that readers not be astonished to discover (for example) sexual content on a page through methods such as (a non-exhaustive example) not putting sexual photos in articles that aren't about sexual content or even (for the advanced class) adding quick descriptions, so that people who might hover over or click on a link will know what it's about, so that "the sexual practice of ____" instead of just "____".

This is not that difficult. We don't "label" the images, as suggested above, but we do generally make decent choices, and where we could do better, we invite editors to WP:Be bold in making Misplaced Pages more closely conform with the long-standing policies and guidelines. WhatamIdoing (talk) 00:02, 19 December 2024 (UTC)

Opt-in content warnings and image hiding

A recent discussion about sensitive images at VPP became quite heated, for reasons, but there actually appears to be little to no opposition to developing opt-in features to help readers avoid images that they don't want to see. Currently the options are very limited: there are user scripts that will hide all images, but you have to know where to find them, how to use them, and there's no granularity; or you can hide specific images by page or filename, which has obvious limitations. I therefore thought I'd bring it here to discuss ideas for improving these options.

My idea would be to implement a template system for tagging images that people might not want to see, e.g. {{Content warning|Violence|]}} or {{Content warning|Sex|]}}. This would add some markup to the image that is invisible by default. Users could then opt-in to either hiding all marked images behind a content warning or just hiding certain categories. We could develop a guideline on what categories of content warning should exist and what kind of images they should be applied to.

A good thing about a system like this is that the community can do almost all of the work ourselves: the tagging is a simple template that adds a CSS class, and the filtering can be implemented through user scripts/gadgets. WMF involvement on e.g. integrating this into the default preferences screen or doing the warning/hiding on the server side would be a nice-to-have, not a must-have. – Joe (talk) 07:34, 11 December 2024 (UTC)

Oh also, I suggest we strictly limit discussion here to opt-in systems—nothing that will change the current default of all images always being visible as-is—because experience shows that, not only is consensus on this unlikely to change, but even mentioning it has a tendency to heat up and derail discussions. – Joe (talk) 07:36, 11 December 2024 (UTC)

Would there be a way to tag or list the images themselves, rather than needing to recreate new template coding for each use? CMD (talk) 08:32, 11 December 2024 (UTC)

That would make sense, but since the images are (mostly) on Commons I couldn't figure out a way of doing it off the top of my head. It would also mean that control of what and how things were tagged would be on another project, which always tends to be controversial on enwiki. – Joe (talk) 08:56, 11 December 2024 (UTC)

From the experience with spoiler warnings, these things tend to proliferate if they exist at all. I would rather stay with the clean policy of no warnings whatsoever than discuss whether to introduce warnings for certain classes of offensive things. I am personally offended by the use of "His Royal Highness" or similar words when referring to citizens of Germany like Mr Prinz von Preussen, but I think it is better not to have a category of pictures offending German anti-monarchists. Even if we do not do the censoring ourselves, I oppose spending volunteer time on implementing something that can be used as a censorship infrastructure. —Kusma (talk) 09:33, 11 December 2024 (UTC)

This would retain the policy of no warnings because they would be invisible to anybody who didn't opt-in. Similarly, only volunteers who want to use their time in maintaining this system would do so. – Joe (talk) 10:45, 11 December 2024 (UTC)

I also was reminded of the spoiler tag fiasco. Only at least we can agree spoiler tags would be on any and all plot summaries. Dronebogus (talk) 17:31, 11 December 2024 (UTC)

Another recent discussion at Misplaced Pages:Village_pump_(proposals)#"Blur_all_images"_switch. Gråbergs Gråa Sång (talk) 10:04, 11 December 2024 (UTC)

Strongest oppose to tagging system, for which there was pretty clear consensus against in the previous discussion. It is against the spirit of Misplaced Pages and would be a huge headache for an end that goes against the spirit of Misplaced Pages. This project should not be helping people hide from information. ꧁Zanahary꧂ 15:33, 11 December 2024 (UTC)

• Support: I don't see why would anyone oppose it. And since I have little knowledge on technical stuff, I don't have anything to add to this idea.

☆SuperNinja2☆ TALK! 17:59, 11 December 2024 (UTC)

@Super ninja2: you don’t vote at the Idea Lab. Zanahary is admittedly falling foul of this rule too but I’ll give it a pass as “I am so passionate about this I will vote rhetorically”. Dronebogus (talk) 18:06, 11 December 2024 (UTC)

Sorry, I didn’t realize we don’t vote here. How are we supposed to voice opposition to an idea? Just exclude the bolded vote? ꧁Zanahary꧂ 18:36, 11 December 2024 (UTC)

You don't. You criticize and give your opinion to fix. ☆SuperNinja2☆ TALK! 18:49, 11 December 2024 (UTC)

I don't voice opposition to an idea? Here's my criticism: tagging to appeal to sensitivities that would have certain types of information and imagery hidden is validating those sensitivities, which is not the place of Misplaced Pages (and is against its spirit), and enables the concealment of informationm which is diametrically opposed to the spirit of Misplaced Pages. My proposed "fix" is to not pursue this content-tagging idea. ꧁Zanahary꧂ 19:23, 11 December 2024 (UTC)

I actually thought so. Saw Zanahary voting and thought maybe I was wrong. ☆SuperNinja2☆ TALK! 18:48, 11 December 2024 (UTC)

I haven’t seen anyone bring this up, but this clearly goes against WP:No disclaimers. Please consider this a constructive note about the obstacles you will face if you try to add content warnings to Misplaced Pages. ꧁Zanahary꧂ 17:23, 16 December 2024 (UTC)

Having a general Opt-in system of blurring or hiding all images would be no problem. Having one based on tags, content, categories... would be largely unmaintainable. If you create an "opt-in here to hide all sexual images", then you have to be very, very sure that you actually can do this and not give false promises to readers. But as there is no agreement on where to draw the line of what is or isn't sexual, nudity, violence, disturbing, ... this will only lead to endless edit wars without possible resolution. Are the images on Breastfeeding sexual? L'Origine du monde? Liberty Leading the People (ooh, violence as well!)? Putto? Pavilion of Human Passions? Fram (talk) 10:03, 11 December 2024 (UTC)

Exactly. One of the issues is that some people think there is a thing such as non-sexual nudity, while others think that nudity is always sexual. —Kusma (talk) 10:14, 11 December 2024 (UTC)

So we could have a category "nudity" instead of or in addition to "sex". Part of the proposal here is coming to a consensus on which categories should exist and on guidelines for their use. I don't see how we can conclude that this is an impossible or impractical task before even trying. We manage to draw lines through grey areas all the time. – Joe (talk) 10:44, 11 December 2024 (UTC)

"Trying" would be a massive task, so deciding whether it seems feasible or not before we start on it seems the wisest course of action. We get endless discussions and RfC about whether something is a WP:RS or not all the time, to have this kind of discussion about which tags we should have and then which images should be part of it will multiply this kind of discussions endlessly. Should The Adoration of the Magi (Geertgen tot Sint Jans) be tagged as nudity? Buttocks? Is File:Nipple of male human.jpg nudity? File:African Breast SG.jpg? If male nipples are nudity, then File:Michael Phelps wins 8th gold medal.jpg is nudity. If male nipples aren't nudity, but female nipples are nudity, then why one but not the other? Fram (talk) 11:04, 11 December 2024 (UTC)

TRADITION!! Gråbergs Gråa Sång (talk) 11:07, 11 December 2024 (UTC)

As with everything, we'd have to reach a consensus about such edge cases either in general or on a case-by-case basis. It's not for me to say how that would go with these examples, but I'd suggest as a general principle we should be descriptive rather than normative, e.g. if there is a dispute about what constitutes male nudity, then break the category down until the labels are uncontroversial – "male nudity (upper body)" and so on. – Joe (talk) 13:50, 11 December 2024 (UTC)

These aren't edge cases though. The more you have to break it down, the more work it creates, and the disputes will still continue. Will we label all images of women/men/children/other? All images of women showing any flesh or hair at all? Basically, we will need to tag every image in every article with an endless series of tags, and then create a system to let people choose between these endless tags which ones they want to hide, even things most of us might find deeply unsettling to even offer as an option? Do we want people to be able to use Misplaced Pages but hide all images of transgenders? All images of women? All images of Jews? Everything that isn't halal? In the 4 images shown below, the one in the bathtub is much more sexual than the one in the shower, but the one in the shower shows a nipple, and the other one doesn't. Even to only make meaningful categories to indicate the difference between those two images would be quite a task, and then you get e.g. the other image showing an artwork, which again needs a different indication. It seems like madness to me. Fram (talk) 14:05, 11 December 2024 (UTC)

There are just so many things that some people don't want to see... Dead Australians or Baháʼu'lláh are among the easier ones that might look near harmless to tag. However, people will also demand more difficult things like "images not appropriate for 12 year olds" that have no neutral definition (and where Europeans and Americans have widely differing opinions: just look for typical film ratings where European censors think sex, nudity, drug use and swearing are ok but violence is not, and American censors will think the opposite). There are also things some people find offensive that I am not at all ok with providing a censorship infrastructure for: images depicting mixed-race couples, images depicting trans people, images depicting same-sex couples. I do not think Misplaced Pages should help people avoid seeing such images, so I do not want us to participate in building a censorship infrastructure that allows it. —Kusma (talk) 11:18, 11 December 2024 (UTC)

Alternatives like Hamichlol exists. Gråbergs Gråa Sång (talk) 11:21, 11 December 2024 (UTC)

The English Misplaced Pages community would control which categories are used for this system and I am confident they would reject all of these examples. "People will make unreasonable demands" does not sound like a good reason not to do something. – Joe (talk) 13:44, 11 December 2024 (UTC)

I am confident they would reject all of these examples Why? On what objective grounds are you labelling those examples as "unreasonable"? Why are your preferences "reasonable"? Thryduulf (talk) 14:14, 11 December 2024 (UTC)

Because if there's one thing the English Misplaced Pages community is known for, it'a always agreeing on everything?

This project already has enough things for ongoing arguments over. Making lists of what people may want to avoid and ranking every image on whether it falls into that list is a tremendous effort that is bound to fail. (The thread calling for such categorization on the policy page is an excellent example.... a user felt they were harmed by an image of a dead man smiling... only it seems not to be a dead man, we were supposed to police that image based on how they would misinterpret it.) I'm also wondering if we risk civil litigation if we tell people that we're protecting against image-type-X and then someone who opted out of seeing such images views something that they consider X.

This is just one more impediment to people adding information to the encyclopedia. I can't see that this censorship system would make more people enthusiastic to edit here (and if it did, I'm not sure we'd really want the sort of editor it would encourage.) -- Nat Gertler (talk) 14:39, 11 December 2024 (UTC)

One more general problem with the proposal is that you do not know whether people will be forced to "opt in" by "well meaning" system administrators trying to censor what can be accessed from their system. Having machine readable tags on images makes it very easy to do so and also easy to remove people's ability to click through and see the content. We should not encourage volunteer efforts on supporting such censorship infrastructures. —Kusma (talk) 11:46, 11 December 2024 (UTC)

I don't think the specific proposal here, placing templates in articles (even if they default to not obscuring any images), would be workable. It's too big of an opportunity for activist editors to go on mass-article-editing sprees and for people to edit war over a particular instance of the template. You'd also have to deal with templates where simply wrapping the image in a template isn't currently possible, such as Template:Speciesbox. If people really want to pursue this, I think it'd be better to figure out how to tag the images themselves; people will still probably fight over the classifications, but at least it's less likely to spill over into disrupting articles. Anomie⚔ 12:45, 11 December 2024 (UTC)

The idea was that, since these templates would have no effect if not someone has not opted-in to hiding that specific category of image, people who do not want images to be hidden would be less likely to fight over it or be worried about what "activist editors" are doing. The idea that Misplaced Pages should not be censored for everyone has solid consensus behind it, but the position some are taking here, that other people should not be allowed an informed choice of what not to see, strikes me as quite extreme. – Joe (talk) 13:40, 11 December 2024 (UTC)

You were given all the information you need by the very fact that this is an encyclopedia. There WILL be things here to upset you. --User:Khajidha (talk) (contributions) 15:06, 11 December 2024 (UTC)

I dispute your good-faith but naive assertion that these templates would have "no effect on people who have not opted in". If you tag images systematically, you make it easy to build proxies (or just censored forks) that allow high schools in Florida to ensure their students won't be able to click through to the photo explaining how to use contraceptives. There is no innocent "only opt-in" tagging; any such metadata can and will be used for censorship. Do you really want us to be in the business of enabling censorship? —Kusma (talk) 15:14, 11 December 2024 (UTC)

Well yes, the proposal literally to enable censorship. For those who want it. It may be that it is used by network administrators as you suggest, we can't stop that, but that's between them and their users. I agree that censorship should not affect what editors include in our content but I find the idea that we can enforce our ideal of Zero Sensitivity Free Speech to a global readership also very naive (and frankly a little creepy; I keep picturing a stereotypical Wikipedian standing in front of a Muslim child screaming "no you WILL look at what we show you, because censorship is bad and also what about Renaissance art"). A silver lining could be that the option of controlling access to our content in a fine grained way may convince some networks to allow partial access to Misplaced Pages where they would otherwise completely block it. – Joe (talk) 16:58, 12 December 2024 (UTC)

We are not in the business of enabling censorship, voluntary or otherwise, because voluntary censorship very quickly becomes involuntary cesnsorship. We are in the business of providing access to information, not inhibiting access to information. Thryduulf (talk) 17:07, 12 December 2024 (UTC)

"We're not in the business of leaving the phrase 'rimjob' to your imagination, Timmy, we're in the business of providing access to artistic depictions of bunny sex!" he screamed, and screamed, and screamed... you guys are really silly sometimes. – Joe (talk) 17:31, 12 December 2024 (UTC)

I've seen enough arguments over people doing mass edits and otherwise fighting over invisible stuff in articles, including complaints of watchlist flooding, to think this would be any different. Anomie⚔ 00:17, 12 December 2024 (UTC)

* I would support an opt-in that turned off or blurred all images and made them viewable with a click. I would absolutely object to anything that used some categorization system to decide which images were potentially offensive to someone somewhere. There would be systemic sexism in such categorization because of different cultural norms. Valereee (talk) 12:56, 11 December 2024 (UTC)

Here are four images of adult women touching their own breasts. Do we categorize all of them as potentially offensive? Valereee (talk) 13:10, 11 December 2024 (UTC)

Yes, or at least the three photographs. I'm standing on a crowded subway car and just scrolled past three pics of boobs. Totally unexpected, totally would have minimized/blurred/hidden those if I could, just for the other people around me. It has nothing to do with being offensive, I'm just in a place where pictures of boobs are not really OK to have on my phone right now. And I live in a free country, I can only imagine what it might be like for others. Levivich (talk) 15:16, 11 December 2024 (UTC)

If you are in a place where images of boobs are not ok to have on your phone, you should turn off or blur images on wikis in general as you can never guarantee there will be a warning. (As an aside, these images are not far from some that I have seen in on ads in subway stations). —Kusma (talk) 16:15, 11 December 2024 (UTC)

Levivich, I sympathize with the desire not to encounter NSFW content while “at work”. But your standard here is “not safe for a crowded American or British public space”, which admittedly is the default for the Internet as a whole. But on Wikimedia we at least try to respect the fact that not everyone has that standard. Dronebogus (talk) 17:49, 11 December 2024 (UTC)

It really doesn't feel like we're trying to respect anyone, based on this and related discussions. We seem to be saying to anybody who has personal or cultural sensitivities about any kind of image (so the majority of humankind) that they can either accept our standard of WP:NOTCENSORED or to not see any images at all. We're saying we can't possibly let your kids have the full experience of our educational images while also avoiding photos of dead bodies or graphic depictions of penetrative sex, because what about male nipples? – Joe (talk) 17:04, 12 December 2024 (UTC)

I don't think anyone is saying that people should not see images at all... simply that if they are concerned about seeing images, they get to be the ones to decide which images they should see by clicking on that image. For them to make it our responsibility to guess which pictures they'll want and be the baddies when we're wrong is not respecting them and their ability to make decisions for themselves. (And I'm not sure that you can say we're giving anyone the "full experience of our educational images" when you are hiding some of them.) -- Nat Gertler (talk) 21:10, 12 December 2024 (UTC)

Yes because what about male nipples. Because what about female nipples? Lots of more liberal-minded legal guardians wouldn’t oppose children seeing those. Or even full nudity. Or even dead bodies and penetrative sex! And then we have to go the whole opposite direction ad absurdum with women in bikinis, and Venus de Milo, and unveiled females, or female humans in general, and Mohammad, and dead aboriginal Australians and spiders and raw meat and Hindu swastikas and poop. Dronebogus (talk) 11:27, 13 December 2024 (UTC)

If a stranger is offended by an image on your phone, remind them that they are being very rude by looking at it. --User:Khajidha (talk) (contributions) 20:57, 11 December 2024 (UTC)

Try that with the policeman looking over your shoulder in the country where accessing "indecent" images gets you imprisoned. – Joe (talk) 17:06, 12 December 2024 (UTC)

Pretty much every image of a human being (and plenty of other subjects) has the potential to be regarded as indecent somewhere. This means there are exactly two options that can achieve your desired outcome: censor all images, or assigned every image, individually, to one or more extremely fine-grained categories. The first already exists, the second is completely impractical. Thryduulf (talk) 17:11, 12 December 2024 (UTC)

Then DON'T GO TO A WEBSITE THAT YOU SHOULD REASONABLY EXPECT TO HAVE SUCH COTENT. Such as an encyclopedia.--User:Khajidha (talk) (contributions) 00:11, 13 December 2024 (UTC)

Someone on the subway asked me to stop looking at pictures of naked people on my phone and I said "WHAT?! I'M READING AN ENCYCLOPEDIA!" Levivich (talk) 00:22, 13 December 2024 (UTC)

I really don’t see why Misplaced Pages should work around the subway-goer looking at your phone and your ability to appease them. Look at another website if you want something censored and safe for onlookers. ꧁Zanahary꧂ 00:28, 13 December 2024 (UTC)

I don't really see why you (or anyone) would be opposed to me having a script that lets me turn off those pictures if I want to. Levivich (talk) 00:36, 13 December 2024 (UTC)

You can have your own script to toggle off every image. You can have a script that runs on an off-wiki index of images you don’t want to see. But to tag images as potentially offensive, I have an issue with, and I hope you understand why even if you don’t agree. ꧁Zanahary꧂ 02:44, 13 December 2024 (UTC)

I’m sorry but your situation is just weird. You should know Misplaced Pages is generally NSFW at this point if you’re complaining about it right now. Dronebogus (talk) 11:45, 13 December 2024 (UTC)

Seems that the problematic behavior here isn't us having the images or you looking at them, it is the random person looking at someone else's screen. We should not be required to modify our behavior because other people behave badly. --User:Khajidha (talk) (contributions) 15:49, 13 December 2024 (UTC)

You can look at other websites if you're in public and an uncensored one would disturb people who might glance at your phone! ꧁Zanahary꧂ 21:06, 11 December 2024 (UTC)

And how do we categorize these in order to allow "offensive" images to be blurred, @Levivich? Valereee (talk) 22:42, 16 December 2024 (UTC)

@Valereee: We don't, we let the people who want to hide images decide which images they want to hide. They can pick specific images, or categories, or use the Wikidata "depict" info (as Izno mentions below), and there's probably some other ways to do it besides those three. Levivich (talk) 19:01, 17 December 2024 (UTC)

Wouldn't it be simpler to set up a toggle on/off applied locally for all images that can be used by IPs as well as registered accounts? Sorry if I'm completely misunderstanding the tech details.

To be clear, I have no objection to allowing people to decide from among WC’s how many hundreds of thousands of categories which ones they don’t want to see. Sounds like a daunting iterative process if there's a lot someone would rather not be surprised by, but it's their time. And if someone wants to go through WC and make sure everything's categorized, ditto. And I guess someone could leave penises on their list all the time and take boobs off once they get off the subway. :D What I object to is for us in any way to suggest/imply which categories someone might want to block. Valereee (talk) 14:16, 18 December 2024 (UTC)

Yes I totally agree with all of that :-) An image switch would be simpler, and compiling a list would take a lot of time, but it's their time. (I would toggle the switch on the subway to protect myself from boobs and penises!) Levivich (talk) 17:12, 18 December 2024 (UTC)

Browsers already have a toggle so they can avoid downloading all images. As I discussed in another thread, users who need to limit their downloads of images are likely to need to do this across all web sites, and so handling this restriction on the client side is more effective. isaacl (talk) 19:54, 18 December 2024 (UTC)

Yeah, but if most of your online time is at, like, art or shopping or recipe sites, it seems like kind of a hassle to make someone flip that toggle every time they come to Misplaced Pages when we could just give them a toggle to set here. Again apologies for my tech ignorance. Believe it or not I was an early adopter when I was young. In the early 90s I taught workshops for my professional association in how to build a website. :D Age. It comes for all of us. Valereee (talk) 16:52, 19 December 2024 (UTC)

Some browsers will let you configure settings for specific sites, so you can block images from only Misplaced Pages. It's just more effective for users to have one interface that they can use across all websites, than to have to make adjustments on every website they want to manage. (For a similar reason, Misplaced Pages doesn't dictate a specific font for the body text; it uses the configured default sans-serif font.)

Regarding the tech side, the most straightforward way to implement a setting for non-logged in users without incurring additional caching costs is to use Javascript that is triggered through something stored on the client (such as a cookie), which is how I understand the Vector2022 width setting is done. That introduces a race condition where images may be downloaded before they can get blocked, and potentially shifting layouts, or the entire page load has to be delayed. isaacl (talk) 17:42, 19 December 2024 (UTC)

I would blur all nude or inappropriate images and ask the people who enter on these pages if they are mature because sometimes kids use wikipedia for research and click things not appropriate for their age I would not blur some images on stuff like breast cancer because sometimes people research stuff on that for only educational purposes. Uknown character (talk) 18:24, 1 January 2025 (UTC)

I'd be ok with such an opt-in too, if it can be made. Perhaps such a link/button could be placed in the main meny or floating header. The hamburger too perhaps, for the mobile readers. Gråbergs Gråa Sång (talk) 13:27, 11 December 2024 (UTC)

The idea is not to decide what is and isn't potentially offensive, but to add descriptive labels and then let readers decide what they do and do not want to be warned about. So for example we would not categorise any of your examples as "potentially offensive", but as containing "nudity" or "nude women" or whatever level of granularity was agreed upon. This idea is a reaction to the proposal to obscure all images (which is being discussed elsewhere) because a) letting users choose whether to see an image is only useful if they have some indication of what's behind the blurring and b) quite frankly, I doubt anyone will ever use such an indiscriminate option. – Joe (talk) 13:33, 11 December 2024 (UTC)

One generally does have indications of what is being blurred, both some sense in a blurred image but more importantly by caption. Some ways of hiding all images would ipresent not a blurred image present a filename, and image filenames are largely descriptive. -- Nat Gertler (talk) 15:59, 11 December 2024 (UTC)

Use alt text, the explicit purpose of which is to present a description of the picture for those that cannot see it, rather than file names which can be completely descriptive without describing anything relevant to why someone might or might not want to view it, e.g. the photo of the statue here is File:Antonin Carlès (1851-1919) - La Jeunesse (1883) (12387743075).png. Thryduulf (talk) 18:22, 11 December 2024 (UTC)

That is actually a much better idea than blurring, thanks! Having a "see alt text instead of images" option would not only be more practical for people wanting to know if images are sensitive before seeing them, it would also give more of an incentive to add alt text to begin with. Chaotic Enby (talk · contribs) 18:31, 11 December 2024 (UTC)

I would also support an opt-in to blur all images (in fact, User:Chaotic Enby/blur.js does about that). However, categorizing images with labels whose only purpose is for reader to decide whether they are offensive is, by definition, flagging these images as "potentially offensive", as I doubt a completely innocuous image would be flagged that way. And any such categorization can easily be exploited, as above.Also, the ethical concerns: if some people find homosexuality offensive, does that mean Misplaced Pages should tag all images of gay couples that way? What is the message we bring if gay people have a tag for blurring, but not straight people? Chaotic Enby (talk · contribs) 14:04, 11 December 2024 (UTC)

You might be able to do it using categories, even Commons categories. Instead of (or in addition to) adding images one by one to special maintenance categories, add entire image categories to the maintenance categories. Keep in mind this isn't the kind of thing that needs consensus to do (until/unless it becomes a gadget or preference)--anyone can just write the script. Even the list of categories/images can be maintained separately (e.g. a list of Commons categories can be kept on enwiki or meta wiki or wherever, so no editing of anything on Commons would be needed). It could be done as an expansion of an existing hide-all-images script, where users can hide-some-images. The user can even be allowed to determine which categories/images are hidden. If anyone wants to write such a script, they'd have my support, hmu if you want a tester. Levivich (talk) 15:30, 11 December 2024 (UTC)

As I commented at Misplaced Pages:Village pump (proposals)/Archive 214#Censor NSFW/ NSFL content last month unless you get really fine-grained, Commons categories don't work. For example all these images are in subcategories of Commons:Category:Sex:

Sex → Books about sex → Books about human sexuality, Books about LGBT

Sex → Biology of sex → Sex determination → Haplodiploidy

Sex → Sex in art → Sex (text) → CIL XIII 000129 → Musée Saint-Raymond, Ra 196

Sex → Ejaculation → Ejaculation of humans → Female ejaculation → Rufus, Le Poil

Sex → Females → Female symbols → Women icons → Blank persons placeholders (women)

To get any sort of useful granularity you have to go multiple levels deep, and that means there are literally thousands (possibly tens of thousands) of categories you need to examine individually and get agreement on. And then hope that the images are never recategorised (or miscategorised), new images added to categories previously declared "safe" (or whatever term you choose) or new categories created. Thryduulf (talk) 15:43, 11 December 2024 (UTC)

c:Category:Penis. If someone wrote a script that auto-hid images in that category (and sub-cats), I'd install it. We don't need agreement on what the categories are, people can just make lists of categories. The script can allow users to choose whatever lists of categories they want, or make/edit their own list of categories. One thing I agree about: the work is in compiling the lists of categories. Nudity categories are easy; I suspect the violence categories would be tougher to identify, if they even exist. But if they don't, maintenance categories could be created. (Lists of individual images could even be created, but that is probably too much work to attempt.) Levivich (talk) 15:53, 11 December 2024 (UTC)

Going that private script route, you could also use the category of the article in which it appears in some cases. But I'd worry that folks would try to build categories for the specific reason of serving this script, which would be sliding from choice to policy. -- Nat Gertler (talk) 16:02, 11 December 2024 (UTC)

Nah, still choice. One option is to create new maintenance categories for the script. Another option is for the script to just use its own list of images/categories, without having to add images to new maintenance categories. Levivich (talk) 16:21, 11 December 2024 (UTC)

Allowing maintenance categories designed to hide images is very much a policy issue, no matter how many times you say "nah". The moment that "pictures which include Jews" category goes up, we're endorsing special tools for antisemitism. -- Nat Gertler (talk) 17:04, 11 December 2024 (UTC)

Nah. See, while we have a categories policy, new maintenance categories are not something we "allow" or don't allow -- they're already allowed -- and they don't create a "policy issue" because we already have a policy that covers it. People create new maintenance categories all the time for various reasons -- it's not like we have to have an RFC to make a new template or make a new maintenance category. This is a wiki, have you forgotten? We need consensus to delete stuff, not create stuff.

And you're totally ignoring the part that I've now said multiple times, which is that no new maintenance categories are required. That's one way to skin this cat, but it can also be done by -- pay attention please -- creating lists of categories and images. See? No maintenance category, no policy issue.

Anybody creating a list of "pictures which include Jews" would be violating multiple site policies and the UCOC and TOS. This is a wiki, remember? Did we not have Misplaced Pages because someone might create an antisemitic article? No! We still had a Misplaced Pages, knowing full well that some people will abuse it. So "somebody might abuse it!" is a really terrible argument against any new feature or script or anything on Misplaced Pages.

What are you even opposing here? You have a problem with someone creating a script to hide images? Really? Maybe just ... not ... try to imagine reasons against it? Maybe just let the people who think it's a good idea discuss the implementation, and the people who don't think it's a good idea can just... not participate in the discussion about implementation? Just a thought. It's hard to have a discussion on this website sometimes. Levivich (talk) 17:09, 11 December 2024 (UTC)

Creating a script to hide images is fine. Curating/categorising images to make them easier to hide is not. You are free to do the first in any way you like, but the second should not be done on Misplaced Pages or any Wikimedia project. —Kusma (talk) 17:30, 11 December 2024 (UTC)

Why yes, I can understand why having people who disagree with you about both intent and effect in this matter would be a disruption to the discussion you want to have, with all agreeing with you and not forseeing any problems nor offering any alternate suggestions. I'm not seeing that that would be particularly in the spirit of Misplaced Pages nor helpful to the project, however. "Someone might abuse it and it might require more editorial effort to work it out, all of which could be a big distraction that do not actually advance the goals of the project" is a genuine concern, no matter how many times you say "nah". -- Nat Gertler (talk) 17:40, 11 December 2024 (UTC)

How would hiding pictures of Jews be an abuse? ꧁Zanahary꧂ 18:40, 11 December 2024 (UTC)

If not categories then perhaps that image tagging system commons has? (Where it asks you what is depicted when you upload something). Not sure how much that is actually used though. – Joe (talk) 17:18, 12 December 2024 (UTC)

Using the sub-cats, you would hide e.g. the image on the right side (which is in use on enwiki). Fram (talk) 16:14, 11 December 2024 (UTC)

Yeah, given how Misplaced Pages categorization works (it's really labeling, not categorization), it's well known that if you go deep enough into sub-cats you emerge somewhere far away from the category you started at.

If the cost of muting the Penis category is having the bunny picture hidden, I'd still install the script. False positives are nbd. Levivich (talk) 16:23, 11 December 2024 (UTC)

This is a bad example. It is only used on the article about the objectionable painting it is extracted from. Aaron Liu (talk) 20:53, 13 December 2024 (UTC)

And...? I thought we were hiding objectionable images (and considering that painting as "objectionable" is dubious to start with), not all images on a page where one image is objectionable? Plus, an image that is only used on page X today may be used on page Y tomorrow ("rabbits in art"?). So no, this is not a bad example at all. Fram (talk) 22:54, 13 December 2024 (UTC)

This is no better than the discussion running at the other VP and is borderline forum shopping. I’m disappointed in the number (i.e. non-zero) of competent users vehemently defending a bad idea that’s been talked to death. I keep saying that the only way (no hyperbole) this will ever work is an “all or nothing” opt-in to hide all images without prejudice. Which should be discussed at the technical VP IMO. Dronebogus (talk) 17:37, 11 December 2024 (UTC)

Reactivating the sensitive content tagging idea here feels like forum-shopping to me too. ꧁Zanahary꧂ 18:41, 11 December 2024 (UTC)

oppose as forum-shopping for yet another attempt to try to introduce censorship into the wikipedia. ValarianB (talk) 18:51, 11 December 2024 (UTC)

If people really want a censored Misplaced Pages, are't they allowed to copy the whole thing and make their own site? One WITHOUT blackjack and hookers?--User:Khajidha (talk) (contributions) 21:02, 11 December 2024 (UTC)

Yes, we even provide basic information on how to do it at Misplaced Pages:FAQ/Forking. Thryduulf (talk) 21:26, 11 December 2024 (UTC)

Actually forget the Misplaced Pages and the blackjack! Dronebogus (talk) 14:55, 12 December 2024 (UTC)

Maybe you missed it, ValarianB, but this is the idea lab, so a) as it says at the top of the page, bold !voted are discouraged and b) the whole point is to develop ideas that are not yet ready for consensus-forming in other forums. – Joe (talk) 17:10, 12 December 2024 (UTC)

Maybe you missed it, @Joe, but forum shopping, spending time developing ideas that have no realistic chance of gaining consensus in any form, and ignoring all the feedback you are getting and insisting that, no matter how many times and how many ways this exact same thing has been proposed previously, this time it won't be rejected by the community on both philosophical and practical grounds are also discouraged. Thryduulf (talk) 17:16, 12 December 2024 (UTC)

...you realise you don't have to participate in this discussion, right? – Joe (talk) 17:20, 12 December 2024 (UTC)

Why shouldn't they? They strongly oppose the idea. ꧁Zanahary꧂ 18:07, 12 December 2024 (UTC)

Yes, that's exactly the problem with forum shopping. If you keep starting new discussions and refusing to accept consensus, you might exhaust people until you can force your deeply unpopular idea through.135.180.197.73 (talk) 18:31, 12 December 2024 (UTC)

Because Thryduulf apparently thinks it's a waste of time to do so. And since the purpose of the idea lab is to develop an idea, not propose or build consensus for anything, I tend to agree that chiming in here just to say you oppose something is a waste of (everyone's) time. – Joe (talk) 18:43, 12 December 2024 (UTC)

How? If I were workshopping an idea to make Misplaced Pages cause laptops to explode, a discussion that omits opposition to that idea would be useless and not revealing. ꧁Zanahary꧂ 19:56, 12 December 2024 (UTC)

Because you're not participating to help develop the idea, your participating to stop other people from developing the idea. Brainstorming is not a debate. Brainstorming an idea does not involve people making arguments for why everyone should stop brainstorming the idea.

To use an analogy, imagine a meeting of people who want to develop a proposal to build a building. People who do not think the building should be built at all would not ordinarily be invited to such a meeting. If most of the meeting were spent talking about whether or not to build the building at all, there would be no progress towards a proposal to build the building.

Sometimes, what's needed (especially in the early stages of brainstorming) is for people who want to develop a proposal to build a building, to have the space that they need to develop the best proposal they can, before anybody challenges the proposal or makes the argument that no building should be built at all. Levivich (talk) 20:13, 12 December 2024 (UTC)

The issue here is that image filtering for this purpose is a PEREN proposal, with many of the faults in such a system already identified. Not many new ideas are being proposed here from past discussions. Masem (t) 20:27, 12 December 2024 (UTC)

I don't think this model works for a wiki. There's no committee presenting to the public. This project is all of ours, and if there's so much opposition to a proposal that it cannot be discussed without being overwhelmed by opposition, then I don't see it as a problem that the unpopular idea can't get on its feet. ꧁Zanahary꧂ 20:43, 12 December 2024 (UTC)

Heh. So if three or four people can disrupt an idea lab thread, then that means it was a bad idea... is what you're saying? Levivich (talk) 21:22, 12 December 2024 (UTC)

Sure. Write up the worst interpretation of my comment and I’ll sign it. ꧁Zanahary꧂ 21:43, 12 December 2024 (UTC)

There's no problem with users voluntarily discussing an idea and how it might be implemented. They should, of course, remain aware that just because everyone interested in an idea comes up with a way to proceed doesn't mean there's a community consensus to do so. But if they can come up with a plan to implement an add-on feature such as a gadget, for example, that doesn't impose any additional costs or otherwise affect the work of any other editor who isn't volunteering to be involved, then they're free to spend their own time on it. isaacl (talk) 19:33, 12 December 2024 (UTC)

My personal thought on how this should work is image sorting by category, the onus is completely on the user using the opt-in tool to select categories of images they don't want to see. We don't need to decide for anybody, they can completely make their own decisions, and there's no need for upkeep of a "possibly offensive image list." Just Step Sideways 02:48, 13 December 2024 (UTC)

It’s interesting but I don’t support it. People don’t necessarily get how categories work. “Sex” isn’t about sexual intercourse, but it’ll be at the top of everyone’s block lists. And blocking a huge over-category like violence will block a lot of totally inoffensive images. In other words, this is too technical for most people and will satisfy no-one while catching mostly false positives. Which is actually worse than all-or-nothing. Dronebogus (talk) 11:19, 13 December 2024 (UTC)

A problem with this is that the tail may begin to wag the dog, with inclusion on block lists becoming a consideration in categorizing images and discussions on categorizations. ꧁Zanahary꧂ 15:00, 13 December 2024 (UTC)

I can see that happening, becoming a WP:ETHNICGALLERY-like timesink. Gråbergs Gråa Sång (talk) 15:07, 13 December 2024 (UTC)

I say let stupid people who don't understand what word means make their own mistakes. It might even teach them something. So long as it is opt-in only it won't effect anyone else. El Beeblerino 07:28, 15 December 2024 (UTC)

Suggestion: we let those who think this is a good idea waste hours of their time devising a plan, and then we oppose it once they bring it to WP:VPPR. I guess they have received enough feedback and can look through the archives to see why this is a bad idea which has been rejected again and again. It's their choice if they want to add one more instance of this perennial proposal, if they believe that either the opposes here are a minority and they represent the silent majority somehow, or if they somehow can find a proposal which sidesteps the objections raised here. Fram (talk) 11:46, 13 December 2024 (UTC)

That'd be great, thanks. – Joe (talk) 11:49, 13 December 2024 (UTC)

Break (opt-in content warnings)

So to summarise the constructive feedback so far:

• It'd be better for labels to be attached to images and not to inclusions of them
• It'd be better to use an existing labelling (e.g. categories, captions) rather than a new system
• However it's doubtful if it's feasible to use categories or if they are sufficiently consistent
• An alternative could be to maintain a central list of labels

This suggests to me three, not mutually exclusive approaches: obscure everything any rely on captions and other existing context to convey what's shown (which is being discussed at Misplaced Pages:Village_pump_(proposals)#"Blur_all_images"_switch); develop a gadget that uses categories (possibly more technically complex); develop a gadget that uses a central list (less technically complex, could build lists from categories). – Joe (talk) 12:11, 13 December 2024 (UTC)

Ah, the dreaded “arbitrary break”. Dronebogus (talk) 14:09, 13 December 2024 (UTC)

…this is your summary of feedback so far? How about "many editors believe that marking content as potentially sensitive violates WP:NOTCENSORED and the spirit of an encyclopedia?" ꧁Zanahary꧂ 14:58, 13 December 2024 (UTC)

Seriously could you two stop? Levivich (talk) 15:10, 13 December 2024 (UTC)

That viewpoint has been well-heard and understood, and any actual implementation plan that develops will have to take it into account. isaacl (talk) 17:23, 13 December 2024 (UTC)

If you don't like it, don't use it. WP:NOTCENSORED applies to features or gadgets just as much as it does to content—Misplaced Pages should not hide information about optional content filtering extensions from users by excluding it from the preferences tab. – Closed Limelike Curves (talk) 19:00, 19 December 2024 (UTC)

My main questions would be what the criteria are for deciding what labels to have, and what steps would be taken to minimize the prejudicial effects of those labels (see Question 7 in this ALA Q&A)? (Asking in good faith to foster discussion, but please feel free to disregard if this is too contrarian to be constructive.)--Trystan (talk) 16:49, 13 December 2024 (UTC)

That is an excellent link. —Kusma (talk) 17:33, 13 December 2024 (UTC)

I think it'd be best if the user sets their own exclusion list, and then they can label it however they want. Anyone who wants to could make a list. Lists could be shared by users if they want. Levivich (talk) 18:20, 13 December 2024 (UTC)

One option would be to start with an existing system from a authorative source. Many universities and publishers have guidelines on when to give content warnings, for example. – Joe (talk) 19:23, 13 December 2024 (UTC)

This is a review of what content warnings and trigger warnings exist, not guidelines on when they should be used. It examined electronic databases covering multiple sectors (n = 19), table of contents from multi-sectoral journals (n = 5), traditional and social media websites (n = 53 spanning 36 countries), forward and backward citation tracking, and expert consultation (n = 15), and no encyclopedia. ꧁Zanahary꧂ 19:46, 13 December 2024 (UTC)

Yep, that's why I linked it; to show that we have at least 136 potential models. Though if you read further they do also come up with their own "NEON content warning typology" which might not be a bad starting point either. – Joe (talk) 20:02, 13 December 2024 (UTC)

Do you want to apply it to sensitive articles, too? That seems more in line with the NEON system. ꧁Zanahary꧂ 20:21, 13 December 2024 (UTC)

No. – Joe (talk) 05:58, 14 December 2024 (UTC)

@Joe Roe: and why not? Dronebogus (talk) 15:45, 15 December 2024 (UTC)

It seems like getting something running for images is enough of a challenge, both technically and w.r.t to community consensus. – Joe (talk) 07:59, 16 December 2024 (UTC)

Since it included NO encyclopedias, it looks to me like we have NO models. Possibly because such things are fundamentally incompatible with the nature of an encyclopedia.--User:Khajidha (talk) (contributions) 23:44, 13 December 2024 (UTC)

Bet you can't name three encyclopedias that contain a picture of anal sex. Britannica, World Book, and Encarta don't, in any edition. Seems that not having pictures of anal sex is quite compatible with the nature of an encyclopedia. Misplaced Pages might be the first and only encyclopedia in history that contains graphic images. Levivich (talk) 00:42, 14 December 2024 (UTC)

Sounds like the problem is ith those others.--User:Khajidha (talk) (contributions) 00:55, 14 December 2024 (UTC)

But it does make me wonder whether anything that appears only in Misplaced Pages and not in other general-purpose encyclopedias is accurately described as "the nature of an encyclopedia". That sounds more like "the nature of (the English) Misplaced Pages". WhatamIdoing (talk) 01:18, 14 December 2024 (UTC)

Misplaced Pages has long ago stopped being similar to old general purpose encyclopaedias; it is a sui generis entity constrained only by WP:NOT. We do have massive amounts of specialist topics (equivalent to thousands of specialist encyclopaedias) and try to illustrate them all, from TV episodes to individual Biblical manuscripts to sex positions. —Kusma (talk) 07:40, 14 December 2024 (UTC)

Or those other encyclopedias are deficient. --User:Khajidha (talk) (contributions) 22:33, 14 December 2024 (UTC)

feel free to argue on the anal sex page that we shouldn’t have any images of anal sex. We do. ꧁Zanahary꧂ 01:19, 14 December 2024 (UTC)

I believe that the argument is that since Misplaced Pages is the only (known) general-purpose encyclopedia to include such photos, then their absence could not be "fundamentally incompatible with the nature of an encyclopedia". If the absence of such photos were "fundamentally incompatible with the nature of an encyclopedia", then Misplaced Pages is the only general-purpose encyclopedia that has ever existed. WhatamIdoing (talk) 02:10, 14 December 2024 (UTC)

Why shouldn’t we operate from the idea that Misplaced Pages is the ideal encyclopedia? To me it clearly is. The spirit of an encyclopedia is obviously better served with photos on the article for anal sex than with a lack of them. ꧁Zanahary꧂ 03:09, 14 December 2024 (UTC)

Because, as people who have a significant say in what Misplaced Pages looks like, that would be incredibly solipsistic and automatically lead to the conclusion that all change is bad. – Joe (talk) 06:00, 14 December 2024 (UTC)

Taken to extremes, all philosophies would pitfall into pointlessness. If we exclude illustrating images because Britannica and World Book do too, then we may as well just fuse with either of those, or shut down Wiki because those others have it covered. Photos of an article subject are educational illustrations, and encyclopedias that lack such photos are weaker for it. ꧁Zanahary꧂ 06:20, 14 December 2024 (UTC)

The point is that you shouldn't take an outlier and declare that unusual trait to be True™ Nature of the whole group. One does not look at a family of yellow flowers, with a single species that's white, and say "This one has white petals, and I think it's the best one, so yellow petals are 'fundamentally incompatible with the nature of' this type of flower". You can prize the unusual trait without declaring that the others don't belong to the group because they're not also unusual. WhatamIdoing (talk) 22:47, 16 December 2024 (UTC)

I honestly don’t care about the other encyclopedias. If they wanted my help, I’d tell them to be more like Misplaced Pages, including by illustrating educatively without regard for offense, sensitivity, or shock. And when I say censorship is incompatible with encyclopedias, I’m not comparing against an average of extant encyclopedias; I am comparing against the principles and essence of what an encyclopedia is, which is an educational, organized, thorough compendium of important information as derived from reliable secondary sources. I consider any sacrifice from the informing mission of Misplaced Pages (like hiding some images, let alone marking them as potentially offensive) to be a loss, and I don’t consider making Misplaced Pages more comfortable or calming to be a benefit. That can be handled by pajamas.com or whatever—or by a Misplaced Pages fork that balances reader comfort and sensitivity with information. Not this one, though. ꧁Zanahary꧂ 01:15, 17 December 2024 (UTC)

A good reference work/encyclopedia on human sexuality probably does, though I haven’t gone and checked. ꧁Zanahary꧂ 03:11, 14 December 2024 (UTC)

Well one obvious example would be the Kama Sutra. Nobody complains about that. Dronebogus (talk) 15:42, 15 December 2024 (UTC)

The right approach to take here is to use the depicts statement on Commons images (see also c:Commons:Structured data). This should have a fairly high true positive ratio (compared either to picking out specific images or using categories) as the intention of the property is to be pretty concrete about what's appearing in the file (see also c:Commons:Depicts and/or c:Commons:Structured data/Modeling/Depiction - it's not obvious to me which is the Commons preference for how to depict things). You'll need to figure out which Wikidata items you want to offer which indicate a screened image, but that can start in the penis, Muhammad, internal organ, and sex directions and go from there. The gadget will probably want to support querying the subclass chain of the Wikidata item (property P279) so that you can catch the distinction between any penis and the human penis. My impression of the problem in using depicts statements is that the structured data work on Commons is much younger than the categories work is and so you're probably going to end up with more false negatives than not. It's a wiki though, so the right way to improve those cases should be obvious, and can perhaps even start with a database query today tracking which images used in our articles do not yet have depicts statements. The other problem this direction is that it doesn't take into account images hosted locally since those don't have structured data, but I anticipate the vast majority of the kinds of images this discussion entertains are free images. Izno (talk) 10:09, 14 December 2024 (UTC)

Nobody maintains those things. They’re almost as useless as captions. Dronebogus (talk) 15:43, 15 December 2024 (UTC)

As I said, the work is much younger. There are also detractors in that community. Yet, I expect that there are many people who do use them, and we can ourselves work just on the set of images that are used in our articles. I imagine that set is both small and queryable, especially for potentially offensive images, which itself is a much smaller set than the nearly 7 million articles we have lying around. Izno (talk) 02:39, 17 December 2024 (UTC)

This is sounds like a very promising approach POV, thanks. I have to say I also had the strong impression that the "depicts" feature was abandonware, but then again maybe having a concrete use for the labels will prompt people to create more of them. – Joe (talk) 08:06, 16 December 2024 (UTC)

It seems to get used a lot be people using c:Special:UploadWizard – half of uploads? I have the impression that using it might increase the likelihood of the tagged images being found in relevant searches, but I don't know why I believe that. But since I believe it, I'd encourage people to use it, at least for images that they believe people would want to find. WhatamIdoing (talk) 22:50, 16 December 2024 (UTC)

Indeed, finding users for it besides c:Special:MediaSearch (which does use structured data) does seem like a way to inspire change, as I alluded to at "it's a wiki". Izno (talk) 02:43, 17 December 2024 (UTC)

• I don't see consensus in this discussion to create a new tagging/labelling system or to use existing Commons categories to hide images. People can argue until they're blue in the face, but the proposal(s) will ultimately be rejected at a community-wide RfC. That aside, I don't believe anyone here is opposed to having a toggle button that blurs or hides all images, right? The toggle switch could be placed in the Settings menu (on mobile view) or Appearance menu (on desktop view), and it would be switched off by default (meaning if editors want to blur/hide all images, they would have to manually switch it on). Only the WMF team has the ability to create such a feature, so that logged-out users can use it and logged-in users won't need to install user scripts. That idea could be suggested at the m:Community Wishlist. Some1 (talk) 15:31, 15 December 2024 (UTC)

  At the VPPro discussion this was forked from opposition has been expressed. Thryduulf (talk) 15:46, 15 December 2024 (UTC)

  @Some1: This is the idea lab. Discussions here are explicitly not about developing consensus one way or another (see the notice at the top of this page). The blur all images approach is being discussed elsewhere (linked several times above) and I would prefer to keep this on the original topic of labelled content warnings. – Joe (talk) 08:01, 16 December 2024 (UTC)

  Probably some of why you're getting so much pushback is because of the first sentence of this section, where you refer to the previous discussion and say "there actually appears to be little to no opposition to developing opt-in features to help readers avoid images that they don't want to see", which is not at all the mood of that discussion. I saw one person saying that making it opt-in would sway them and a great many people saying that the very existence of such a system would be ripe for abuse. Also, this is the Idea Lab, it is for developing ideas, not staying fixed to the original proposal. Please stop bludgeoning the discussion by repeating your original proposal and allow people to develop a form of the concept that is more likely to have community support, such as blurring all images.135.180.197.73 (talk) 23:54, 17 December 2024 (UTC)

• I feel like this section is trying to give false legitimacy to a widely opposed idea by saying the longstanding consensus that “content warnings and censorship are bad” (and by extension the opinions of anyone supporting that position) is illegitimate because it’s not “constructive”. People have a right to not help you “construct” an idea that’s against policy and been rejected time and time again. If you don’t want negativity don’t make a controversial proposal. Dronebogus (talk) 15:40, 15 December 2024 (UTC)

  Nobody is asking you to help. Several of us have politely tried to get you to stop bludgeoning the discussion by stating your opposition over and over again. – Joe (talk) 08:04, 16 December 2024 (UTC)

  It's not happening here. You have been told where to go to copy the entire site and modify it to fit your ideas. --User:Khajidha (talk) (contributions) 13:07, 16 December 2024 (UTC)

  I find it curious how nobody ever calls opinions they support “bludgeoning”. Levivich and WhatamIdoing have contributed almost as much, and as repetitively, in agreement with you. I know idea lab is supposed to be all about open-mindedness and positivity but this is a perennial proposal that clearly violates WP:NOTCENSORED and WP:NPOV, two of the most fundamental policies of Misplaced Pages. You’re building something up that will inevitably get shot down if it actually made it to RFC. Dronebogus (talk) 00:00, 17 December 2024 (UTC)

Joe Roe, I remember reading somewhere on a wikimedia project (maybe it was Phabricator) thoughts about implementing a tool called OpenNSFW, which from my non-technical understanding, it's able to look at an image and label it as safe or NSFW. I don't know how accurate it is, whether it could be implemented on such a scale, etc, etc but I thought it might be relevant. JayCubby 00:41, 19 December 2024 (UTC)

OpenNSFW is not something I've heard of previously. A few minutes research and all I can tell you about it is that it categorises images as either "safe for work" or "not safe for work" the latter being images containing either "pornography" or "nudity" but nowhere I've found are those terms defined. I was not able to find any independent analysis of how accurate OpenNSFW is, but other machine learning algorithms that attempt the same task seem to have best-case results between 79% and 94% accuracy. I was not able to find any indication of detail about how accuracy was determined beyond "it's subjective" and one inaccurate result being an image of a clothed young woman sat on the ground leaning against a wall playing a guitar being classed as not safe for work by one model (that was not OpenNSFW), my guess is that this was due to low contrast between the guitar and the woman's skin tone. Even if OpenNSFW equals the 94% success rate of the best model tested, that still leaves 6% of images wrongly categorised. Even in extremely unlikely case the errors were all safe-for-work images wrongly categorised as not-safe-for-work, this requires the viewer to have the same (unknown) definitions of "pornography" and "nudity" as the model's developers and for those two categories to cover 100% of images they regard as not safe for work (e.g. they are happy to view images of violence, drug use, medical procedures, war, disease, death, etc). It is also worth noting that these models are described as "computationally expensive", so are unlikely scale well. Unless someone is able to show that this model performs very significantly better than the others reviewed (on all metrics), this is not practical for Wikimedia projects even if this sort of censorship was something we would entertain (which it is not). Thryduulf (talk) 01:51, 19 December 2024 (UTC)

Let's say, for the sake of argument, that OpenNSFW could correctly label 80% of images deemed to contain nudity (which is what I think it's mostly trained for). It probably doesn't make sense to scan all images on Commons, a good deal of categories could be excluded (like the literally millions of pictures from the ISS, or ethnographic toplessness). Other offensive subjects or categories (graphic violence, gas gangrene) could be blanket-included and resulting false positive excluded by hand (let's say experienced users could apply for a patrol-type right).

https://www.researchgate.net/publication/345162125_Classification_of_Not_Suitable_for_Work_Images_A_Deep_Learning_Approach_for_Arquivopt might be helpful, but it's too technical for me. JayCubby 02:32, 19 December 2024 (UTC)

Once again you are simply assuming that your definitions match other people's definitions. For example, many people who object to images of nudity do not distinguish between "ethnographic nudity" and other types, but many people do - who is right? Anything requiring human input (e.g. your "patrol-type right" suffers all the same problems that you are trying to solve by using machine learning in the first place (see extensive documentation of these problems in this discussion). Thryduulf (talk) 02:39, 19 December 2024 (UTC)

Misplaced Pages, at least the English version, is Western-leaning. In the West, there's some distinction between ethnographic and non-ethnographic toplessness and their perceived offensiveness, but I'm not trying to rigidly define offensive material, as a broad definition would be impossible. I don't want to censor everything possibly objectionable, only what readers of an encyclopedia really wouldn't expect to jump out at them. On the patrol bit, I'm saying there will be false positives and negatives, but likely a small enough number to be correctable manually. JayCubby 02:49, 19 December 2024 (UTC)

Misplaced Pages, at least the English version, is Western-leaning this is a bug. We attempt to avoid systematic biases like this because our goal is to create a neutral encyclopaedia, not a western-leaning encyclopaedia. In the West, there's some distinction between ethnographic and non-ethnographic toplessness and their perceived offensiveness while this is true for some western people in some western places, it is not true of all western people in all western places. For example the distinction would matter in a UK university geography lecture, it would not matter in a UK university maths lecture. , I'm saying there will be false positives and negatives, but likely a small enough number to be correctable manually. If you think that a 20% incorrect categorisation rate (or even 2%) would produce manageable numbers then you haven't appreciated how many images are on Commons. You have also ignored (again) all the problems that are not about numbers. Thryduulf (talk) 03:11, 19 December 2024 (UTC)

On the accuracy bit, the accuracy numbers appear to be for people alone based on the paper I found. This would be a silly thing to implement if it falsely flagged tens of millions of images, .

On the distinction bit, I'm saying people would be less offended by the images in Himba than Topfreedom.

On the numbers aspect, yes, there are 99,475,179 images on Commons, but by my very rough estimates the vast majority of those could be excluded without creating many false positives.

I could do an in-depth analysis of this, yes, but it's a big enough subject that the only effective way to approach it is through numbers. JayCubby 03:26, 19 December 2024 (UTC)

I'm saying people would be less offended by the images in Himba than Topfreedom. and I'm saying that while this is true for some people (group A) it is false for other people (group B). People from both groups will be using this filter. If you do censor ethnographic nudity then group A will rightly complain about being denied access to appropriate content (false positive), if you don't censor ethnographic nudity then group B will rightly complain about seeing inappropriate content (false negative). You cannot both censor and not censor the same image. Which group do you choose to side with? How are you explaining to the other group that their standards are wrong?

yes, there are 99,475,179 images on Commons, but by my very rough estimates the vast majority of those could be excluded without creating many false positives. even if you exclude 95% of images, that is still almost 5 million that you need to deal with by hand. If 95% of the 5% are automatically categorised correctly and you somehow don't need to check them, that still leaves about 250,000 images. All this assumes that there is no miscategorisation, no new images or categories, no renamed categories, and no instances of categories in your exclude/include sets being merged together (all but the last is provably false, the last is unknowable either way at this level of detail). Whose standards are the patrollers applying to the images they check? Why those standards? What happens if patrollers disagree?

the only effective way to approach it is through numbers. except considering only numbers is not effective, because the vast majority of the problems with this and similar proposals are nothing to do with numbers. Thryduulf (talk) 03:44, 19 December 2024 (UTC)

On the first, I think there should be a minimum of images that should be obscured. Maybe select ones on anatomy, I don't know.

On your second point, I'm not too sure of Commons' category structure, I'd like to see numerical distribution of images into different categories. JayCubby 03:49, 19 December 2024 (UTC)

The Commons category structure is so disorganised that the backlog for images lacking any categories is six years old. (Not a knock on Commons editors, it's just such an overwhelmingly huge yet entirely thankless task.) Any system with heavy reliance on those categories would be at the whims of this. CMD (talk) 04:29, 19 December 2024 (UTC)

The following is a (genuinely) brief overview of Commons categorisation with relevance to this discussion. Commons categories come in multiple types.

• Some categories are not relevant to image subject (e.g. user categories, copyright licenses, files by copyright license, project administration categories, etc).
• Meta-categories - i.e. ones that should contain only subcategories (e.g. Commons:Category:People with objects by material). Note that many of these incorrectly contain images that should be in subcategories.
  • All these categories (and their subcategories) should be sub-categories (at some level) of Commons:Category:Topics, but I don't know if they all are. I also don't know whether that category contains any non-content subcategories, nor whether there are root categories that should contain all and only non-content categories (my guess is that in practice there isn't).
• Mid-level categories that contain both images and sub-categories
• Bottom-level categories that contain only images.

Of those categories that contain image, some contain only a single image others contain thousands (although no category should contain this many, there is no exact threshold for when a category needs diffusion, no guarantee it will get diffused, and some categories need perpetual maintenance.

Many (most?) images are in multiple content categories, e.g. File:Cosplayer of Ellen Joe at ICOS04 (20241019153251).jpg is in Commons:Category:Cosplay of Ellen Joe (15 images), Commons:Category:Cosplayers with motorcycles (18 images), Commons:Category:High-heeled shoes in cosplay (575 images, 11 subcategories), Commons:Category:ICOS04 (31 images) and Commons:Category:Women with chains (3 images, 2 subcategories).

Some categories contain only images that unambiguously show nudity, some contain only images that unambiguously don't show nudity, others contain both of the above and images that are ambiguous (e.g. Commons:Category:Fantasy Fest 2012, is opaque body paint nudity? what about translucent body paint? nipple pasties?).

Subcategories can be surprising, e.g. you'd expect Commons:Category:Nude women standing to only contain photos of nude woman standing, but it also contains Commons:Category:SVG nude standing women, which contains Commons:Category:SVG nude standing women, which includes File:290 Venuso el Willendorf.svg. Is that pornographic? Nudity? If so is it ethnographic? Are your answers the same for File:Misplaced Pages 20 - AT Niederösterreich Venus.svg from the same category? How does that make you feel about the completely innocuous-sounding Commons:Category:Misplaced Pages 20 derivatives/svg which the second image is directly in.

All files should be categorised when uploaded, categories exist for media needing categorisation for each year since 2018, each one contains between 34,000 and 193,000 files. Commons:Category:Media needing categories requiring human attention has over 2,500 subcategories, each with several tens of images. Thryduulf (talk) 05:18, 19 December 2024 (UTC)

this is a bug. We attempt to avoid systematic biases like this

In some cases it's a bug. In other cases, it's just about being useful. enwiki is meant for English-speaking internet users. If we randomly rewrote 10% of each page in Chinese, that would be less "linguistically biased", but very annoying for the 99% of enwiki users who can't read Chinese. In the same way, a filter should try to match the preferences of the median English-speaking internet user (on a "prudishness" scale). We'll never do a perfect job of that, but we can definitely do better than implicitly bowing to the preferences of the most extreme 1% of users (who think all images should be treated as safe-for-work). – Closed Limelike Curves (talk) 03:30, 21 December 2024 (UTC)

a filter should try to match the preferences of the median English-speaking internet user (on a "prudishness" scale). 1. Why? 2. What is a "prudishness scale"? 3. How are you determining the median on it? 4. How are you assessing each image on the scale? Thryduulf (talk) 12:34, 21 December 2024 (UTC)

The median is “whatever I personally consider it to be”; it’s a generalization of something Ellen Willis once said: “In practice, attempts to sort out good erotica from bad porn inevitably comes down to 'What turns me on is erotic; what turns you on is pornographic.” Dronebogus (talk) 09:26, 22 December 2024 (UTC)

This is exactly the opposite of my point (see below). The median is whatever readers consider it to be, completely independent of my opinions. My opinion is that no image should be censored or blurred. If the tool I proposed below existed, I'd personally vote "0 years old" on every image (because I don't think anything should be censored). But that's my personal opinion, as an extremely culturally liberal/libertarian kind of person. It's not my place to impose that on the readers. – Closed Limelike Curves (talk) 19:21, 22 December 2024 (UTC)

“Whatever readers consider it to be” yeah good luck finding anything within 20 parsecs of a consensus from the collective readership of the largest website on the planet. Dronebogus (talk) 08:51, 26 December 2024 (UTC)

For #1, see median mechanism and an introduction to mechanism design/collective choice theory for an overview of desirable properties. In a sense, the median identifies the unique "consensus" position, because a majority of voters will oppose any other setting (a majority of voters will prefer the median to the alternative). For #2-4: a prudishness scale is a scale that measures prudishness. A simple example would be to ask every reader "at what age would you let your kids see this image?" For each image, we calculate the median to get that image's age rating. Users then get to select what age ratings they want to hide in their preferences.To clarify, this is a thought experiment; I'm not suggesting the WMF create an actual polling tool just for this. (Though I'd be very interested in it if we could use it for other things too, e.g. readers rating articles on their quality or neutrality.) Instead, my point is:

1. You can give a neutral definition for whether an image is appropriate or not, which has nothing to do with any editor's personal opinion; it's just a statement about readers' preferences. Every image already has an "age rating" (even if we haven't measured it), just like how every politician has an "approval rating" (even if we haven't polled anyone).
2. Having zero image filtering isn't some kind of magic "neutrality" that keeps us from having to make difficult choices—we're still making all of those decisions. We're just choosing to take the most extreme position possible on every image, by setting all of their ratings to "0 years old" (regardless of what readers think). That's a very opinionated decision—it's just as "neutral" as banning every image because someone might consider it inappropriate.

– Closed Limelike Curves (talk) 19:11, 22 December 2024 (UTC)

As you've now admitted you're just wasting everybody's time here with a thought experiment rather than an actual proposal, I shan't go into detail about all the ways you're comment is fundamentally wrong, but the most basic is that a majority of voters will prefer the median to the alternative is intended to apply to voting for a political candidate (which we are not doing here) and assumes a one-dimenional spectrum and, as the article states It is impossible to fully generalize the median voter theorem to spatial models in more than one dimension. What images to censor is almost fractally-dimensional - even if you take what appears to be a single dimension at first glance, say nudity, you quickly realise that you need to split that down further - the subject's age, gender, topless/bottomless/full nudity, pose, context (e.g. ethnographic or not), medium (e.g. painting, photograph, cartoon, sculpture, diagram, etc), prominence in the image, etc. all matter to at least some people, and they all vary differently. e.g. a sculpture of a topless elderly adult male hunched over is very different to an impressionist painting of a beach scene with a topless pre-pubescent girl in the background is very different to a medical photograph of a topless transgender 20-something man immediately post top surgery, etc. Thryduulf (talk) 18:53, 23 December 2024 (UTC)

@Closed Limelike Curves, the WMF already did that, though before your time; see WP:AFT5.

@Thryduulf, I believe this "impossible" thing is already being done at Common Sense Media, which appears to be a US website for telling parents whether the book-shaped object their kid is reading is age-appropriate and contains any of multiple specified taboo elements (e.g., tobacco, alcohol, kissing). If we really wanted to pursue something like this, we could look at how it's being done elsewhere. I would not be surprised to find that it is already happening in other places (just perhaps without the specific goal of masking images). WhatamIdoing (talk) 04:15, 26 December 2024 (UTC)

The fact that CSM gives ratings from their particular point of view does not mean they are succeeding at what Thryduff noted. They are an advocacy group with their own point of view of what is appropriate. -- Nat Gertler (talk) Nat Gertler (talk) 04:25, 26 December 2024 (UTC)

When I looked at it, it gave age ratings based on what their userbase said. Whether a book contains any references to tobacco is objective, so one would not expect to find differences of opinions about that. WhatamIdoing (talk) 06:02, 27 December 2024 (UTC)

I think there's a lot of subjectivity to what counts as a "reference to tobacco". If Sherlock is puffing on his Meerschaum pipe, certainly. If there's a Meerschaum on the mantlepiece, probably. If he's wearing a smoking jacket? If Watson tells him he looks smokin' in that jacket? If he mentions that Martin Luther King Jr worked a plantation in Simsbury, Connecticut?? How close to tobacco does the reference have to be in order to be a reference to tobacco? -- Nat Gertler (talk) 06:15, 27 December 2024 (UTC)

And yet we manage somehow to decide what belongs in Category:Tobacco, so presumably this would also be manageable. WhatamIdoing (talk) 06:18, 27 December 2024 (UTC)

Which articles belong in category Tobacco is determined by whether tobacco is a defining feature of the article subject and no more specific subcategory is more appropriate. If you cannot see how this is qualitatively and substantially different to determining whether an image contains a reference to tobacco then you do not have the competence required to usefully partake in this discussion. Thryduulf (talk) 22:13, 27 December 2024 (UTC)

We are not CSM, and we should not take a position on the propriety of imagery and information related to nudity, profanity, alcohol, and consumerism! This is an encyclopedia, not a morality police. Speaking of, this is also proven possible a project by Iran’s Morality Police, by the Indonesian Ministry of Communication and Digital Affairs, and by the Chinese Ministry of Public Security. It is indeed very possible to censor and deem certain information offensive. We are just not willing to do that. ꧁Zanahary꧂ 07:58, 26 December 2024 (UTC)

I feel like you have been consistently struggling with the gap between "identifying" and "censoring". We already put photos in c:Category:Pornography at Commons. Editors have figured out how to do that, and it does not involve "taking a position on the propriety" or becoming "morality police", nor does it involve "censoring and deeming certain information offensive". Putting some sort of #sexual-content hashtag on the image would not require a materially more complex process.

Again, I don't believe this will happen unless and until the WMF is forced to do so, but I think we should be realistic about the challenges. There are parts of this that are quite simple and very similar to what we're already doing, every single hour of the day. WhatamIdoing (talk) 06:11, 27 December 2024 (UTC)

Looking at the non-subcategorized photographs in that category.... most of them are not pornography. -- Nat Gertler (talk) 06:18, 27 December 2024 (UTC)

Sure. I wouldn't use that particular cat (or any of them) as a substitute for a purpose-built system. But we seem to figure out what's relevant for each type of category, so I believe that people could do the same kind of mental work in a different format, and even use the same kind of dispute resolution processes if editors had difficulty agreeing in any given case. This is not rocket science; this is not brain surgery. (It's also IMO not going to happen.) WhatamIdoing (talk) 06:21, 27 December 2024 (UTC)

It's also IMO not going to happen. Then why are you dragging out this discussion on an overwhelmingly opposed idea supporting an idea you know will almost certainly fail? When an idea I support doesn’t gain momentum I’ll throw out a few counter-arguments and tweaks and move on. Dronebogus (talk) 20:43, 28 December 2024 (UTC)

The selection of criteria to filter for is mired in POV. If the provided filters were for content related to Jews, images of people with green eyes, and images of unpierced ears, you’d probably scratch your head at the apparent fact that the designers of these filters thought that these categories of information were problematic and worth working around and validating sensitivities towards. ꧁Zanahary꧂ 17:29, 27 December 2024 (UTC)

the WMF already did that, though before your time; see WP:AFT5.
sigh—of course this tool already existed, then got killed off in an RfC by angry editors.I can at least partially agree with the spirit of the comments, which is that if people were just giving feedback along the lines of "How do you rate this article from 1-5?", that wouldn't be super useful (even if there's no downside either, and it's a huge improvement over our current system of article ratings).OTOH, A/B tests comparing two versions would probably be very useful for settling some kinds of disputes (especially those about article wording, e.g. what readers find more intuitive). – Closed Limelike Curves (talk) 05:17, 27 December 2024 (UTC)

Well, the reason it's AFT5 is because there were five versions, and none of them worked out very well. Ratings-only versions didn't provide actionable feedback. Free-form text let people spam garbage, and expose their personal/identifying information on talk pages. It caused a lot of extra work for WP:Oversighters.

The bigger problem was that the utility varied by subject area. The feedback on medical articles was pretty consistent: readers want images, and they want to know the prognosis. AFT5 comments are one of the sources for my oft-repeated line that if the Misplaced Pages article is written correctly, and you get a text message saying "We're at the hospital. They think the baby has scaryitis", then the article will tell you whether the correct response is "What a relief" or "I'm so sorry". The feedback on pop culture articles was also pretty consistent, but in a rather unfortunate way. The feedback there was largely "I loooove this band!" or "This show is overrated" (and the ratings were about the person's opinion of the topic, not about the Misplaced Pages article). WhatamIdoing (talk) 06:17, 27 December 2024 (UTC)

Makes sense. (Although I'm not sure why this created lots of work for OS—I was under the impression that people are allowed to disclose their personal information on WP themselves, if they want.) My complaint is mostly about killing this rather than trying to improve it. I can think of two quick major improvements—

1. Worst-case scenario, just go back to the "unactionable" 5-star ratings. That's already a big improvement on B/C/Start ratings as a metric of article quality (since it's not based entirely on how picky a reviewer you got updating the rating 12 years ago). Using an average rating cutoff could be a good first step in prioritizing or weeding out GANs.
2. Have some kind of "reviewer reputation", so feedback from people who left good comments gets sorted to the top and low-reputation comments are hidden. Bonus points if you let people upvote/downvote comments.

– Closed Limelike Curves (talk) 18:14, 27 December 2024 (UTC)

1. The problem with the 5-star ratings was that they were meaningless because the metric was undefined e.g some people ranked an article 5 stars for being fully referenced, copyedited, written in good quality prose, adequately illustrated and not missing information about whatever the reviewer was looking for, others would rank the same revision as 3 or even 2 stars. Some reserved 5 stars for articles than could not be improved. Others ranked the article based on how useful it was to them (a stub would rank 5 stars if it contained everything they were looking for, which might just be e.g. a birth date, a featured article might get 1 star if it didn't answer their specific question), yet another set of readers ranked the article based on how much they liked (or didn't like) the subject of the article.
2. This would not solve the problem of reviews containing spam or personal information, nor would it be possible to assign a reputation for readers who are not logged in.

Read the discussions about the article feedback trials, they were discontinued because nothing that was tried worked, and nothing else that was suggested was workable (and lots of things were suggested). Thryduulf (talk) 22:26, 27 December 2024 (UTC)

The problem with the 5-star ratings was that they were meaningless because the metric was undefined
The very vague nature of B/C/Start ratings by a single person is what makes them borderline-meaningless. The good news is if you average over enough ratings, that's fine—different definitions of each rating cancel out. (Especially if you do a basic adjustment for each rater's POLR intercepts, i.e. how "strict" they are when they're rating articles.) – Closed Limelike Curves (talk) 23:23, 27 December 2024 (UTC)

B/C/Start ratings have limited usefulness but are not meaningless: They are sort-of defined and measure a single thing (article quality at given point in time). Thryduulf (talk) 23:39, 27 December 2024 (UTC)

That's precisely correct, and also exactly how 5-star ratings work (sort-of defined, and measure article quality at a given point in time). The main difference is with a larger sample size (e.g. all readers, rather than the occasional editor), the usefulness of these ratings increases (since idiosyncrasies start to cancel out). – Closed Limelike Curves (talk) 01:02, 28 December 2024 (UTC)

For articles with lots of ratings the ratings did not produce any useful feedback and did not reliably correlate with article quality, because not everybody was rating article quality. Lots of articles did not get many ratings. It worked in theory, but it did not work in practice. Seriously, actually read the old discussions, it will save you and everybody else a boatload of time. Thryduulf (talk) 08:17, 28 December 2024 (UTC)

@Closed Limelike Curves, the problem was that most readers weren't rating article quality.

We'd given them an FA- or GA-quality article, and the responses would be "One star. I hated this movie." We'd give them a badly written, unsourced stub, and they responses would be "Five stars. This is the best movie ever."

A larger sample size does not solve this problem. You cannot cancel out individual raters' idiosyncrasies about quality when the raters aren't rating quality in the first place. WhatamIdoing (talk) 21:04, 28 December 2024 (UTC)

I'm not saying the implementation was great or didn't need major changes, just that the idea of soliciting feedback from readers was good, and the issues with the system can be fixed (filtering out "unhelpful reviewers" is a classic unsupervised learning task, or article ratings could be replaced with simple A/B tests to compare before/after an edit). Even if it wasn't, though, there's no harm in holding onto the ratings—if they're not helpful, just don't use them—or in keeping the interface on and limiting it to logged-in users. – Closed Limelike Curves (talk) 21:39, 28 December 2024 (UTC)

the idea of soliciting feedback from readers was good it was (and is) a good idea. However nothing that was tried worked in practice to produce feedback even close to useful enough to outweigh the costs of collecting it - except the one thing we currently still have: talk pages. Talk pages give us less feedback than the AFT, but a much greater proportion of it is useful feedback and a much lower proportion of it is spam, personal information, or just plain irrelevant. We tried fixing the system - not just once but five times - and you can be certain that if there was 'one simple trick' or anything like that then it has been tried and didn't actually solve the problems. If you had either actually read the links you've been given or actually listened to what other people have told you on multiple occasions you would know all this though. Thryduulf (talk) 22:49, 28 December 2024 (UTC)

Yes, I read through the links the first time they were provided. I don't see anything about switching to A/B testing by paragraph, or consulting with experts in statistics or ML to address problems in the data. (It turns out this is a very common problem on the internet—but despite this, every website except Misplaced Pages has managed to find a feasible solution by spending 5 minutes talking to a data scientist.) – Closed Limelike Curves (talk) 17:07, 3 January 2025 (UTC)

a filter should try to match the preferences of the median English-speaking internet user (on a "prudishness" scale). I actually do not understand how one can think this is the job of an encyclopedia. ꧁Zanahary꧂ 20:12, 22 December 2024 (UTC)

Because you can change the settings to let you see whatever you'd like? This is just my suggestion for how to choose a sensible default—default to whatever most people would pick anyway. – Closed Limelike Curves (talk) 22:00, 22 December 2024 (UTC)

You assume that most people want to block images in the first place.--User:Khajidha (talk) (contributions) 17:32, 23 December 2024 (UTC)

I explicitly do not. If a majority of people don't want to block any images for people of any age, the median age rating for all images would be 0 in the mechanism I described above. – Closed Limelike Curves (talk) 00:19, 24 December 2024 (UTC)

The default on an encyclopedia is the revelation of pertinent information. ꧁Zanahary꧂ 18:05, 23 December 2024 (UTC)

Though there is a point at which too much information, to the point of irrelevancy, can be given. We, I fear, are approaching that point with our use of images at times. JayCubby 18:52, 23 December 2024 (UTC)

What you are saying is that some images are WP:UNDUE, which is completely separate from anything being discussed here. Thryduulf (talk) 18:54, 23 December 2024 (UTC)

That is completely unrelated to the concealment of sensitive images, and is instead pertinent to, as @Thryduulf has said, WP:UNDUEness. ꧁Zanahary꧂ 19:14, 23 December 2024 (UTC)

There's also MOS:IMAGERELEVANCE. Some1 (talk) 15:52, 28 December 2024 (UTC)