Talk:Image filter referendum/en/Categories

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search


5-10 categories? A few dozen?[edit]

Who decides what the 5-10 objectionable categories are?[edit]

From the penultimate question on the survey, it looks like 5-10 filtering categories are being considered. I was thinking that as a practical matter, it might be difficult for many thousands of editors across different cultures to decide what those categories were, so I assumed someone else might be deciding, i.e. the WMF. Phoebe's statement implies my assumption might be wrong. If 5-10 categories are to be used, who will (and how will they) decide what those categories are?--Wikimedes 07:04, 21 August 2011 (UTC)

Actually, the research shows that providing three broad groups would make nearly everyone around the world happy (except, naturally, the small but vocal "anti-censorship" people who think it the inalienable right of every vandal to display any image he wants on my computer screen):
  • Certain religious images (e.g., en:Mormon temple garments and paintings of Mohammed)
  • Sexual images (definitely including porn, but possibly going as far as general nudity)
  • Violent images (e.g., mutilated bodies and abused animals)
I don't know, but I'd guess that five to ten were proposed because it would be useful to introduce some granularity. For example, I could imagine that a person might want to avoid stills from pornographic movies, but wouldn't mind the naked woman in en:Pregnancy. (It may amuse you to know that the "anti-censorship" crowd has successfully prevented the inclusion of even one image of a fully dressed pregnant woman in that article: all images show light-skinned women either completely naked or in their underwear.) WhatamIdoing 20:09, 23 August 2011 (UTC)
I am arachnophobic, and I get panic attacks from looking at spider images. A friend of mine suffers from BPD and he gets anxiety attacks from looking at images of tall structures. Et cetera. "Argument": destroyed. Contrary to the "censorship" strawman people like you keep raising (you guys are almost like an anti-anti-censorship false flag operation), there are many different overwhelmingly powerful arguments against a project-wide filter category system, not a single one of which has got anything whatsoever to do with censorship. At all.
Also, you are destroying your own "argument" without even realizing it: If there are actual, encyclopedic reasons that speak against the set of images included e.g. in en:Pregnancy, then we should solve that problem at its root (ie. go to the article and improve it), not throw a completely superficial "solution" like an image filter at it. That said, an image filter has my tolerance, if not enthusiastic support -- but only if it is implemented as a general image filter, without any project-wide filter categories. -- 20:54, 23 August 2011 (UTC)

As I noted at some point on the mediawiki discussion page: I'm also in favor of any implementation being a general image filter. I haven't seen any other option proposed that sounded like it would work out socially or philosophically. SJ talk | translate   01:23, 6 September 2011 (UTC)

We need this but it should be made sure to be flexible.[edit]

I agree with one of the above users, but only partially, that this could become imposed censorship, but if implemented correctly it won't be. In fact the point of user controls is to AVOID global censoring. Just as one must make sure a governing body doesn't impose censors, it should also be made sure that they don't impose no possibility for the individual to censor the content on their own computer. One can always avoid certain pages even now, but if one wants to read a topic that may contain offensive images, they should not be forced to view those images if they don't want to, any more than they should be forced to not see them if they do want to see them.

For more flexibility the 2 proposed categories ("nudity/sex" and "violence") should become 3 categories ("nudity", "sex", and "violence). Otherwise an image of a XXX sex act might be censored just as much or little as an image of the nude painting Birth Of Venus.

A user's personal control panel should contain not only an "on/off" switch for each category, but also a 0-10 slider under each category for showing or not showing an images of a certain "strength" of that category. It should also allow the image to be delayed showing or blocked entirely based on a check box (not forced to be delay method only as suggested in the official statement). Specific images that you believe should or should not be shown based on your image category sliders (but had been shown or not shown opposite what you wanted) should be allowed to be added to a user's blacklist or whitelist manually by the user on an image by image basis. Of course all the above settings should be reversible by the user so that no change to the settings made by a user is permanent for that user.

This is very important it be done though as NO web-client software package (i.e. Firefox) is capable of blocking some images but not others, and blocking images but not text. Instead of such software will just block WHOLE SITES that have "bad content" in general above a certain user set level (no distinction of pictures versus text, or other such surgical precision). Instead pages with such offending material will simply be completely blocked from being browsed to by the user. This browser based "complete censorship" is a FAR STRONGER method than that being proposed on Wikimedia, and makes "collateral damage" by blocking completely lots of web pages even with only a small amount of offensive content. Until browse censors allow finer control of the censoring, then Wikimedia is just doing what it has to. And keep in mind this is USER CONTROLLED! It is NOT "we wiki get to censor what you see cause we're evil big brother ha ha ha ha ha".

To the above user who is worried about institutions policy, my only reply is this. Follow the rules of that institution when at that institution. If you want to be able to brows freely with no censoring, then do it on your own computer at home or at an institution or other place with internet access who's rules don't restrict content accessed on campus. I'm guessing those institutions already have a policy in place preventing offending pages from even being accessed. At least with this in place they could block the images but not the whole page. There are MANY places with their own rules. If you go to a fastfood joint with a sign that says "no pets" are you going to bring your pet with you just because you don't like the rule? No, because then you won't get served food. You are correct that the US government is bound by the constitution. However schools and other privately run companies and institutions have the right to impose their own restrictions and rules. And you will need to follow them if you don't want to get in trouble. This applies for a private school/college, restaurant, grocery store, etc.

Animedude5555 07:14, 20 August 2011 (UTC)

About the institutions, overall I agree that public computer labs have a right to control what is seen. As my point above still stands, though, this greatly hurts a massive demographic. Most Wikipedia editors have personal computers, are well-educated, etc., and so have this false view that everybody owns a computer and lives in some sort of metropolis with a thousand internet cafes. But we can't ignore the demographic with only access to one public computer, especially in conservative or religious areas where administrators are likely to enforce filters. Many of the pictures we're thinking of allowing to be filtered have crucial educational content. And I know so many people, some street kids, some homeless, some deadheads, some just minimum-wage families without a computer, that if the local school or library enforced a policy would have no alternative Wikipedia access. I can only imagine how many people are in this situation in extremely poor, rural areas.
To protect this demographic at large, I don't think we should even consider giving institutions this tool in the first place Rodgerrodger415 16:36, 20 August 2011 (UTC)
You do not seem to understand the proposal. This system lets any reader, with a single click, see any image the reader wants to see. This system will not work for schools, libraries, or parents. Think about it: the school says "Hide these images from the children". The student says "Ha ha, I click here to see the image anyway!". There's nothing the school's computer systems can do to prevent the individual user from overriding the settings on this system. If you really want to censor your users' images, you need to get a en:web proxy (or similar, paid systems, like "CyberPatrol") to do this. This system simply will not work for that purpose. WhatamIdoing 20:21, 23 August 2011 (UTC)
1. As discussed above, "nudity," "sex" and "violence" are merely three of countless types of image content regarded as offensive by some individuals in some cultures.
2. Also discussed above is the inevitable difficulty of determining whether to include a given image in a particular filter category (and the likelihood of controversy, argumentation and edit wars). If I understand correctly, you propose that this be further complicated by requiring that each image's "strength" (whatever that means) in a given category be ranked on a numerical scale. How is that remotely feasible? Via what process do you envision this occurring? —David Levy 07:32, 20 August 2011 (UTC)
You missed the part where I said the individule could force a image blocked under a certain user's setting to be shown (via a user account specific or cookie specific whitelist) if that user wanted a specific image to be shown that they believe should not have been blocked. Likewise if I find an image that I believe should be blocked under my current settings but isn't, then I could "blacklist" it.
Keep in mind this policy gives the USER (not Wiki) control of the censoring. It is NOT a violation of speech, and given certain images on certain wikis I may well choose to NOT BROWSE TO A CERTAIN WIKI ENTIRELY if this new feature is not put in place. By forcing me to NOT go to a certain wiki (based on my values) they would be in effect censoring THAT ENTIRE WIKI for me. Now THAT would be the unfair thing to do. What they are proposing gives the user MORE POWER, rather than taking away power as you claim. It does NOT infringe on their right to free speech in ANY WAY!
As for the implementation, it would be simple. It should use the world's best shape/image recognition software implementation ever invented. Not sure what the name of the software is, but it would be expensive, VERY expensive. But in the end it would be worth it to allow for a very good implementation of my suggestion. It would (for nudity) auto detect the presence of certain body parts, how much of the different parts is shown, the pose used (does it look like a "slutty" pose like in a playboy magazine, or just a "plain nude" pose like nude art in an art museum), the overall realism of the image (from "flat" images like a cartoon, to realistic 3D computer graphic renderings, to actual photos), and also in general what fraction of the skin is exposed (bikini coverage, versus fully clothed, etc). All these factors would contribute to the final computer determined "strength" rating in nudity category.
Of course violent images and images with sexual content would also have similar things looked at, and ratings calculated automatically by computer based shape/image recognition technology.
And don't think my suggestion is science fiction. There already exists forensic software that the police can use to detect child porn on a computer that indeed depends on implementation of image/shape recognition algorithms to determine in an image the age of the person depicted and also what amount of nudity and/or sexuality is present in the image. This allows the images the cop is searching for to be narrowed down, and he/she can then manually inspect these images to make sure that no "false alarms" are marked as illegal images.

Animedude5555 07:57, 20 August 2011 (UTC)

You missed the part where I said the individule could force a image blocked under a certain user's setting to be shown (via a user account specific or cookie specific whitelist) if that user wanted a specific image to be shown that they believe should not have been blocked. Likewise if I find an image that I believe should be blocked under my current settings but isn't, then I could "blacklist" it.
I "missed" nothing. I didn't comment on that portion of your proposal because it's irrelevant to my main concerns. I don't know why you've reiterated it, as it doesn't address anything that I wrote.
Keep in mind this policy gives the USER (not Wiki) control of the censoring.
...within a predetermined set of filters.
It is NOT a violation of speech, and given certain images on certain wikis I may well choose to NOT BROWSE TO A CERTAIN WIKI ENTIRELY if this new feature is not put in place.
That's your prerogative. Under the above proposal, persons objecting to image content other than "nudity," "sex" and "violence" face the same dilemma.
By forcing me to NOT go to a certain wiki (based on my values) they would be in effect censoring THAT ENTIRE WIKI for me. Now THAT would be the unfair thing to do.
Note the phrase "my values." Why do you believe that your values take precedence over those of other individuals/cultures?
What they are proposing gives the user MORE POWER, rather than taking away power as you claim. It does NOT infringe on their right to free speech in ANY WAY!
I don't recall claiming any such thing.
As for the implementation, it would be simple. It should use the world's best shape/image recognition software implementation ever invented. Not sure what the name of the software is, but it would be expensive, VERY expensive. But in the end it would be worth it to allow for a very good implementation of my suggestion. It would (for nudity) auto detect the presence of certain body parts, how much of the different parts is shown, the pose used (does it look like a "slutty" pose like in a playboy magazine, or just a "plain nude" pose like nude art in an art museum), the overall realism of the image (from "flat" images like a cartoon, to realistic 3D computer graphic renderings, to actual photos), and also in general what fraction of the skin is exposed (bikini coverage, versus fully clothed, etc). All these factors would contribute to the final computer determined "strength" rating in nudity category.
Of course violent images and images with sexual content would also have similar things looked at, and ratings calculated automatically by computer based shape/image recognition technology.
Okay then. Thanks for sharing your thoughts on the matter. —David Levy 08:19, 20 August 2011 (UTC)

Objective/neutral filtering is easy : just pick a few dozen keywords[edit]

First off, I'm rather appalled at the "Wikipedia (and I) know(s) better than you what images you should see" attitude. Also, "the government will start censoring Wikipedia if we add filters" is an absurd slippery-slope arguement.

More importantly, it's extremely easy to have an objective filtering system. You just need to use objective keyword system. For example, you tag all images that have nipples in them. The reader can decide whether they will accept images with nipples in them. The presence of an exposed nipple is not subjective. No one has to determine how sexually explicit the image is on some sort of scale. There is a nipple in it, end of story.

This would easily cover the vast majority of content that could be considered offensive in any culture. Think of the keywords you can easily and objectively add to an image:

  • Nipples, genitals, penis, vagina, anus
  • Intercourse/penetration, sexual stimulation (or masturbation and variations thereof), oral sex/stimulation
  • Blood, exposed organs, disease, surgery
  • Suicide, deceased person (or corpse, cadaver)
  • Carnivorism, skinning
  • Religious imagery (or, Catholic imagery, Hebrew imagery, etc, when the particular religion is not questionable)
  • Muhammad, Jesus, God, etc
  • Weapon, firearm
  • Hijab

Keywords that could potentially be ambiguous could be defined or not used. There could also be potentially combination filters, for example, do not display an image tagged "female" unless it is also tagged "hijab".

I think this is a reasonable approach to the idea. 13:35, 21 August 2011 (UTC)

Why do we need an "objective" filtering system? People get to chose to use it or not. If they disagree with it then they don't need to use it. Ottava Rima (talk) 13:39, 21 August 2011 (UTC)
To satisfy concerns a filtering system will be non-neutral in any way, including cultural bias, and prevent the need to come up with subjective image categories. 13:44, 21 August 2011 (UTC)
A filter isn't supposed to be neutral. It is stupid for a filter to be neutral. A filter is supposed to filter what the -user- wants, which makes it 100% subjective and non-neutral. Otherwise, it isn't a filter. Ottava Rima (talk) 16:11, 21 August 2011 (UTC)
No matter how many times people explain the distinction between "the reader deciding which filter(s) to use" and "the project deciding what filter categories are available and what images are placed in said categories," you continue to conflate the two. It's rather frustrating.
The concern isn't that a reader's personal beliefs are non-neutral. That's a given. The concern is that the filter options presented to the reader will be non-neutral (i.e. only some "potentially objectionable" images will be assigned categories, with subjective determinations of the files contained therein).
For example, if we have a filter category for sexual content, there will be significant disagreement regarding what images are "sexual" in nature. Meanwhile, if we don't have filter categories for unveiled women or homosexuals, this will constitute a formal declaration that persons/cultures deeming those subjects "objectionable" are wrong to do so. (And if we do have filter categories for those subjects, the opposite value judgement will be inferred.)
The only neutral approach, short of providing filter categories for literally everything (which obviously isn't feasible), is to provide filter categories for nothing. (This, of course, doesn't bar the introduction of a system providing the native capability to block all images and manually display them on a case-by-case basis.) —David Levy 17:05, 21 August 2011 (UTC)
After several pages of well-argumented debate, where an overwhelming majority of veteran wikipedia editors vent their concerns about the matter, dismissing the argument as "absurd slippery-slope" seems rather asinine to me. As for your proposal, although slightly less dangerous, it shares with the original one a mix of good intentions, leniency towards moralism and practical dangers. To begin with, governments could filter based on legitimate tags for oppressive reasons: governments in North Africa, for example, could censor all non-muslim religious images just for the sake of bothering religious minorities, which they already systematically do. On the other hand, images having nothing to do with, say, genitals, could be tagged as such if they are found to be inconvenient. And yes, wikipedia is an encyclopedia, so it does know a lot of things better than you (as in the general you), probaly including, but not limited to, which images you should see. Complainer
Actually, the majority of users are for the filter. The "majority" you are speaking of on this page (which is barely over 50%) are mostly IPs from those logging out and using multiple IPs to make it seem like far more people. If we implemented a bot that CU checked everyone and blocked those who log out edit, a lot of users would be blocked from their actions on this page alone. Ottava Rima (talk) 16:11, 21 August 2011 (UTC)
Actually, most users are opposed to this filter, especially outside the US. (If you don't need proof for your claims, neither do I. Helps the quality of the debate, doesn't it?) Kusma 16:23, 21 August 2011 (UTC)
Most users? No. Not even close. Most users in the world are actually against porn. Stop trying to project a tiny, vocal minority that lives in the Netherlands or Germany upon the rest of the world. China and India are 1/3rd of the world and their culture is very anti-pornography as a whole. As is the Muslim world, which is another 1 billion. That is half the world that is completely opposite of a tiny few, and we haven't even discussed the Americas in that number. Ottava Rima (talk) 16:30, 21 August 2011 (UTC)
How do you know that the majority of users is for the filter? Adornix 16:25, 21 August 2011 (UTC)
Because you can take the talk page, which pornography related talk pages are always heavily libertine even though the raw votes never suggest even close to that. Then you can take out the duplicate IPs and the logged out editing and realize that those in for this still represent a vast majority. With the bias acknowledged, it is probably about a 80% for the filter. After all, this is about giving people choice, not tyranny of a few people in the Netherlands or Germany that wish to force their views on sexuality upon the rest of the world that wants moderation. Ottava Rima (talk) 16:30, 21 August 2011 (UTC)
Even if all IPs are the same person, your count seems completely off. Kusma 16:45, 21 August 2011 (UTC)
Maybe you need to do a recount. At least two people admitted to "accidentally" editing logged off. IPs have no way of really knowing about this except to have a user account. It is a majority for the proposal, and most polls that aren't done by having canvass shoutings on talk pages have always shown a vast majority against the porn. Only about 2% of the world's population lives in such a libertine culture. Ottava Rima (talk) 16:59, 21 August 2011 (UTC)
The Wikimedia Foundation's fundamental principles cannot be overridden by a majority vote.
It's reasonable, of course, to opine that this idea complies with said principles. But if it doesn't, no amount of support can change that. —David Levy 17:05, 21 August 2011 (UTC)
The primary principle of Wikimedia is accessibility to an encyclopedia. Encyclopedias do not require image. By denying a large portion of the world access to an encyclopedia that those like -myself- wrote because -you- demand that they have no ability to make it so pornography or other problematic images don't appear for them is one of the most egotistical things ever, and it has no purpose on this site. The WMF is acting completely based on what ethics and its principles demand. You are being incivil by attacking it for doing so. Ottava Rima (talk) 18:58, 21 August 2011 (UTC)
I tell you for the third time, OR, STOP calling people names. Like it or not, this is NOT how you conduct a discussion on wikipedia. Anyhow, I think there is some confusion here. Nobody here is arguing that people should not have access to wikipedia. We are arguing that not looking at pictures they consider indecent is their responsibility and neither wikipedias nor their teachers or whoever happens to be in control of their firewalls. You provide the perfect example of the concept, being obviously opposed to much of our image bank and still an active contributor. If it didn't bother you, why should it a billion Chinese? Complainer
The Wikimedia Foundation hosts several projects other than Wikipedia, so let's extend "an encyclopedia" to include all of them. You're correct that our written content can be useful (and preferable in some cases) without the display of images. (We haven't even touched on bandwidth issues or visual impairments.) That's why I support the introduction of an optional image filter, provided that it covers all images (and enables their manual display on a case-by-case basis). Problem solved. No more "pornography or other problematic images" appearing against the reader's wishes.
I merely object to the establishment of a system in which images are singled out and formally deemed "potentially objectionable," which cannot be accomplished in a neutral or non-contentious manner, let alone an efficient or reliable one.
You accuse me of incivility for expressing disagreement with the Board of Trustees (on a page set up to solicit feedback regarding the idea) while simultaneously describing my position as "egotistical" (one of many such insults that you've hurled). There's no need for that. —David Levy 19:29, 21 August 2011 (UTC)
"STOP calling people names" - I haven't called anyone one name. However, you are making up false statements about what others say, which is a major breach of civility. You really need to stop. You have already made it clear that you don't have respect for cultures different from you and that you want to ensure that hundreds of millions of people can never have access to Wikipedia because you don't want to give them a choice. It is highly incivil, selfish, and not appropriate. It is also racist because you are intolerant of other cultures that are not like you. You wish to deny them a simple choice because they are different. That is not what Wikipedia is for. Ottava Rima (talk) 20:11, 21 August 2011 (UTC)
1. You appear to have mistaken Complainer's message for part of mine.
2. As noted elsewhere on this page, you've somehow misinterpreted my messages to mean exactly the opposite of my actual sentiment. Among other reasons, I oppose the idea because I believe that it would discriminate against members of various cultures by affirming some beliefs regarding what's "potentially objectionable" and denying others. You needn't agree, but attributing my stance to intolerance and racism is uncalled-for. —David Levy 20:42, 21 August 2011 (UTC)
1. I didn't mistake it. I assumed that you had an extra colon than what you did and it formatted behind yours. 2. And don't dissemble about your claims. It is impossible for anyone to argue that a few people will be discriminated against when not doing anything continues the discrimination against hundreds of millions of people. You wish to impose your fringe view on the world to deny people a choice not to look at porn while using the encyclopedia. In essence, you deny children, those at work, and the billions of people in places like the Middle East or China who are not legally allowed to look at porn from having access to the encyclopedia. That isn't appropriate no matter how much you try to claim you are protecting people. Ottava Rima (talk) 21:25, 21 August 2011 (UTC)
1. You keep referring to "Wikipedia" and "the encyclopedia." Again, this pertains to every Wikimedia Foundation project.
2. As I've noted repeatedly, I support the introduction of an optional image filter, provided that it covers all images (and enables their manual display on a case-by-case basis). Problem solved. No display of "porn" against readers' wishes.
I merely object to the establishment of a system in which cultural beliefs (whether yours, mine or someone else's) of what constitutes "objectionable images" are formally codified, thereby deeming them correct and everyone else's incorrect. In addition to opposing this on philosophical grounds, I believe that such a method would be highly contentious, inefficient, resource-draining and unreliable.
But again, I'm not asking you to agree with me. If you feel that I'm wrong, that's fine. But please drop the allegations of intolerance and racism. There's no need for that. —David Levy 22:32, 21 August 2011 (UTC)

porn -- An encyclopedically valuable depiction of a nude body or bodypart is "porn" in your worldview, Ottava Rima? "Porn" as in "portrayal of explicit sexual subject matter for the purposes of sexual arousal and erotic satisfaction"? Seriously: Is e.g. commons:file:Human vulva with visible vaginal opening.jpg "porn" to you? This is important because if it is, then you should understand that the problem is all yours. -- 21:40, 21 August 2011 (UTC)

You completely missed the entire point of this section. It is unnecessary to define porn if you use appropriate objective keywords. If he doesn't want to see a vagina, he can block images tagged vagina, and nobody has to care what his definition of pornography is. 04:41, 22 August 2011 (UTC)
I was replying to Ottava Rima. -- 09:07, 22 August 2011 (UTC)

"so it does know a lot of things better than you (as in the general you), probaly including, but not limited to, which images you should see" - this is arrogant idiocy. Anyway, arguing about how many people are on each side is incredibly childish and pointless. The final vote will be the final vote. You shouldn't try to sway people's opinions by telling them they are in the minority. Each individual should vote for him or herself - to try to tell them otherwise is authoritarian, elitist behavior which is absolutely incompatible with the views you are supposedly attempting to uphold.
"Lots of people are concerned about a slippery slope" doesn't necessarily make it a less fallacious argument. People are herd animals and tend to panic in large numbers. Why would governments censor images based on keywords for images but not based on keywords in articles? It would be easy to block every page with the words "Jew", "Hebrew", etc in them for example. On the other hand, they might uncensor entire articles when an "offensive" image is removed. Why is this any less likely than "Africa will censor all non-Muslim images to religiously oppress the people"? Why wouldn't Africa block Wikipedia entirely? 04:39, 22 August 2011 (UTC)
I support's idea. The categories should have strict definitions easy to measure. So instead of a vague "partial nude", we could say "underbody clothes (shorts, skirt, etc) shorter than X cm/in below crotch, cleavage below X cm/in from a certain point, visible belly". We could discuss about how many cm, of course, but this way each picture will be easy to judge and discussion would restrict to the definitions themselves. -- 01:06, 24 August 2011 (UTC)
I assume this is sarcasm, but it's a long leap of logic to assume objective filtering means guessing measurements. 02:43, 30 August 2011 (UTC)

Working definition of "controversial" + categorization scheme[edit]

I've been looking more closely at the 2010 Wikimedia Study of Controversial Content. The Study notes, among other things, that Wikimedia projects already have working definitions for "controversial" text or topics. As an example, the Study cites the English Wikipedia's list of controversial issues. The Study suggests that a similar, though of course different, working definition of "controversial" could be achieved for images.

I have two questions:

1. Does this mean that the Foundation is planning to have one single catch-all category of "controversial images" which users could choose to filter out? I'm confused on this point. From what has been said on the main page for this referendum, I was under the impression that the plan was to categorize images as belonging to various categories (e.g. nudity, violence, drug use) and allow users to filter out specific categories. Is the plan to have both a single catch-all category of "controversial images" and various sub-categories (e.g. nudity, violence, drug use), with the option of choosing which sub-categories to filter out? Please clarify.

2. The English Wikipedia defines a controversial issue as "one where its related articles are constantly being re-edited in a circular manner, or is otherwise the focus of edit warring". The Study of Controversial Content cited this in a footnote as an example of an objective, verifiable criterion for controversial content. In that case, is the standard for controversial images envisioned as being based on the number of complaints or edit wars associated with an image? Or is there no definite idea yet as to how images will be categorized as "controversial"?

Note: I'm looking for some clarification here, and ideally some discussion on how best to establish a categorization system for images, should the image-filtering plan proceed. Please don't comment here if you're just going to argue for or against the filtering plan. --Phatius McBluff 16:24, 18 August 2011 (UTC)

I would be also interested to see this questions answered. But I have to add that I'm strongly against the introduction of such a categorization, regardless in which way it will be done. --Niabot 16:34, 18 August 2011 (UTC)
Hello Phatius McBluff, and thank you for reading through all of the information in such detail! It's good to know that people are actually reading the stuff. :-) The controversial content study is just the background for the referendum and part of the inspiration. Robert Harris was contracted by the Foundation to look into the issue of controversial content on Wikimedia projects, and prepared that report. After that, the Board discussed it and felt it agreed with some and disagreed with other points raised, etc. (Phoebe's comment above in #discussion could help more, since she's on the Board and I'm not.) Then the image filter referendum proposal came up, which is what we're discussing now.
The current proposal is just what you see on the Image filter referendum page and its subpages; the controversial content study is just background information. The current proposal does not use a single catch-all "controversial" category, but instead has various subcategories that people can choose from. (Who are we to say what is "controversial"? Things like "nudity", "violence", etc. are a lot more straight-forward and neutral. Granular settings are good. :-)) The idea is also that it would fit in with our existing category system on Commons and the various projects, rather than by creating an entire new set of categories. Cbrown1023 talk 16:39, 18 August 2011 (UTC)
The current proposal does not use a single catch-all "controversial" category, but instead has various subcategories that people can choose from. (Who are we to say what is "controversial"? Things like "nudity", "violence", etc. are a lot more straight-forward and neutral. Granular settings are good. :-)) The idea is also that it would fit in with our existing category system on Commons and the various projects, rather than by creating an entire new set of categories.
I'm glad to hear this, if only because it confirms that I was correct in my initial impression of what the plan was. However, just to clarify, by "our existing category system on Commons and the various projects", do you mean the system of (say) putting this image into Commons:Category:Shamanism in Russia? In other words, is the plan to allow users to block images from whatever image category they please? I'm puzzled on this point, because the main page for this referendum mentions a possibility of only "5–10 [filterable] categories". --Phatius McBluff 17:04, 18 August 2011 (UTC)
Phatius, it's not just you being confused; there's been a lot of discussion around this, and nothing is etched in concrete. I think in a perfect world being able to filter any image category you chose would be nice; but there are pretty serious usability and technical concerns around doing that. So the proposal is to filter 5-10 categories. -- phoebe | talk 18:16, 18 August 2011 (UTC)
Okay, I think I now understand where things currently stand. Cbrown and phoebe, thanks for your prompt responses! I do hope that we get to have more input from ordinary users on exactly how the category/tagging system will work. Most of the feasibility (and at least some of the philosophical) concerns surrounding this proposal will probably hinge on that. --Phatius McBluff 19:49, 18 August 2011 (UTC)
I too took a closer look at the report. One salient point was that the recommendations were addressed to the community, and explicitly not to the board. Rich Farmbrough 17:09 18 August 2011 (GMT).
A companion report was delivered to the Board; the body was unchanged but it included an introduction directed explicitly to the Board. SJ talk | translate   01:23, 6 September 2011 (UTC)
"Things like "nudity", "violence", etc." If the purposes of this project are to allow people to filter out what they do not want to see, that's a pretty big etc. How do we select the 5-10 top objectionable categories, as measured by audience aversion? I'm very leery about a filter that could potentially invite users to filter out:
  • Nudity
  • Sexuality
  • Violence
  • Gore & Dead Bodies
  • Blasphemous Images (Christianity)
  • Blasphemous Images (Islam)
  • Blasphemous Images (Hinduism)
  • Sexist images
  • Racist Images
  • Depictions of Homosexuality
--Trystan 19:35, 21 August 2011 (UTC)
I'm also very worried about whether we can objectively assign images to categories. I can think of one objective measure: "Has been objected to by at least one known human". That category, over time, would approach filtering all images, but it would be completely fair. Additionally, we can fairly use mathematical criteria like "frequency of objection" to classify images.
5-10 categories would be very hard to sustain. People who object to classes of content will legitimately want equal footing for their personal objections, and we'll be hard pressed to find any fair way to assign categories or justify any particular assignment. Most of all, when someone is offended by an image not already categorized, what can we tell them? "Your offense isn't a valuable as other types of offense"? 5-10 categories will tend to slide towards unlimited categories, and I think that will work. --AlecMeta 00:20, 23 August 2011 (UTC)
It will not necessarily extend very much, as there are usability limits. But we will honour only some of the requests. We cannot rely on the number of complaints as anybody can create enough accounts to object many more times to a single image than the realistic number of real complaints. That means it is up to subjective decisions by editors who have an interest in editing the filters or related categories (depending on how it is set up).
For the encyclopaedic articles we can rely on editors wanting to improve the quality of articles on noteworthy subjects (and POV editors being in minority). Editors will get their reward by seeing those quality articles (and the quality Wikipedia). I am not convinced the same will work with the filter categories. The work may indeed be dominated by POV pushers - or take energy from the poor souls already fighting with backlogs and uncategorised media on Commons. I have seen no suggestion on how this will be solved or any reasoning about how this will not be so.
--LPfi 14:02, 23 August 2011 (UTC)

What will happen? All users who are offended by the filter will switch it of. And all the censors will put over time tousends of images into one or the other offending category. They will visit controversal articles in search for images to ban. And finally they will put classical works of art into such categories. And then the editors of value content are gone. Hopefully somebody has made a fork under a liberal jurisdiction by then. --Bahnmoeller 18:56, 23 August 2011 (UTC)

What categories are admissible?[edit]

The elephant in the room here is the incredible amount of handwaving that goes on when it comes to discussion of the categories that will be available to filter. We're to understand that they will include graphic sexual images, possibly nauseating medical images, ETC. Well the ETC is a big deal. If John Q. Antisemite and friends want the facility to hide images of Jews on the grounds that Jews have demon eyes that corrupt and defile all who meet their gaze, are we going to tell them that's unacceptable and we won't make such a category available? If so, what are the criteria for determining the admissibility of a filter category? This is crucial information that needs to be spelled out in advance rather than left until after this is a fait accompli. It's no good saying we will leave it to readers to determine which categories they want. I strongly suspect some will be more welcome than others, and this will have a lot to do with the particular cultural biases of the people overseeing this new tool. It will not be long at all before someone proposes a category that is judged unacceptable. What do we do then? How do we decide whose aversions to cater to? Beorhtwulf 15:53, 23 August 2011 (UTC)

You're making an excellent point! Huge and inevitable drawbacks like that are the reason why I believe only a general image filter (possibly with a per-user blacklist/whitelist system) could ever work. Other examples: A friend of mine has BPD and cannot look at images of tall structures without getting an anxiety attack. Do we create a filter category for people like him? A female friend of mine was raped by a black guy and she suffers from PTSD and cannot look at images of black guys without panicking. Do we create a filter category for people like her? Myself, I'm arachnophobic, but I have no problem with images of insects and do not want them thrown into the same category with spider images. Et cetera, et cetera, et cetera. -- 20:02, 23 August 2011 (UTC)
The reason that "etc" gets thrown about is that the system is not fully designed. The WMF has a reasonable point in making sure that this is generally supported before they invest more staff time in this.
I'm reminded of an old Doonesbury comic strip. A political candidate proposes an unpopular (but possibly necessary) economic policy and says that it is likely to result in some people losing their jobs. The media say, "But can't you be more specific?" He responds, "Yes, in this town, this particular factory will fire exactly this many workers". And the media say, "But can't you be more specific?" He responds, "Yes, the following people will lose their jobs: John Smith of 123 Maple Street, Alice Jones of..."
Back here in the real world, we can't foresee every single consequence. We can't predict exactly which categories people will choose to include. We can't even predict exactly which categories will exist: they change all the time. In just the last hour, about 75 new categories were created on Commons and about 20 were deleted.
So you're basically saying "Please show me the fixed, never-changing, guaranteed, final version of the product, so I can see exactly which images in which articles are going to be affected forever", and the fact is that what you want not only does not exist now, but it will never exist. This is a wiki: it will always change. We can give you general descriptions—there's pretty strong support for giving readers the option of not seeing certain general types of images—but the exact specifics, and how any given image fits into that, will definitely change over time. WhatamIdoing 21:43, 23 August 2011 (UTC)
the fact is that what you want not only does not exist now, but it will never exist -- Exactly. Which is why we are opposed to any project-wide filter category system. Beorhtwulf's argument illustrates that not only is the WMF proposal not perfect as can be expected from a work in progress, it is flawed beyond potential practicability as long as it includes special filter categories. There are most definitely far, far more than "5-10" types of images people will want to have filtered. On top of that, as Beorhtwulf correctly argues, there are types of images for which a category will be rejected out of reasons of political correctness (should e.g. my friend get the filter category for images of black men? are you saying we should have such a category? or are you saying that my friend's very real PTSD-based problem is not grave or widespread enough to warrant an image filter? how would you or anyone go about defining objection to a certain type of images as suffiently grave and widespread for a filter category? does that mean we only pay respect to widespread and politically correct objections to images?). All of that is conveniently hidden under the ETC rug. -- 22:06, 23 August 2011 (UTC)
Even if the system were fully implemented today, it would still be possible to change it in the future. We should not pretend that it won't. It would be silly to make promises today about exactly which types of images might or might not be included in the future.
In general, though, by limiting it to 5 to 10 groups of images, the WMF is signalling that the intention is not to provide thousands of options to cover every possible individual's phobia. If only 5 to 10 groups of images are available, then only the 5 to 10 groups deemed to be the highest priority will be implemented. If "images of men who look like the one who raped me" is in the top 5 or 10, then it might well be implemented; if it's not, it won't. People who need something else will have to do what they are doing now, which is either avoiding WMF projects altogether, or buying a third-party filtering system. WhatamIdoing 23:44, 23 August 2011 (UTC)
If only 5 to 10 groups of images are available, then only the 5 to 10 groups deemed to be the highest priority will be implemented. -- Who defines "highest priority"? According to what parameters?
People who need something else will have to do what they are doing now, which is either avoiding WMF projects altogether, or buying a third-party filtering system. -- Ah, so you're saying that the needs of the few should be ignored but the needs of the many shouldn't. That probably makes sense in your own worldview, but not in mine and hopefully not in the Wikimedia Foundation's. Also, there are alternatives (especially the adjustable general image filter) that take of everyone, not just the many.
Why are you so hellbent on the image filter being based on categories, especially when there are alternatives and when people are pointing out numerous fatal flaws with any project-wide filter category system?
Two blunt but honest questions at this point: (1) Are you being intellectually dishonest here (i.e. secretly recognizing mine and other people's valid points)? (2) Do you personally have a problem with any particular type of images that is likely to be deemed "highest priority" (e.g. nudity)? -- 00:10, 24 August 2011 (UTC)
"Highest priority" is determined by the community, which, in the end, is likely to mean what the most people (around the world, not just the young, single, childless, white males who dominate this page) demand.
The point behind using categories is that they already exist, and therefore impose far less burden than alternatives, like tagging specific images. If (to use your example of nudity) we wanted a category for nudity, it is the work of mere seconds to say, "Put Commons:Category:Human penis (or selected subcategories) on the list". Doing the same thing with a separate tagging system would require editing about a thousand image description pages. Doing the same thing with special, new categories would similarly require adding the new category to every one of those pages. Furthermore, categories are a well-understood technology, and we have a long history that helps us figure out whether any given file should be placed in any given category. Whether File:____ belongs in Cat:Human penis is an easy question; whether File:____ should be tagged as "Images that are offensive on sexual grounds" is not.
I have no expectation of enabling any of the filters (unless, I suppose, it becomes so widely used that it is useful for checking what other readers will see). I do not think that I need to protect myself from any images. WhatamIdoing 17:17, 24 August 2011 (UTC)
"Highest priority" is determined by the community, which, in the end, is likely to mean what the most people (around the world, not just the young, single, childless, white males who dominate this page) demand.
In other words, we'll vote on what is and isn't reasonably "objectionable," with the majority beliefs formally deemed valid and the minority beliefs dismissed.
If (to use your example of nudity) we wanted a category for nudity, it is the work of mere seconds to say, "Put Commons:Category:Human penis (or selected subcategories) on the list".
Does this count as "nudity"? Who decides? —David Levy 21:05, 24 August 2011 (UTC)
Who will decide how to categorize that image? The same people that decide how to categorize it now. There is no proposal to change the method by which images are categorized on Commons.
Who will decide whether the subcats under Commons:Category:Bikinis will be included under a "sex and porn" tickbox? The same community that does not connect those categories to any of the sex, porn, or nudity categories now. WhatamIdoing 17:58, 25 August 2011 (UTC)
1. The image filter system won't be limited to Commons; many Wikimedia Foundation wikis host images separately.
2. You're mistaken in your belief (also stated elsewhere) that no new categorization would be necessary. As has been explained in great detail, our current category system isn't remotely suited to the task planned. There's a world of difference between subject-based image organization and sorting intended to shield readers from "objectionable" material. —David Levy 19:02, 25 August 2011 (UTC)

You believe our current cat tree won't work. What does that mean? Most probably, you mean that you looked at a category like Commons:Category:Violence and decided that in your personal opinion, many of the images were not objectionable (however you are defining that), like a scan of a 17th century publication and a logo for television ratings. Then you assumed (wrongly, I believe) that this category would definitely be included in any list of "Tick here to suppress violent images".

The cat includes a few images that people who didn't want to see violent images probably wouldn't want to see. The first line, for example, shows a very bloody wrestling match. But that image would be suppressed by including either Commons:Category:Hardcore wrestling or Commons:Category:Professional wrestling attacks, basically all of which are images of violence or its aftermath. It is not necessary to include Cat:Violence (with its wide variety of image types) to filter out this image.

In other cases, the images are currently not categorized correctly. The images of the victims of violence belong in "Victims of violence", not in "Violence". The old paintings in belong in "Violence in art". The presence of so many images from movies suggests that Commons needs a "Category:Movie violence"—regardless of what happens with any filter. This is routine maintenance; people do this kind of work at Commons all day long.

There will be false positives on occasion. This is not a bad thing. It should be possible for people to identify most false positives by reading the image captions, and they will click through in those instances. If the contents of a category change so that are a lot of false positives, we can—and will—remove those categories, or substitute specific subcats. If the filter is restricting far more than people want restricted, they will turn off the filter.

I'm just not seeing the problem. We don't need to categorize specific images according our subjective view of whether the image is "objectionable". We're not promising perfection; perfection is not feasible. A reasonable list taken from the existing cat tree—based less on what should be considered "objectionable" and more on what actually produces objections—should be sufficient. WhatamIdoing 22:04, 25 August 2011 (UTC)

Most probably, you mean that you looked at a category like Commons:Category:Violence and decided that in your personal opinion, many of the images were not objectionable (however you are defining that)
I've been exceedingly clear in conveying that I strongly oppose reliance on my (or anyone else's) personal opinion of what's "objectionable."
Then you assumed (wrongly, I believe) that this category would definitely be included in any list of "Tick here to suppress violent images".
No, I've made no such assumption.
There will be false positives on occasion.
The greater problem, in my view, would be false negatives. As others have pointed out, our current categories are based upon what the images are about, not what they contain. For example, a photograph with a bikini-clad woman in the background (who isn't the main subject) currently is not categorized in any way enabling persons objecting to her sight to filter it. Using the current categories for this purpose obviously would dilute them beyond recognition, so for the system to work, we would need to create a separate set of categories. (Otherwise, there would be never-ending battles between those seeking to maintain the categories' integrity and those attempting to include off-topic images on the basis that incidental elements require filtering.)
In some cases, even an image's main subject isn't categorized in a manner indicating the "potentially objectionable" context. For example, many people strongly object to miscegenation. These individuals would have no means of filtering images such as this one.
And what about readers whose religious beliefs dictate that photographs of unveiled women (or any women) are "objectionable"? What current categories can be used to filter such images with any degree of reliability? —David Levy 23:18, 25 August 2011 (UTC)
I see no significant harm in false negatives. Sure: someone's going to upload another goatse image, and it's not going to get categorized instantly, and some innocent reader will want to bleach his brain. But—so what? How is that different from what happens now, except that we might reduce the odds of it happening by an order of magnitude? We're not promising 100% success.
We're actually promising that every single preference by tiny minorities won't be accommodated. Limiting the tickboxes to 5–10 options means that we won't be able to provide a filter that accommodates every single request. We'll have to focus on the 5–10 filters that are wanted by the largest numbers of users. This, by the way, means making those filters both as broad as our readers want, and as narrow as they want. For example, a filter that hides everything even remotely connected to nudity, from images widely considered benign (e.g., anatomical line drawings and marble statues) to images normally considered non-benign (e.g., photographs of sexual intercourse) is not likely to be as popular as a more narrowly tailored list. We might offer options that indicate degrees, like "serious sex and porn" and "most photographs showing genitals", but we are unlikely to waste one of our limited slots with "nudes in art" or "images of women" or "images showing people of more than one race". WhatamIdoing 17:28, 26 August 2011 (UTC)
"...images widely considered benign... to images normally considered non-benign" Perhaps my standards are not "normal", but if I was using a filter (for example, to make WP 'safe' for a child), I would want any and all nudity filtered, without pausing to consider how benign it is.--Trystan 18:30, 26 August 2011 (UTC)
Sure: someone's going to upload another goatse image, and it's not going to get categorized instantly, and some innocent reader will want to bleach his brain. But—so what? How is that different from what happens now, except that we might reduce the odds of it happening by an order of magnitude?
Under the current setup, readers familiar with our content have no expectation of protection from images to which they object. Those for whom this is a concern might disable the display of images via their browsers or add-on scripts.
As soon as a native filter system is implemented, readers will place their trust in it (and regard false negatives as unacceptable).
We're not promising 100% success.
That won't prevent readers from expecting it.
We're actually promising that every single preference by tiny minorities won't be accommodated. Limiting the tickboxes to 5–10 options means that we won't be able to provide a filter that accommodates every single request.
Indeed. That's a major problem, and one that won't be limited to "tiny minorities."
The belief that a photograph of an unveiled woman is objectionable is not rare. Will this be one of the five to ten filter options? Probably not. Why? Because we're allowing cultural bias to shape determinations of which objections are and aren't reasonable.
We'll have to focus on the 5–10 filters that are wanted by the largest numbers of users.
In other words, "if you aren't part of the majority, we don't care what you think."
We might offer options that indicate degrees, like "serious sex and porn" and "most photographs showing genitals", but we are unlikely to waste one of our limited slots with "nudes in art" or "images of women" or "images showing people of more than one race".
Right, we mustn't "waste" resources on those silly, non-majority-approved cultural beliefs.
To be clear, I agree that it would be impractical to dedicate slots to such subjects. We could have 100 slots and come nowhere near covering every "objection." That's one of the reasons why I oppose this implementation (and support the one discussed here). —David Levy 05:38, 27 August 2011 (UTC)
The literature on the history of warning labels in libraries does not support the suggestion that the community will happily settle on a few objectively identifiable, broadly agreeable categories. If we say that people have the right to filter images of nudity, and use Commons:Category:Human penis as one indicator of nudity, I think you will find that a very determined group of editors will be using that category to tag any image containing any depiction of a human penis, from Michelangelo's David on down. The WMF-endorsed user right of filtration will override good classification principles; it's not a very good reply to "I have the WMF-endorsed right to filter human penises," to say, "Well, yes, but this image isn't really about human penises, it just shows one." So any categories used as part of the system would cease to be organizational and descriptive categories, and become instead broad warning labels. You could certainly automatically populate the new warning label using existing categories, but they serve very different purposes and it would be vital to implement them separately.--Trystan 00:02, 25 August 2011 (UTC)
Trystan, I'm not sure I follow your reasoning. If you believe in the ALA's view of the world, there is no way to implement separate labels that would be appropriate and non-prejudicial. Allowing people to choose existing descriptive categories and say "I'd rather not see these images" is the only variant that might fit their criteria - with a style guideline that categories should be [remain] purely organizational and descriptive. That might not perfectly match the expectations of some readers, but then no automated system would. [By the way, if you think that there aren't already edit wars about whether or not to include certain images in controversial categories, you should spend more time on Commons :)] SJ talk | translate   01:23, 6 September 2011 (UTC)

"5-10" had better include the seven categories shown for this to be seen as legitimate[edit]

I would suggest that the "5-10" categories from question five should at least include the seven categories shown in this image which was linked from the voting instructions at or this vote will probably not be seen as legitimate. Those categories are: children's games; gladiatorial combat; graphic violence; medical; monsters and horror; other controversial content, and sexually explicit. 16:31, 24 August 2011 (UTC)

I doubt that anyone will mind if "children's games" disappears entirely from the list, or if "gladiatorial combat" and "graphic violence" are merged. WhatamIdoing 17:19, 24 August 2011 (UTC)
I think there's a good reason to exclude games: they can distract students from assigned work. I'd also like to see gladiatorial combat separate from graphic violence because a lot of very agressive forms of combat don't result in particularly violent images. If I had my way I would add religion and fiction to the categories too. 20:35, 25 August 2011 (UTC)
The games themselves might distract students from school work, but the pictures of the games will not do that any more than the text of the article will. This is only about hiding (temporarily) the images. It will not keep the students from reading the articles. WhatamIdoing 22:06, 25 August 2011 (UTC)

general image filter vs. category system[edit]

the only way forward is with a general image filter[edit]

I've noted this at other places, but would like to open a discussion specifically about this issue. Some proponents of the image filter appear to be hellbent on an implementation that relies on categories.

Me and many others argue that any possible category system would bring a lot of problems and is ultimately not a workable approach at all. The many varied problems, including neutrality, and catching all possibly objectionable images, have been discussed at length elsewhere on this talk page.

Personally, I believe that if we go forward with the image filter, it can only be as a general image filter:

  1. It could be instantly activated and deactivated by the user (or temporarily disabled on a per-image, per-page, or per-session basis).
  2. It is the only way to ensure that nobody will accidentally see images to which they object. At the same time, it is also the only filter variant that is actually "tailored" to each and every single last individual user.
  3. It would avoid the whole (imho completely inevitable) infighting and disagreement and drama over which images to include into which categories and so on. It would also exclude simple oversight.
  4. The general image filter could of course be made customizable by the user, who would sort individual images, articles, or entire categories into a user-specific blacklist and whitelist. Blacklisted images (or images included in blacklisted articles/categories) wouldn't show up for the user even when they disable the filter, while whitelisted images (or images included in whitelisted articles/categories) would be displayed even when the user browses with enabled filter.

In all, I can think of no argument that speaks for a category system rather than the general image filter. Imo, this is the single most important referendum-within-the-referendum. -- 01:07, 23 August 2011 (UTC)

I could support that, with one caveat. If it allows blacklisting of entire categories, there should be a policy in place stating that the purpose of categories is organizational, and not censorial (essentially the distinction drawn by the ALA above.) That way if people argue that an image should be placed in a category just so that it can be filtered (as opposed to it being a good fit for that category based on its subject matter), the policy rests on the side of maintaining the integrity of the category system.--Trystan 02:08, 23 August 2011 (UTC)
Yes, that sounds reasonable. -- 04:28, 23 August 2011 (UTC)
Both sound reasonable to me, if one intends to provide an image hiding feature. Both points are well defended by claims made elsewhere above. I agree that this is the essential implementation detail if we do indeed make it possible to set a personal preference for hiding images. There is only one argument I have heard for a category system: that it might be technically easier to implement. I don't think this is a good argument - a bad solution here could be worse than none. SJ talk | translate   01:23, 6 September 2011 (UTC)
What's the differences between your suggestion and the current proposed one? It's almost the same. -- Sameboat (talk) 06:15, 23 August 2011 (UTC)
The difference is: No labeling.
  • If you're looking at all this as "how can we make a simple filter system", this is merely a subtle point.
  • If you're coming at this from "how do we prevent people from exploiting the situation and doing us permanent harm", this is actually a really really important point.
Labeling appears to be exploitable.
--Kim Bruning 11:16, 23 August 2011 (UTC)
The 4th point is lame as I previously have explained. Without labeling by an organized group, it will be practically unmanageable by user alone. Don't give me that "if you want self-censorship, do the dull works all by yourself" nonsense. Our labeling does not do further damage than the existing censorship options. It will be too forgiving to be exploited by conservatives, authoritarian and communist. -- Sameboat (talk) 13:29, 23 August 2011 (UTC)
Mate, trying to explain things to you is like trying to play chess with a pigeon. But here goes anyway: I added point #4 to further expand my proposal with an option for the users to adjust the image filter according to their liking. They do not have to do the dull work at all. They can decide to simply switch the filter on and off and either filter all or none images. So, now you can knock over the pieces and so on. -- 13:36, 23 August 2011 (UTC)
Are you telling me the category you stated in #4 is the general category rather than filter category? That's pointless, category may contain images both worth filtering or not. -- Sameboat (talk) 13:46, 23 August 2011 (UTC)
the category you stated in #4 -- What "category" did I "state" in #4? I was talking about a user-specific and user-defined blacklist and whitelist. The categories I mentioned are the existing file categories. But that's just a tiny aspect of my proposal. If it turns out that blacklisting/whitelisting entire categories is not a good idea for some reason, I imagine it will be more because of what Trystan addressed above.
category may contain images both worth filtering or not -- In that case, it's for the user to decide the trade-off. Consider that a filter category would be largely out of control for the individual user and will in all likelihood also contain images that the individual user doesn't deem problematic.
Let me ask you something in return now: what exactly are the arguments that speak for a category system rather than the general image filter? -- 13:57, 23 August 2011 (UTC)
This IS the trade-off I can't accept. I need a separate category system for filtering. That's the whole point of this proposal. I need specific images to be filtered and the others unfiltered, the WM proposal will save me tons of time for that purpose. Do you know how many files uploaded to Commons right now for an individual to customize their black/whitelist? Commons:Special:Statistics gives you the answer. -- Sameboat (talk) 14:06, 23 August 2011 (UTC)
This IS the trade-off I can't accept. -- You're making progress! At least you're now talking about your own preferences. The question with regard to the image filter though is what is best for most people.
the WM proposal will save me tons of time for that purpose -- It may save you time, but it will cost the community a lot of time and stress. Time better spent writing articles.
That's the whole point of this proposal. -- Wrong. The whole point of the image filter is to give users the option to filter images. How exactly this should be implemented is completely up in the air. The possibility of a Wikimedia-hosted and -maintained filter category system is just one of the things determined via the referendum. It is not in any way, shape or form a necessary part of the final image filter. Fait accompli much?
I need a separate category system for filtering. -- That is exactly what the whitelist/blacklist system is: A separate, per-user filter category system, just better in that it allows each individual user to adjust the image filter exactly to their liking -- something which no project- or Wikipedia-wide category system could ever achieve. Alternatively, if they don't have the time to define their own blacklist/whitelist, users can simply enable/disable the filter at any time.
You still haven't explained the drawbacks you see with a general image filter, as they apply to most or all people, not just yourself. Enable the general image filter, you won't see any images. Want to see an image or all images? Disable it with a click, or whitelist it. What is the problem with that? The huge problem with any project-wide category system in that regard remains that different people all have a different set of images they deem objectionable. This is why the referendum asks users about 5-10 categories. But that arbitrary range of 5-10 categories is highly questionable, because it presupposes that there are only 5-10 different groups of images which people will want to have filtered. It also relies on the likewise doubtful presupposition that people can agree on which images to include into even one of those categories. So which direction is all this going to? Exactly right: individual per-user filter categories. Ergo I argue that the category system with all its inevitable drawbacks (neutrality, catching all objectionable images, agreeability etc) is dropped in favor of a general image filter, with the proposed additional option of a per-user blacklist/whitelist system -- which, again, is essentially a per-user filter category. -- 20:00, 23 August 2011 (UTC)
So this is an ALL images on/off function that already provided by browsers innately, case closed! OK not yet. The WM proposal is exactly allowing every user to customize their own categories. And this can be changed if 5-10 categories is deemed to be insufficient, that's why we're discussing it right here in search for the equilibrium of the amount of category and how categorization should be conducted objectively based on visible characteristics. -- Sameboat (talk) 22:15, 23 August 2011 (UTC)
See the section below this one for a deconstruction of that "argument". Project-wide filter categories are an utterly unworkable approach. That is just the fact of the matter, whether you are capable of and willing to wrap your mind around it or not. -- 22:22, 23 August 2011 (UTC)
I regard this approach as the best by far, for the reasons eloquently stated by I've previously expressed support for such a setup, and the implementation described above seems ideal.
Sameboat: You need to understand that a user could simply enable the all-images filter and whitelist individual images upon encountering them (e.g. when viewing an article and seeing a blocked image whose caption describes something that he/she wishes to view). Likewise, he/she could blacklist image categories and whitelist individual images contained therein. That's substantially more advanced than the functionality included in browsers.
This method is vastly simpler and easier to implement, requires no special community efforts, and eliminates numerous major flaws. —David Levy 22:29, 23 August 2011 (UTC)
The all image on/off does not conflict with the categorized image filter. Somebody want to examine it to decide whether they want to add it to whitelist or blacklist. But some user definitely don't want to come in contact with images they don't want to see at all but in the mean time still show the image that is acceptable for them. This suggestion is not a trade-off, but striping the right to hide SOME image initially apart from some simple category selection for the enduser. -- Sameboat (talk) 00:00, 24 August 2011 (UTC)
There are no "rights" in a legal sense on Wikipedia (except the right to leave), only privileges. And no, you are of course once again completely and utterly wrong. The adjustable general image filter is the only filter variant that does give each individual user complete control over what they want to see and what they don't. Now, since you seem unwilling and/or unable to understand anything explained to you, would you mind telling me what you are doing on an encyclopedic project? Just honestly curious. -- 00:16, 24 August 2011 (UTC)
Check my contributions on zh & en WP if you like, but this is irrelevant, the image filter is designed for readers, not editors. What the WM proposal gives is the ability to filter image according to user's preference without requiring them to examine every image they come in contact beforehand. Any inferior suggestion (your suggestion) which does not provide an equal substitution is not gonna be helpful. Many users just do not need complete control over the filtering preference. If some volunteers of the WM community are willing to do all the dull work to label most images for them, why reject it? You said user can use 3rd party script for the same purpose. I would prefer the filter being exploited than the user's computer being infected thru 3rd party script. -- Sameboat (talk) 01:07, 24 August 2011 (UTC)
The above proposal does not require anyone to "examine every image they come in contact beforehand." It enables readers to easily view a desired image with one click (without ever needing to review anything beforehand or display a single image of which they don't approve).
You're asking why we would reject volunteers' offer to label certain images "objectionable," thereby formally establishing that various cultures' and individuals' moral beliefs are valid (if reflected in the categorization) and invalid (if not reflected). You don't understand why anyone would oppose that? —David Levy 02:55, 24 August 2011 (UTC)
You're missing the point I care the most: how does the reader is supposed to know if the image in question should be filtered or not when the hide all images option is enabled? It is not that "if you're not confident, just don't open any image at all" because this suggestion is no more than a minesweeper game to me. -- Sameboat (talk) 03:17, 24 August 2011 (UTC)
Actually censors care the most, that others don't see the images or read the book. To cite Joseph Henry Jackson: Did you ever hear anyone say, "That work had better be banned because I might read it and it might be very damaging to me"? --Niabot 03:21, 24 August 2011 (UTC)
WP is not just "a" book, it is an internet encyclopedia covering almost every topic possible. Citing that quote is merely telling people to avoid obtaining any knowledge if they don't wanna be offended by some images. Or if I interpret that political quote correctly, it tells people who are too easily offended have no right in obtaining knowledge. -- Sameboat (talk) 03:53, 24 August 2011 (UTC)
You must be silly to think that this only referrers to a book. It aims at all knowledge, all book, all articles. It says that you can't judge about something you never have experienced. It also says that only censors will read (to late for them) and that they make the decision for others. Making decisions for other is censorship. --Niabot 11:34, 24 August 2011 (UTC)
This is another misconception against proponent of image filter. We want some of the image filtered because we already have experienced many similar images previously. In order to avoid experiencing that unpleasure once more we need the filter. And this is not censorship when it is the end-users asking for filtering some of the images for themselves ONLY, not for others, make no mistake. -- Sameboat (talk) 13:06, 24 August 2011 (UTC)
It is already censorship to label content as objectionable:
"Labels on library materials may be viewpoint-neutral directional aids designed to save the time of users, or they may be attempts to prejudice or discourage users or restrict their access to materials. When labeling is an attempt to prejudice attitudes, it is a censor’s tool. The American Library Association opposes labeling as a means of predisposing people’s attitudes toward library materials." [1] --Niabot 13:14, 24 August 2011 (UTC)
Applyin ALA opinion on labeling here is seriously wrong to begin with. Library labels are applied to each book while image filter categories are applied to image, self-explanatory. Image may be included in the book or article. Do you see the difference now? Labeled book in the library might turn off some potential reader from picking it, while image filter does the exact opposite because reader might know some degree of unpleasure caused by sensitive images can be alleviated, they can be more confident in accessing the article after adjusting the image filter preferences properly. And I want to add one thing of the general image filter, turning off all images is problematic because there're many images should never be hidden such as diagrams and statistic graphs which are equally text in the form of image. -- Sameboat (talk) 13:45, 24 August 2011 (UTC)
It isn't wrong. You might label books, articles, images, comments,... It's everytime the same question and problem. If you label images you exclude images. If you label articles you exclude articles. This has nothing to do with the fact, that you can still access the article, while the image is hidden.
What you propose is to rip the images out of a book and hand them out if the user requests them! Absolutely unacceptable. --Niabot 14:12, 24 August 2011 (UTC)
Whether the image will be hidden or not will be decided by user's filter preference, not WMF. Censorship is about thought/information control, while image filter is giving users freedom to choose. What WMF endeavors now is clearly the latter. -- Sameboat (talk) 15:09, 24 August 2011 (UTC)
Whether the image will be hidden or not will be decided by the censors that categorize the images for the filter. The WMF provides the tools for the censors. --Niabot 15:13, 24 August 2011 (UTC)
Disable the filter if you condemn it, then you're living in your colorful world without censorship, your life will be totally untouched by this feature. -- Sameboat (talk) 15:22, 24 August 2011 (UTC)

That isn't the problem we are talking about. We speak about prejustice of content, optional or not. --Niabot 15:30, 24 August 2011 (UTC)

Indeed. The issue, Sameboat, isn't that anyone will be forced to filter images. It's that the Wikimedia Foundation and/or its volunteers will formally judge what images reasonably should/shouldn't offend people and offer inherently non-neutral filter options based upon said judgements. —David Levy 15:41, 24 August 2011 (UTC)
The filter categories are not set in stone. This is a wiki, meaning things can be changed and refined to avoid prejudiced categorization and come up with more neutral and objective selections. And ALA's statement is ultimately made in fear of "restricted access" by the public, which may be a concern of information access policy in different states/countries/nations. But the thing is, whether accessing the particular information violates the law is judged by the local authorities/judiciary departments themselves, rather than depending on our labeling system. Which means if accessing particular media (for example, image of pedophile) is a crime in that region, labeling it or not does not change the fact that it is against the law. -- Sameboat (talk) 15:58, 24 August 2011 (UTC)
You can't change this by redefining categories. Any answer to the question "What is objectional content?" is "non-neutral". It will always be a "non-neutral" judgement. This starts with the names of the filter-categories itself. There are only two categories that are neutral. A category that contains not a single image and category that contains all images.
You don't need to put the terms "pedophelia" and "law" on the table, since non of our used images is pedophelia or in violation with the law. We don't need a filter for non existing content. --Niabot 16:07, 24 August 2011 (UTC)
I never define if the content "objectionable" to me. It is the visible characteristics of the image which should be used to shape the filter categorization system so there's no concern of prejudiced labeling. Pedophile is just an example. In reality there're many kinds of images on WM server which are legally acceptable in your hometown but illegal in many other countries, particularly the Muslim. But I almost forgot that I brought this out simply for refuting the ALA's statement, law is a reasonable mean to restrict information access. If law is not the concern, may the labeling hurt the people's self-esteem for accessing the labeled media? -- Sameboat (talk) 16:34, 24 August 2011 (UTC)
What visual characteristics define objectionable content? A question related to the previous one. But there are many more questions that come closer to the point: Why does the filter not allow the exclusion of flowers? What are the default categories? Which categories will be listed? Why are this the categories that are listed?
If you follow this questions back to core, you will giving an answer that provokes the question: "What is objectional content?" There is no way you could avoid this non-neutral question and to give an neutral answer.
Again. Law isn't a reason. We are only bound to US-law, with or without filter. The people in foreign countries are bound to the law, with and without filter. The filter doesn't have any impact on law. It's not a necessary tool to comply with the law. As such, law is not an argument of the debate. Thats exactly what the ALA states. If we introduce this censors tool, then without any need to do so. --Niabot 16:49, 24 August 2011 (UTC)
Simple enough: common sense and understanding of different cultures/religions. Note that I'm not trying the say this is an easy task, but if the slightly detailed categorization than the currently proposed 5-10 categories can cover the preferences of more than half of the users on WM, it is already a good start. -- Sameboat (talk) 17:01, 24 August 2011 (UTC)
At some point i will repeat myself like an record player... It should be obvious that there is no common sense that is acceptable by liberals and conservatives. There is also no room for compromise. An image is either inside a category or it is not. That is a straight line between yes or no. You will never have the case that all opinions are under or over the line. That makes a decission non neutral. A filter that allows all images is guaranteed to fulfill neutrality. It is a line at the top. No opinion is above it. A filter that excludes all images is a line at the bottom. Again, all opinions are on one side. The filter is neutral. Anything else is up to the pattern/opinions. You can lower or rise the line as long as you don't make a opinion change from yes to no or the other way around. That is the maximum tolerance for neutral judgment. For example: All would agree, that this is a flower. That someone disagrees is very unlikely, but it already limits maximum tolerance. What you try to find with "common sense" (i doubt that it exists in the first place), is the median line trough all opinions, described as a rule. If you found this median and you make a decision after this rule. The problem with "yes/no" categorization is, that 50% will agree with the decision and 50% won't. A non-neutral judgment is programmed in this case. You can't avoid it. If you would like, you could proof it with mathematics. --Niabot 17:33, 24 August 2011 (UTC)
Suppose that most users of Wikimedia Foundation websites are Christian. (I'm not claiming this to be true, but let's imagine that it is.) Would you support a filter system based strictly upon Christian beliefs? In such a scenario, this would "cover the preferences of more than half of the users," so would that be "a good start"?
Any system favoring particular sets of value judgements (whether yours, mine or someone else's) is unacceptable. Even if we could reliably include a vast majority (which I don't regard as feasible), we mustn't discriminate against minorities by formally establishing that their definitions of "objectionable" aren't worth accommodating. Neutrality is a core principle of the Wikimedia Foundation — one that can't be voted away by a majority. —David Levy 21:05, 24 August 2011 (UTC)
I think someone with better knowledge/reasoning than me should take my place in this discussion, otherwise just let this section fades over time. Maintaining neutrality does not mean denying or condemning the existence of unneutrality. True Neutrality cannot be achieved without accepting the biased opinions. That's why we never have neutral POV in our WP articles but stating every POV possible with citation. So is the filter system. It is meant to accept different preferences without imposing your so-called neutrality on other users. I'm out of it, but that does not mean this suggestion gained my acceptance. Making it a choice between black and white just disgusts me. -- Sameboat (talk) 01:32, 25 August 2011 (UTC)
Indeed, maintaining neutrality does not mean denying or condemning the existence of non-neutral views and cannot be achieved without accepting biased opinions. That's why I oppose implementations that would ignore/exclude certain cultures' beliefs and support one that would accommodate "every POV possible" to the best of our ability. —David Levy 02:22, 25 August 2011 (UTC)
Your arguments appear to be three:
  • that making a maintenance-oriented list of pages that concern people is prohibited by NPOV (which it isn't, or the anti-spam folks would be in trouble),
  • that NPOV is a mandatory policy on all projects, including Commons (which it isn't), and
  • that since we can't please 100% of users with a "perfect" solution, we should refuse to take even a small step towards an improvement that would please most of them (a fallacy known as making the best become the enemy of the good). WhatamIdoing 17:48, 25 August 2011 (UTC)
  • Filter categories like "violence" are no maintenance-oriented lists. Prejudicial labels are designed to restrict access, based on a value judgment that the content, language or themes of the material, or the background or views of the creator(s) of the material, render it inappropriate or offensive for all or certain groups of users.
  • The images are filtered and excluded/hidden from articles inside other projects. That the technical implementation will be hosted on commons has nothing to with this.
  • Labels on images or articles may be viewpoint-neutral directional aids that save the time of users, or they may be attempts to prejudice or discourage users or restrict their access to materials. When labeling is an attempt to prejudice attitudes, it is a censor's tool. [2]
--Niabot 18:01, 25 August 2011 (UTC)
1. I reject your description of this system as a "maintenance" task. It will be a reader-facing site feature.
2. The image filter system is to be implemented across all Wikimedia Foundation projects, including those for which NPOV is a nonnegotiable principle.
3. Setting aside the neutrality issue, I don't believe that the proposed setup is even feasible, let alone a step toward an improvement. The amount of resource consumption and extent to which perpetual controversy would be provoked are staggering.
Returning to the neutrality issue (and setting aside the issue of feasibility), I don't regard "pleasing" majorities (at the expense of the minorities whose value judgements are deemed invalid) as a laudable goal. We could "please" most people by customizing each language's projects to declare that whatever religion predominant among its speakers is correct, but that doesn't make it a sensible idea. —David Levy 19:02, 25 August 2011 (UTC)
An image, whether accessed directly or via one of the Wikimedia Foundation's various projects, typically is accompanied by descriptive text, which would inform the reader of the blocked file's subject and enable him/her to decide whether to view it. —David Levy 04:01, 24 August 2011 (UTC)
In order to inform the reader what characteristics that image contains, the description must be more detail than it normally requires. In the end, it is almost like labeling the image by filter category (I mean the labels of objective characteristics, not subjective interpretation). Not to mention its inconvenience that requires the user to access the file description page before filtering and possible lacking of description text or the language that user understands. Even though alt text can more or less fulfill that purpose, it is not an integral part of the file description page. The image alt text needs to be written per article which is the source of impracticability. Filter category eliminates the troublesome of internationalisation. While your suggestion greatly relies on the file uploader's self-discipline and more translation works for worldwide users. -- Sameboat (talk) 04:27, 24 August 2011 (UTC)
Actually, I'm thinking primarily of the text accompanying images in articles (or equivalent pages), which should clearly explain their basic nature. —David Levy 05:23, 24 August 2011 (UTC)
This is even more inefficient than labeling filter category because there can be more than 1 article/page embedding the same image. -- Sameboat (talk) 13:06, 24 August 2011 (UTC)
How does an image's use on more than one page render the setup inefficient? If a reader were to whitelist an image via any page, it would be unblocked for him/her on every page containing it. —David Levy 13:26, 24 August 2011 (UTC)
I mean the description text of image for reader to judge if that image worth being filtered or not. If there're more than 1 article including the same image, you need to copy and paste that description text to all including pages, this is needlessly repetitive. -- Sameboat (talk) 13:51, 24 August 2011 (UTC)
Copy and paste what description text? The pages in question already contain (or should contain) such text, which is essential for context.
For example, someone viewing the English Wikipedia's "Penis" article would see the top image described as "a selection of penises from different species at the Icelandic Phallological Museum." A reader with the general image filter enabled would choose to whitelist that image only if he/she wished to view "a selection of penises." —David Levy 14:11, 24 August 2011 (UTC)
The caption text is not always that obvious when the images are hidden indiscriminately. It may be sufficient in well-written articles, but generally not in most stub to decently short articles. Also considering the "random page" option and articles of unknown/utterly professional/ambiguous title (medical/surgery terms for example), the general image filter cannot prevent such surprising moment because the general image filter requires the user with some degree of vigilance of being offended before turning on/off the filter, while WM proposal does not. -- Sameboat (talk) 14:38, 24 August 2011 (UTC)
The caption text is not always that obvious when the images are hidden indiscriminately. It may be sufficient in well-written articles, but generally not in most stub to decently short articles.
If an article is missing reasonably descriptive captions, its images lack the context needed to be useful. This problem can be addressed in a straightforward, neutral and uncontroversial manner, thereby providing a valuable service to all readers. This is a far better use of volunteers' time than tagging "objectionable" images (an inherently complicated, non-neutral and contentious task) would be.
Also considering the "random page" option and articles of unknown/utterly professional/ambiguous title (medical/surgery terms for example),
If a user is unfamiliar with a subject, he/she can simply read the page (or relevant portions thereof) before deciding whether to unblock an image.
the general image filter cannot prevent such surprising moment because the general image filter requires the user with some degree of vigilance of being offended before turning on/off the filter, while WM proposal does not.
Sorry, I don't understand what you mean.
But if you want to discuss "vigilance," I can think of nothing in the Wikimedia Foundation ecosystem requiring more vigilance to implement and maintain than a category-based image filter system would. Millions of images would need to be manually checked (with countless tagging disagreements and edit wars inevitably arising), with thousands more uploaded every day. —David Levy 15:41, 24 August 2011 (UTC)
One of many practical reasons not to use custom categories. But note that this section was about not using a small # of categories at all, whether or not they were already in place and maintained by the community. SJ talk | translate   01:23, 6 September 2011 (UTC)
Indeed. I wrote the above in support of the "general image filter" implementation proposed in this section. —David Levy 12:16, 7 September 2011 (UTC)
I'm not sure this is the only way we could implement an image filter, but I'd be happy if we introduced it this way and when I use a slow connection I'd definitely use it with "hide all images" set on. I'm assuming that blocked images would have their caption shown as well as the click here to see the image box. We need to decide whether it would also display alt text or it would give people the choice whether alt text was displayed. I think my preference would be for the former. WereSpielChequers 10:20, 18 September 2011 (UTC)
If it is technically feasible, then I support the inclusion of alt text. WhatamIdoing 16:23, 18 September 2011 (UTC)

Give choice - not one central category list, but several user defined[edit]

I am strictly against a centralized categorization list. This is not self-censorship, this is censorship! The only option for me is - selecting each image by yourself OR -

having the choice to select from user-, or user-groups defined censorship lists. So every user can select pre-defined censorship list from users or groups he trusts. This is the only way to go for me, a centralized Wikipedia censoring-commitee is inacceptable. 07:09, 21 August 2011 (UTC)

Fully agreed. It's also amusing that some proponents of such a central, Wikimedia-hosted "objectionable content" list seem to believe that the same community that in their view is unable to make sound editorial decisions regarding images in articles should somehow be perfectly able to produce a workable filter category. Hilarity will ensue. -- 07:19, 21 August 2011 (UTC)
Please, WMF will not enforce users to use the filter. If you're worrying about the filter being abused by 3rd party organization, that's not our problem at all. The major problem of user-defined filter category is that you actually have to see then add the images to the filter categories all by yourself. It is also equally as bad as adding the original categories into your filter category. Considering the absurd amount of images and their categories on the server, it is too inefficient and impractical to be done by a single user for himself. The filter itself is meant to hide the image from users if they don't want to see it in the first place. What's the meaning if user must see each unwanted image first before filtering it? -- Sameboat (talk) 07:56, 21 August 2011 (UTC)
How is a category of "objectionable content" compatible with the core idea of an encyclopedic project, with core tenets like neutrality? A centralized blacklist would essentially label parts of Wikipedia's content as objectionable (by whose standards?) and non-essential (according to whose encyclopedic judgment?).
Even more importantly, how much sense does it make to label some content as objectionable and non-essential, yet at the same time to consider the same content as important enough for full coverage of the subject matter to include it in the article? -- 08:13, 21 August 2011 (UTC)
Nothing worth discussing with you if you think the filter category is identical to a blacklist. We're not talking of morality of the labeling but people who need this feature to avoid things they don't wanna see. Disable the filter if you don't want to use it. Objectionable or not doesn't concern you. -- Sameboat (talk) 08:35, 21 August 2011 (UTC)
Nothing worth discussing with you -- Oh wow. Insulting others because you're out of arguments. Well done, considering.
if you think the filter category is identical to a blacklist -- Nothing worth discussing with you if you intentionally seek any opportunity to evade the arguments. So you don't think that a centralized filter category that labels content as objectionable and enyclopedically non-essential is anything like a blacklist. Then ignore the word "blacklist" and focus on the reasoning I presented. Or do you need a filter button for the word "blacklist" in order to enter meaningful discussion?
Objectionable or not doesn't concern you. -- At least change your tone if you don't have anything of value to contribute to this discussion. And yes, it does concern any editor. Read my arguments, respond to them, or be quiet. Thank you. -- 08:44, 21 August 2011 (UTC)
You're the one who is evading. I've already explained the user-defined filter is practically inconvenient, unusable. All you do is complaining on the labeling to be objectionable. You, the one who will definitely not using this feature, are asking to shut the project down which is asked by a considerable amount of other users. We believe WM user group can do a better job than the 3rd party censoring option which may censor everything indiscriminately, because our filter is tailor-made for WM projects which can create less controveries. -- Sameboat (talk) 09:15, 21 August 2011 (UTC)

complaining on the labeling to be objectionable -- I am saying that an image filter category hosted and maintained within the Wikimedia project and specifically created for the purpose of filtering a community-decided set of images would, by sheer virtue of being offered to the user as a part of Wikipedia, become a part of the encyclopedic project itself. We'd be offering (i.e. suggesting!) images sorted into one or more filter categories to the users for blocking, thereby agreeing as a community that these images are objectionable according to some general/common/basic/universal? sense/feeling/opinion/consensus? regarding the objectionability of images. Like I also pointed out above, by including an image in a filter category (and offering=suggesting it for blocking), we'd also be labeling it as non-essential to the article. Which in turn would then bring up the question of why those images are there in the first place. An image either contributes to the encyclopedic coverage in an irreplaceable way, or it doesn't. If the image is and integral part of and essential to the article, then how could we be suggesting it for blocking? Conversely, if the image is not essential, then why is it in the article?

We believe WM user group can do a better job than the 3rd party censoring option -- I don't think they can. You can install Firefox, install the Greasemonkey add-on, install the userscript and all images on Wikipedia will be hidden by default, unhide-able with a click and with additional unobtrusive, Wikipedia-style links at the top of each page, giving the user the options to show all hidden images on that page and to disable image hiding for the rest of the session. It hardly gets any more convenient than that. When I know that I'm looking at articles that are unproblematic for me, I can e.g. disable image hiding, otherwise it's activated by default and thus failsafe and foolproof.

Again: This is all technology that already exists and which is freely available and which I am already using. And which, by not being part of Wikimedia, poses no problems with regard to the project's basic tenets. -- 11:15, 21 August 2011 (UTC)

Just because the filter categories are being stored in WM server, doesn't mean WM acknowledges/declares those images to be objectionable or non-essential to the article. You're making a fallacy here. I like to read some medical article but somehow the real photo may be too gory for me. I understand the image is essential to the article but I just want these kind of images hidden from my sight until I want a peep at them. A truly workable filter for WM must be hosted by WM itself and managed by its users, otherwise the whole project is to difficult to be effective and up-to-date. -- Sameboat (talk) 11:29, 21 August 2011 (UTC)
You keep ignoring the crucial points. I was talking about an image filter category hosted and maintained within the Wikimedia project. A filter category built e.g. by the English Wikipedia community means it's a part of the project itself. You cannot argue your way around that.
I just want these kind of images hidden from my sight until I want a peep at them -- 1, 2, 3, done. You're welcome. -- 11:59, 21 August 2011 (UTC)
I never ever trust 3rd party script. It is the best news the filter WILL be part of the WM project. It does not discriminate the labeled images objectionable or unessential to the project. That's all. -- Sameboat (talk) 12:13, 21 August 2011 (UTC)
I am tired of throwing reason at you, but I hope that someone else may profit more from my postings in this section than you managed to do. -- 12:21, 21 August 2011 (UTC)
You're making a huge fallacy to support the shut down of the project, that's why no one else is willing to reason with you except your fellow who echos your vague point. -- Sameboat (talk) 12:25, 21 August 2011 (UTC)
Ahm, yeah, right. The original poster of this thread, me, and a couple more people on this page, as will you will see if you at least skim over it. Farewell. -- 12:29, 21 August 2011 (UTC)
Ooh, perfect. That greasemonkey script might cover most issues nicely. Next problem! --Kim Bruning 13:53, 21 August 2011 (UTC)
It's only "perfect" if you think "see zero images no matter what" is a reasonable response to "I don't want to see images of mutilated bodies by default, but I'd sometimes like to be able to click through to them anyway". If a reader finds 2% of images distressing, do you want the reader's choices to be limited to (1) giving up all of the other 98% or (2) being distressed by 2%? WhatamIdoing 20:27, 23 August 2011 (UTC)
Greasemonkey can be enabled and disabled with a single mouseclick. Nobody's choices are affected in any way, shape or form with a general image filter, that's exactly the beauty of it. -- 23:15, 23 August 2011 (UTC)
I don't think that "nobody's choices are affected in any way, shape, or form" is an accurate description of "unable to see the 98% of images of images that I want to see without also seeing the 2% that I don't want to see", unless by that you mean "the only choice anybody should ever be given is 'all or nothing'". WhatamIdoing 23:46, 23 August 2011 (UTC)
This should never operate invisibly. It's important to make it clear to the user in every instance that some content has been concealed, why it was concealed, and what they need to do to view it. - Elmarco 21:10, 27 August 2011 (UTC)
Yes, that would be an absolute requirement. To the original idea of having a set of user-defined categories: you don't even need to have a visible list of what other users use. You could make it easy for readers to identify the categories that a given image is in, and choose a set of them to shutter. (One of the long-term technical improvements that this discussion highlights a need for, is access to metadata about media embedded in a page. It currently takes a number of aPI calls to get author, upload date, and category information about a file; and there is no quick way to see all of its parent categories. So for instance there is no easy way to identify all pictures of Books, since most will be in a sub-sub-category.) SJ talk | translate   01:23, 6 September 2011 (UTC)

Categories are not filters or warning labels[edit]

Categories are inclusive, filters are exclusive[edit]

Oppose for reasons quoted by many other posters, basically, the fact that offering filter tags makes government censorship a lot easier.

Remark on implementation: if this proposal does make it through the democratic scrutiny, then I remain opposed to the idea of it being implemented through categories. Categories were designed to be able to find pages, not to exclude them. Their purpose is inclusive, i.e., when you are searching for something then it is better to have a few results too many than to miss out on the exact page (image, resource...) you were looking for.

Censorship (aka "filtering") tags, if implemented at all, should have the opposite philosophy: it is better to accidentally come across an offensive image every now and then than to miss out on important information that may not be so offensive, after all. Resources should only be labeled 'offensive' after they have been established to be so, not 'potentially offensive' because someone thinks its a good idea to do so. In a debate on whether a certain resource 'potentially' has a certain property, the 'yes' side stands a better chance of winning by definition (in case of uncertainty, the claim remains 'potentially' true).--Lieven Smits 08:05, 22 August 2011 (UTC)

This is why images should be given objective tags such as "nipples, genitalia, blood, intercourse" rather than an "offensiveness" rating. 10:54, 22 August 2011 (UTC)
Well, keeping in mind that I am still very much opposed to the whole idea of tagging objectionable images, if we remain within the limited discussion of whether or not to use categories, I think that you mostly confirm my point: the 'normal' category Blood [3] would have a rather different set of pictures from the 'objectionable because too bloody bloody' tag. Hence: two different concepts, two different implementations.--).--Lieven Smits 12:02, 22 August 2011 (UTC)
Yes. The filtering categories are going to have to be totally totally different than categories like 'Shuttering requested due to _____ '. People will want to add images to the filter categories that don't objectively meet the stated criteria, and over the long haul we won't have a way to stop them from doing it. Admitting up front that the 'filter categories' are different is a good insight that will save us a lot of trouble later on. --AlecMeta 16:48, 22 August 2011 (UTC)
You're trying to invent a problem where there doesn't need to be one. We shouldn't be trying to define what is objectionable. You tag an image with blood, and someone determines whether or not they will allow images with blood to display on their computer. It doesn't matter how much blood or how objectionable anyone else might think the blood is. The presence of blood in an image is at least 99% of the time going to be an objective fact which is indisputable. You may get people who want to add improper categories, but this can be undone, just as people add bad content to articles and it is reverted. It will still be an encyclopedia anyone can edit. If such an objective system is used your concerns become moot, and you let the end user worry about which tags they want to block, if any. 19:46, 22 August 2011 (UTC)
In my theory, the desire of people to censor will be inarguable-- people will always find an image they dislike and then use any excuse they can to get it in a filter. And since bigotry is multiplicative, black gay men will get filtered all the time if they stand anywhere near ANYTHING potentially objectively objectionable. And that won't do.
We can't "filter neutrally"-- instead let's admit upfront all filtration is, by its nature, non-neutral. Stop and think about it-- we can live with this. Individual filters don't have to be neutral as long as there is no "one" filter or one set of filters or a pretense that filtration is objective, authoritative, or morality-based. We say upfront it's all prejudice, but if you REALLY need a very-slightly-prejudiced content, we can give you the tools to do that on your own time. --AlecMeta 21:39, 22 August 2011 (UTC)
It's still an encyclopedia anyone can edit. Why are black gay men more likely to be filtered than deleted entirely in the absence of filters? If you properly use objective filters, the problem you propose can't happen. Say a black gay man is standing nude, and you filter "penis". The presence of the penis is not debatable, and he will be just as filtered as an image containing a straight white man's penis. Now suppose a multiplicative bigot sees a picture of a clothed gay black man, standing in a room in his house, everything normal and everyday about the scene. What is the bigot going to tag? Wall? Chair? How many other users will filter photos by the tag "chair"? If the black gay man is at a pride parade, maybe the image will be tagged with "homosexuality" or "gay pride" or something of that nature, which some could consider objectionable, but only those who choose to filter those contents will have the image blocked. If you filter purely by keywords, and don't assign an offensiveness rating, it will be nigh-impossible to game the filtering system to prevent others from seeing the image you don't want to see, since everyone will filter different keywords rather than choosing the "moderate" setting. You could vandalistically add a false tag, but someone will notice and remove it, just as articles are vandalized and reverted. 04:58, 23 August 2011 (UTC)
I may not have been clear on the example of 'Blood' categories/filters/tags. It may be objectively verifiable whether an image contains blood, but that is beside the point. The category 'Blood' should not contain all images with blood on them, but only images whose content is relevant to someone who intends to learn about blood: blood groups, microscopic images of blood cells, chemical formulae, test tube images, blood drives etcetera. The other 'thing', the category that you intend to use as a filter against objectionably bloody images, would presumably not even include any of those; it would contain depictions of humans and possibly animals who are bleeding. The two concepts of categorisation, while both deserving the name 'blood', are very different both practically and theoretically.
My principal opposition against creating the other 'thing' remains. It can and will be used to limit other people's access to information. Apart from being almost the exact opposite of NPOV.--Lieven Smits 07:15, 25 August 2011 (UTC)
You are confusing Wikipedia's existing category system and a content filter keyword system. They would be separate systems. 02:47, 30 August 2011 (UTC)
I agree that Wikipedia categorisation serves a very different purpose from content filtering. I do not think that I have mixed the two concepts up.
The purpose of the proposal is essentially content filtering. But several contributors on this page have argued that the proposal includes using the existing category system for that purpose. Thus, confusion appears to be in the air. I have tried to argue that the category 'Blood' should be distinct from the content filter 'Blood' if there is ever a need for the latter. The category 'Blood' is unfit for filtering out shocking images of bleeding people and animals.--Lieven Smits 09:42, 30 August 2011 (UTC)

Your point about categories having an inclusive, not exclusive purpose is a good one. Nevertheless, for sufficiently specific categories, the two become the same. SJ talk | translate   01:23, 6 September 2011 (UTC)

Libraries have fought warning labels for years[edit]

When it comes to passionate defenders of intellectual freedom and access to information, one need look no further than the library profession. They have opposed warning label systems, like the one being proposed, for years. From the American Library Association:

Labels on library materials may be viewpoint-neutral directional aids that save the time of users, or they may be attempts to prejudice or discourage users or restrict their access to materials. When labeling is an attempt to prejudice attitudes, it is a censor's tool. The American Library Association opposes labeling as a means of predisposing people's attitudes toward library materials.
Prejudicial labels are designed to restrict access, based on a value judgment that the content, language or themes of the material, or the background or views of the creator(s) of the material, render it inappropriate or offensive for all or certain groups of users. The prejudicial label is used to warn, discourage or prohibit users or certain groups of users from accessing the material...

I would encourage anyone interested to read the full statement and seek out some of the substantial body of library literature on why warning labels are incompatible with intellectual freedom.--Trystan 13:59, 22 August 2011 (UTC)

Trystan, while I know some proposed implementations have involved a custom set of 'controversial' categories (which would qualify as prejudicial labels), I believe the more general quote from that document which is relevant here is: "Directional aids can have the effect of prejudicial labels when their implementation becomes proscriptive rather than descriptive." We would need to be careful in any use of existing categories as well not to present them in a proscriptive way. SJ talk | translate   01:23, 6 September 2011 (UTC)
Read w:Wikipedia:Free speech please. Your right of speech here is a privilege granted by WM, not US government. -- Sameboat (talk) 14:12, 22 August 2011 (UTC)
You get to choose what 'warning labels' you want. No body is going to paste a 'label' on your 'book' for you. --Bencmq 14:18, 22 August 2011 (UTC)
@Sameboat: I don't get your point at all: Trystan is not even American, as far as I know. @Bencmq: you are not getting Trystan's: the issue is that, once the labels are there, you can be denied access to the books by third parties. Besides, the labelling would be done by the wikipedia community, including, but not limited to, shills of the Chinese government, not by the single user. --complainer 14:24, 22 August 2011 (UTC)
On that point I agree. There is inevitably going to be ridiculous edit warring over images whether they should go into a certain category that is used for the filter - but I think the availability of the tool itself is as important as well. I hope, before we see anything, the likelihood for such inevitable dispute could be minimised through technical means.--Bencmq 14:30, 22 August 2011 (UTC)
Sameboat, I am not American, and am aware that free speech guarantees don't apply to private organizations regardless. The issue is not free expression but the related concept of intellectual freedom. I chose the ALA wording because it is well-expressed, but there are librarians world-wide that support intellectual freedom.
Bencmq, the warning labels show up in at least two places that have the potential to prejudice users in the sense that the ALA is talking about. First, in the filter itself ("Do you want to avoid looking at controversial images like sex, nudity, violence, immodestly dressed women, blasphemy, or gay people?") Second, on the image pages themselves, which would need to be appropriately tagged with those labels for the filter to work.--Trystan 14:32, 22 August 2011 (UTC)

Yet every library I've been to is divided into a Children's section, Young Adults' section, etc. And there's librarians and parents around to ensure that children don't wonder off. There's a difference between a government telling organizations what to do and organizations using their own judgement. --Michaeldsuarez 16:08, 22 August 2011 (UTC)

That's an excellent point, thanks for bringing it up. Material is organized to help the reader find it, not to warn about it's content. It is a frequently attempted form of censorship to try and get material aimed at children or youth filed in the adult section based on ideological objection (e.g. Children's books like Heather Has Two Mommies or Judy Blume Young Adult books dealing with sexuality.)
As for librarians monitoring children, absolutely not. Libraries very explicitly do not stand in the place of parents. We do not scrutinize or evaluate what patrons are doing, and are most certainly not censors for other people's children.--Trystan 16:26, 22 August 2011 (UTC)
(ec) You raise some great points, Trystan. w:Wikipedia:Free speech not withstanding, I actually do think of labeling offensive content raises very real 'free speech' issues, because if we did filtering "wrong", we might empower or legitimize real-world censors taking away our readers' real-world rights.
But labeling done "very right" can be way less bad once you take away any sense of "authority" behind the labelling. If the foundation personally decided on "the best" labelling scheme, it would come with authority. But filtering will be massively parallel, "buggy", controversial, and it won't work the way users might expect it to. The kind of "prejudgment" our labels will contain is far less reliable and monolithic-- nobody will agree on where the lines are or if there should even be lines. It will become apparent to anyone using the feature that its logic is irrational, idiosyncratic, and non-ideal-- in short, our labels won't be monolithic and authoritative.
This isn't exactly what our 'heart' tells us to do, but let's be realistic. A library is usually located in a specific location-- we are everywhere-- buses, subways, airplanes, offices, mosques, and everywhere else. We also are a 'face of the internet' to whole cultures that are getting online for the very first time-- imagine our ancestors at the dawn of film, when the silent black and white image of a train heading towards the viewer could actually elicit enough fear to be called a "scary movie"-- and now realize that whole cultures are going to be meeting us at the same time they meet any internet technology. I don't know who these cultures are, I don't know where they are precisely, and I have no idea what they need in a "library".
If they really deeply want a very simple 'privacy shutter', should we really let our original vision of the ideal global library stand in the way of them having it? I ask this none-rhetorically, but my emotions are saying that this is 'close enough' that we should give it to the people who want it. --AlecMeta 16:17, 22 August 2011 (UTC)
Hm, but. For that to make sense, we would have to place a permanent and quite prominent note on every single content page (not just the main page, where first-time users may or may not arrive!) drawing attention to the filter option. Otherwise, new and technically inept users in particular may keep running headlong into objectionable material (which probably has done untold damage over the past years, including work stoppages, people poking out their own eyes, and spontaneous combustion). -- 16:30, 22 August 2011 (UTC)
I agree that we'll have to advertise the filter on either all content or content that has objectively been shown to upset people to a certain _objective_ level. It will still involve the reader having to do some effort on their own, so it won't be perfect from their point of view, but it will be as close to perfect as we can get. --AlecMeta 18:54, 22 August 2011 (UTC)
Nope. The closest we can get to a perfect filter is a general image filter. That is the only way to (i) guarantee that --once they activate the filter-- people won't see any objectionable images and (ii) to prevent any completely unnecessary snafu that would inevitably accompany any filter category system.
Bottomline: IF the image filter is implemented, it should most definitely be done as a general image filter, which users can enable and disable at will. What exactly, in your opinion, actually speaks for a category system rather than the general image filter? -- 00:44, 23 August 2011 (UTC)
I would certainly like to see what sort of interest there is in a general image filter. That is the only version of this feature that I could imagine myself using. (for a small and annoying phobia.) SJ talk | translate   01:23, 6 September 2011 (UTC)
«labeling done "very right"» How can we expect to do it right. It might be right for yourself, but not for others. You will be in the trouble that the average American, the liberalst German and the conservativest Arabian have to agree that the labeling is done right. Otherwise you simply don't do it in a "very right" way and all your argumentation is for nothing. I hope that this is clear. --Niabot 18:25, 22 August 2011 (UTC)
Either that, or we create several different filter categories, one "Typical Arab Sensibilities Filter", one "Typical Conservative Christian Sensibilities Filter", one for mom, one for dad, one for grandma, one for granpa, one for average children, one for smart children who are ahead of their age. I can see it now, it suddenly all makes perfect sense. Seriously, to anyone easily offended by sarcasm: I feel sorry. For you. -- 18:38, 22 August 2011 (UTC)
Yes! Precisely. I'm not sure if all the participants who want a filter yet realize this basic insight that you capture so well-- there is NO SUCH THING as a filter everyone can agree on. The liberalist American and the most conservative American will NEVER agree on a filter, and that's just Americans. There is absolutely no consensus to find-- it's JUST opinion.
So "Very Right Labelling" is infinite labelling-- it's a wiki with infinite forks-- as many categories as people want, right down to per user. I could make a whole category with no better label than "images that offend AlecMeta" and that would be fine. No one would care what offends me, no one would expect ANY labelling be perfect any more than we expect any article to be perfect.
Doing this "very right" would actually be something we could actually be proud of, at the end of the day. We could show Youtube and GoogleSafeSearch that there's a better way than a "one-size-fits-all" labeling. We could, if we chose to, make filters in infinite diversity in infinite combinations. NO two people would EVER agree on a filter, and we could prove it in front of the whole world.
I still don't see it as a priority. But if it has a low price tag, is it an exciting and educational project? absolutely! --AlecMeta 19:02, 22 August 2011 (UTC)
Alec, the point is that no filter category or set of filter categories will ever be sufficiently evolved, nor will they evolve towards anything useful in the first place. You're trying to play the eventualist / incremental improvement card: "it doesn't need to be perfect right now, but it's approaching perfection". No, it's not. Not only will it never reach perfection (as in: working really well as a filter category/categories), it will actually never even be approaching anything like that. -- 19:13, 22 August 2011 (UTC)
There is no perfection to find for this kind of project. It will never make anyone completely happy. But if we want to actually understand what images upset people, Mediawiki is a powerful technology, and we can use it to let people make filters that are "as perfect for them" as time and effort allows. Most of them will always be upset that their filter isn't imposed on everyone, but that's okay. --AlecMeta 19:20, 22 August 2011 (UTC)
But thats all under assumption that you could create personal filters. Or at least a bunch of them. I doubt that it would be practical to have a list with 100 kind of different filters from that some might hit your expectations. At least for the endusers this would be a pain. "Oh, cool. I can choose from 100 filters. I will spend an hour in reading and choosing. But, damn, i just wanted to read something about XYZ". --Niabot 19:55, 22 August 2011 (UTC)
Indeed, it is all under that assumption that we "do it right". If we do this very wrong, it would blow up in our faces very badly. If we do it only semi-wrong, I predict the Image Filtration project will quickly evolve unlimited categories as we learn, experimentally, that there are no consensuses and no hard lines. What's on the table is close enough to perfect that either it will be "good enough" (and I'm wrong) or it will develop infinite categories with time. Even better, of course, would be to get it right the first time for a low price, if that's an option.
The first screen would probably only show the n most popular filters, and trying to make sense of infinite filters would be a big challenge. It's just about offering people who are displeased with the status quo a limited SOFIXIT power. People are coming to us upset with our content, the solution is to give them a 'project' and let them try their best to solve it for themselves. If they fail, that's okay-- the experiment doesn't have to succeed for it to be a success. Offering people the chance to fix it is enough. --AlecMeta 20:02, 22 August 2011 (UTC)
Now you did a whole lot of good faith assumptions. At first it is a bit doubtfull, that people, don't wanting to see the images (offended), will actively search for such images and categorize them in a neutral way. The second point is the default filtering. It's still on the table. Which will be the default filter? (More objections to follow, but two for the start) --Niabot 20:35, 22 August 2011 (UTC)
People won't categorize in a neutral way-- there will be no neutral way, there will be no neutral categories, just like real life. There is no default-- default is off. Even when turned on there is no default, only 'most popular'. (and I admit a LOT of assumptions, so by all means pick my arguments apart, that's what they're there for.) --AlecMeta 20:46, 22 August 2011 (UTC)
Right, right. It's about the children, all about the children, always and only about the children. Won't somebody please think of the children?! Great argument! So new and fresh and powerful, kinda. -- 16:19, 22 August 2011 (UTC)
I just can't express deeply enough how much this is NOT about the children. The children don't want this, the children don't need this, and the children can get around this without blinking. Won't somebody please think of the grandparents who can never unsee Goatse! --AlecMeta 19:28, 22 August 2011 (UTC)
  • 50 years ago, childrens libraries were indeed restricted to material thought suitable for them. Movies were censored also, on the same traditional grounds as that proposed here: sex, and violence. For that matter, so were encyclopedias, which carefully avoided anything that might disturb the readers. It was a survival of what once was a important cultural value, paternalism in the quite literal sense, where (usually upper-class) adult males prevented other people from reading or see what these supposedly more responsible people thought them unfit to see. We did not make an encyclopedia on that basis. We were most of us raised in a world that was free from this, or at least becoming free from this, and we did not make an encyclopedia to suit the views of our great-grandparents. We're a contemporary encyclopedia, and contemporary means uncensored. It was one of our first principles. We also believe in the freedome of our readers. If they choose of their own will to censor it with their own devices they are free do do so. regardless of what we may personally think of that decision. But we an not in conformance with our principles encourage them to do so, or adopt a system aimed at facilitating it. The board seems not to have been as much committed to free expression as the community, and the members of it need to rethink whether they are in actual agreement with the goals of the organization they are serving. DGG 17:31, 22 August 2011 (UTC)

Any idea, why the American Library Association has made such a proclamation? Because they were and are under pressure from local politicans - and we are just the next one on their targetlist. --Bahnmoeller 18:32, 22 August 2011 (UTC)

It is worth noting that we are not, in fact, under pressure from local or global politicians on this score. A few countries block us, but not primarily for our use of images. And some filters may block commons, but we do not receive significant political pressure to change as a result. SJ talk | translate   01:23, 6 September 2011 (UTC)
[Warning: the following post contains feminist, gay, and blasphemous content, which is offensive to many users. If you are offended by this content, please make sure your filters are set to hide it.]
...but in all seriousness, you raise some excellent points, Alec.
I don't think taking the labeling authority away from WMF and giving to users makes much difference. If someone tags a picture of a woman wearing shorts and a T-shirt as "Image filtering: Immodestly dressed women", there really is no response to that. I can't sensibly argue that the person who tagged it doesn't find it offensive, as it is completely subjective. And WMF will have validated their right not to be offended by such images, so I guess it should be tagged that way. But labeling an image with such a pejorative, moralistic label is antithetical to intellectual freedom principles. Whether it's done by an authority or a collective, I simply don't see a right way to indicate that wanton women with bare arms and legs/sinful gays/blasphemous Pastafarians are offensive and that the user might want to think about blocking them.
"I have no idea what they need in a "library". We don't really need to know in that sense. Whatever the culture, we provide free access to information (free from warning labels prejudicing users against that information) and let individuals access what they want to. In every culture there are those who want to freely access information and those who want to stop them. A commitment to intellectual freedom sides with the former group and actively opposes the latter.
There are also, in every culture, disenfranchised and unpopular minorities. Time and time again, calls for warning labels disproportionately target these groups. Empowered female sexuality, cultural minority traditions, gay and lesbian issues, and religious minorities are frequently held to stricter standards than what the majority deems proper. Adopting a warning label system just helps perpetuate this.--Trystan 00:12, 23 August 2011 (UTC)
Agreed. If we make a category "Immodestly Dressed Women", we should naturally expect it to grow to include every possible image that could be added to the category in good faith-- and on earth, that's going to wind up meaning "Immodestly Dressed Women" ultimately filter all images of women that show more than eyes-- there would be no way to stop that expansion. The whole system will be prejudice embodied-- but prejudices are just another form of information, and if people want to tell us about their prejudices, we can treat those prejudices as information to be shared, categorizes, forked, remixed, etc.
We have no clue how to reach consensus across languages, across cultures, across projects, but we are going to learn those things. We also have no clue how to reach consensus on 'offensiveness'-- and I think consensus has to fail there because no consensus exists. So we use forks instead of consensus-- that's okay. the image filtration project, by it's nature, is nothing but POV-forks already, because POV is all it is. --AlecMeta 00:38, 23 August 2011 (UTC)
I have great interest in participating in a project committed to neutrality and intellectual freedom. I have absolutely zero interest in participating in a project that rejects those principles and sets out to embody the prejudices of its users.--Trystan 02:17, 23 August 2011 (UTC)
Mmm-hmm... but you seem to have forgotten the important distinction between "censorship" and "selection". Have you ever worked for a library that has a collection of Playboy magazines? No? Why not? They're popular, they're informative, they're in demand by readers, they have cultural importance—but you haven't ever recommended that any library acquire them, have you? there are just ten libraries in all of Canada that include even a single copy of this magazine. Nine of them are university libraries. The other nearly 2,000 don't list it in their catalogs at all.
So if not acquiring Playboy doesn't violate your commitment to intellectual freedom, then how could letting me decide whether to display video stills from hardcore porn movies on my computer screen possibly violate intellectual freedom? Is skipping porn a "legitimate acquisition decision" only when it's done by a paid professional and affects every single user in the system, but merely base "prejudice" when the individual reader is making the decision to choose something else? WhatamIdoing 21:00, 23 August 2011 (UTC)
Selecting and proposing images for filtering equals labeling them as (i) non-essential to the article (in which case they shouldn't be included in the article in the first place) and (ii) as somehow generally potentially objectionable (according to whose standards?). The problem is not the "censorship" strawman you keep raising (although it is detrimental to the level of discourse that you keep raising it), but the fact that proposing (ie. suggesting!) images for filtering equals labeling them as somehow negative, and as non-essential to the articles -- to which all images are supposed to make an irreplaceable contribution. -- 21:15, 23 August 2011 (UTC)
Every image does not make an irreplaceable contribution to an article, just like every book does not make an irreplaceable contribution to a library, and every artwork does not make an irreplaceable contribution to an art museum. A brief look at the battles that the English Wikipedia has had with keeping vanity shots and extensive galleries out of some articles is sufficient to disprove this idea. To give one example, en:Bride has been reduced to "only" nine images, half of which are still unimportant.
You are proposing that a person going to a museum not be allowed to pick and choose what he (or she) looks at, but must look at every single image in every single room merely because the curator thought it worth displaying to the public. Why should I have to look at every single image? Why should I not look at the images I want to see, rather than the images that someone else wanted to display? Does your notion of free speech include an inalienable right for speakers to have an audience, but no freedom for you to use earplugs if you don't want to listen to the man yelling on the street corner? WhatamIdoing 23:28, 23 August 2011 (UTC)
Playboy is an interesting example. You make a very good case for libraries carrying it, and some choose to,[4] with somewhat predictable results.[5] Whether each individual library's decision with respect to that title is censorship or selection is a very complex and interesting and debatable question.
I wouldn't say that skipping "porn" per se is based on prejudice, depending on how pornography is defined. The difficulty is that in defining pornography for people who wish to avoid it is very difficult. The prejudice we are likely to introduce into our system is that people tend to be much more offended by (and tempted to label as "pornographic") certain types of images. For example, women are often held to different standards than men in terms of acceptable dress. Images of same-sex couples attract disproportionate criticism compared to similar images of opposite sex couples.--Trystan 23:39, 23 August 2011 (UTC)
I don't think that anybody believes that the filter could be 100% effective, if for no other reason than the undeniable fact that commons:Category:Media needing categories has thousands of uncatted images, and new images turn up every second of the day. We already have problems with correctly categorizing borderline images, and nothing here will change that.
It's true that the borders of categories are sometimes hard to define. We already see people getting blocked for edit warring over the cats, and that is unlikely to change. It's likely to be a 98% solution rather than a 100% guarantee. But none of those limitations are good reasons for saying that it's unethical to give readers the ability to use neutral information to make their own choices about which parts of our "library" they personally choose to see.
In fact, I believe that the greater ethical transgression is forcing them to see what they do not want to see. If your library included Playboy, you would never shove it in front of an uninterested reader, and you would stop any patron who did that to other patrons. You'd be calling security, not telling the other patrons that the library had a legitimate reason to include porn magazines and if they didn't like having them shoved in their faces unexpectedly, then they should stop using the library. Similarly, I think we should support allowing a Wikipedia reader to say, "I don't want to see that, so please stop shoving it in my face". WhatamIdoing 00:03, 24 August 2011 (UTC)

Negative effects of using categories as warning labels on the usefulness of categories[edit]

This topic has been mentioned in a few places above, but I thought I would add it under a separate heading here for a fuller discussion.

Categories, effectively used, attempt to capture that elusive quality of aboutness; the main subject of an item, in this case an image. They let us organize and find items based on content. Warning labels are very different; they attempt to flag any image which contains the offending content, regardless of what it is primarily about.

For example, we have a category about Nudes in Art. The implied scope of this category, based on how it has been used, is something like "artistic works primarily featuring the nude form." The scope is not "artistic works which contain any nudity whatsoever, however trivial." That would not be a useful scope for it to adopt. But if Nudes in Art becomes the warning label for people who want to filter out nudity, it will become diluted in that way. A category being asked to serve double duty as a warning label will invariably become diluted and much less useful.

There is also a detrimental effect on the images being categorized. A good practice is to categorize something based on a limited number of categories that best describe it. For example, this image of the Barberini Faun has a very well-chosen set of descriptive categories, but (goodness!) they forgot to include the all-important fact that he is naked. Let's add that, and any other potential objections people might have about it, and drop off some of those categories which merely tell us about its significance as a work of art. If we make warning people a primary function of our categorization system, it is to the detriment of its descriptive and organizational power.--Trystan 22:44, 21 August 2011 (UTC)

This is no difference than stating the filter will make the labeled image "unessential to article". I have enough such fallacy here already. He may not think so or just worry other's thought. But WM can and should establish a statement that filter categories do NOT reduce the priority and usefulness of labeled images to be included in the article than unlabeled images. -- Sameboat (talk) 23:28, 21 August 2011 (UTC)
the filter will make the labeled image "unessential to article" -- Funny that you would mention it, seeing as you didn't have a good reponse to that powerful argument above, either. Again, not for you, Sameboat, but for others who may inadvertently think that you have any sort of legitimate point (which you do not): Sorting an image into a category of objectionable content and offering it to users for blocking equals (among other things) labeling it as non-essential to the article. That's not a fallacy, nor a statement of opinion. It's a fact, and as such non-negotiable reality.
An image either contributes to the encyclopedic coverage in an irreplaceable way, or it doesn't. If the image is an integral part of and essential to the article, then how could we be suggesting it for blocking? Conversely, if the image is not essential, then why is it in the article in the first place?
WM can and should establish a statement that filter categories do NOT reduce the priority and usefulness of labeled images to be included in the article than unlabeled image -- Unfortunately for your line of reasoning, that statement is blatantly untrue, whoever says it. -- 01:32, 22 August 2011 (UTC)
Of course if labeled image would be refused by editors to include it in the article, WM would not propose the image filter in the first place. Your statement may be true for some editors, but definitely does not represent the whole community. -- Sameboat (talk) 04:22, 22 August 2011 (UTC)
Your statement may be true for some editors, but definitely does not represent the whole community. -- What on god's green earth are you blathering about? -- 09:03, 22 August 2011 (UTC)

When labeling is an attempt to prejudice attitudes, it is a censor's tool. The American Library Association opposes labeling as a means of predisposing people's attitudes toward library materials. --Bahnmoeller 18:20, 22 August 2011 (UTC)

True, but irrelevant.
Labeling an image of the Prophet Mohammed as being an image of the Prophet Mohammed is informative and appropriate. It is not a means of predisposing people's attitudes towards the image: it is a method of telling them what the image contains. The ALA would object to labeling such an image as "artwork created by evil people" or "artwork whose existence is an offense against all morally sound people". Those labels would prejudice the viewers' attitudes, but nobody is proposing that we do that.
Similarly, the ALA does not object to labeling pornography as pornography. This is informative, appropriate, and non-prejudicing. It would object to labeling it as "things decent people would never look at" or "materials only of interest to sexual perverts", but nobody is proposing that we do that.
The ALA does not oppose labels; it opposes labels whose purpose is to impose the librarians' prejudices on the users. The ALA does not object to readers using informative labels to help them choose which materials they want to see. WhatamIdoing 21:21, 23 August 2011 (UTC)
Labeling an image of the Prophet Mohammed as being an image of the Prophet Mohammed -- Ah, but that is not all of what we'd be doing if we implement a variant of the image filter that relies on special filter categories. We'd be categorizing that image and suggesting it for filtering! -- 21:39, 23 August 2011 (UTC)
Only in the sense that my local library indicates to me that an entire section of the library is mostly garbage by labeling it en:Young-adult fiction. I am independently prejudiced against such (commonly depressing and frequently idiotic) books. Are they wrong to label the books in a way that allows me to exercise my prejudice against them, by choosing books from other parts of the library? WhatamIdoing 23:18, 23 August 2011 (UTC)
Do you see how you need to go back to the library analogy in order to make a "point"? If libraries had a user interface and were suggesting certain books or types of books for "filtering" of some sort, yes, they would be wrong to do so and it would be very different from simply labeling the books according to their content. -- 23:22, 23 August 2011 (UTC)
My library provides exactly that filter in their online user interface. They also provide filters according to format, subject, language, publication date, and more.
You are claiming that libraries do not provide labels that readers can use to filter the contents of the library because you think the ALA prohibits it. I am proving that you are wrong: they do this routinely. Their goal is to have "a book for every reader, and a reader for every book", not "read this because the librarian know better than you what you should read". That means letting the readers ignore and reject the books that they do not choose to read—just like Wikipedia should allow readers to ignore and reject images they do not wish to look at. WhatamIdoing 23:35, 23 August 2011 (UTC)
You are claiming that libraries do not provide labels that readers can use to filter the contents of the library because you think the ALA prohibits it. -- I don't think I said anything like that. In particular, I have not expressed any opinion (nor do I have one) with regard to the ALA, not here, nor in any other section. So you are, in fact, not proving anything I said wrong. More importantly however, you are not even proving wrong what you think I said.
"read this because the librarian know better than you what you should read" -- You accuse the people opposed to an image filter that relies on project-wide filter categories of an overly authoritative and normative approach? Also, not having an image filter (you may now realize how I keep going back to the Wikimedia situation at hand) does not equal an advice, let alone a suggestion or order ("read this"). By contrast, suggesting images for filtering by presenting them to the user in his user interface does exactly that. So you are guilty of what you are wrongly accusing me of. Coincidence?
just like Wikipedia should allow readers to ignore and reject images they do not wish to look at. -- You are, consciously or not, presupposing that there could be no image filter variant that does not rely on special filter categories. Also, the "should" is your personal opinion. -- 23:50, 23 August 2011 (UTC)
A library could well adopt a "pornography" label as a finding aid, helping locate sexually material explicit material designed for sexual gratification for those users with that need. But when people say to public libraries that they want to avoid pornography, their objections are almost never limited to that class of works.--Trystan 23:45, 23 August 2011 (UTC)
Labels always work both ways. One patron could use a "pornography" label to find a work, and another could use the same label to avoid the work. I am certain that my local library adopted the "Young adult" label to help people find works of interest to them; I am equally certain that I use that label to avoid works not of interest to me. You should label books anyway.
It is true that most people object to many kinds of works. A person who does not want to see pornographic magazines might not want to see sexually explicit romance novels, either, or possibly any romance novels at all. I do not want to read depressing YA fiction; I also have no interest in similarly depressing works of fiction that happen to be outside of the YA section. It happens that I also have no interest horror, vampires, romance, sports, or dinosaurs. That I don't want more than one type of book does not mean that it is inappropriate for you to provide neutral, factual labels for all the books. We can provide factual information, like "pornography" or "photographs showing human genitals" without using subjective qualities like "depressing" or "sexually immoral". We have been doing this for years, after all, in categories like Commons:Category:Human penis. WhatamIdoing 17:06, 24 August 2011 (UTC)
The key distinction is that categories like Commons:Category:Human penis haven't been singled out as "potentially objectionable." No matter how factual the image filter categories are, the fact that they've been compiled on the basis of what should (and shouldn't) be deemed "objectionable" renders them inherently non-neutral. —David Levy 21:05, 24 August 2011 (UTC)
I disagree that making a maintenance-oriented list of categories that people might want to filter under a tickbox labeled something like "sex and porn" is inherently a non-neutral activity—or that NPOV applies to maintenance activities rather than to the main namespace—but let me try to focus your attention on this fact:
Commons does not have a neutrality policy. Commons rejected NPOV. NPOV is not required for WMF projects. So your argument sounds an awful lot like "But it might violate a non-existent policy", which is not a convincing argument. WhatamIdoing 17:40, 25 August 2011 (UTC)
1. Please explain how a formal determination of what content is and isn't reasonably regarded as "objectionable" is neutral.
2. Are you suggesting that the feature won't directly affect the main namespace?
3. The image filter system is to be implemented across all Wikimedia Foundation projects, including those for which NPOV is a nonnegotiable principle. 19:02, 25 August 2011 (UTC)
When categorizing images, we don't and won't decide what is "reasonably regarded as objectionable". We will instead decide (for example) if an image is reasonably regarded as a photograph of sexual intercourse, and place it (or not) in Commons:Category:Photographs of sexual intercourse based on that determination. That determination has everything to do with what our typical reader expects to find in a category with that name, and nothing to do with anyone's views on the morality of the images.
When assembling a list of categories for "Images to suppress if the person ticked the 'no sex or porn pictures, please", we will not decide whether the contents of a category are "reasonably regarded as objectionable". We will instead decide whether the category is reasonably regarded as images of "sex and porn" based on that determination. That determination will have everything to do with what our typical reader expects to have hidden (and expects not to have hidden) if he ticks that box, and nothing to do with anyone's views on the morality of what is or isn't hidden.
This process of implementing the principle of least astonishment is every bit as neutral and factual a process as the one that ALA members use when they list Hustler magazine under "Pornography -- periodicals" (which is exactly what they do with that magazine). You would expect to find Hustler in such a category; you would not expect to find The Last Temptation of Christ in that category. The fact that some people believe one (or the other, or both, or neither) is "objectionable" in some sense is completely unimportant. WhatamIdoing 21:27, 25 August 2011 (UTC)
You're missing the point. Even if the "sex and porn" filter's coverage is 100% accurate and undisputed, its creation — combined with the non-creation of a [something objectionable to someone] filter — will constitute a formal declaration that "sex and porn" is reasonably objectionable and [something objectionable to someone] isn't. —David Levy 23:18, 25 August 2011 (UTC)
"...what our typical reader expects to have hidden..." I don't know that such a creature as a typical reader exists. Personally, if I activated a nudity filter, I would expect all nudity to be filtered, including artistic, medical, educational, and pornographic. If I went to nudity-related categories, I would expect a very different set of images (those primarily or significantly about nudity.) Similarly, we could certainly create a filter for works about pornography, but I think that would be a rather surprising scope for most people who use it. In the library setting, complaints about pornography are almost never about anything I would classify as pornography.--Trystan 02:31, 26 August 2011 (UTC)

Can filtering be done in a way that minimizes the prejudicial effects of warning labels?[edit]

I've been re-reading the Harris Report to highlight areas that I agree with and pin down the basis of disagreements. The report talks about balancing intellectual openness with other goals; which I think is a fair point, and I could agree with it if the language was strengthened. I also retain my inherent dubiousness towards labeling based on its prejudicial power. However, I think I could support a filtering system that identifies 5-10 controversial areas for filtration if it was founded on the following principles:

  1. We acknowledge that warning labels prejudice users against certain classes of content, and are therefore an infringement on intellectual freedom.[6]
  2. In a few extreme cases, this infringement on intellectual freedom is justified, in order to give users control over what images they see, where an objectively definable set of images can be shown to cause significant distress for many users.
  3. The scope of a warning label must be clearly and explicitly defined based on image content.
  4. In order to be a reasonable infringement of intellectual freedom, warning labels must be minimally discriminatory.
    1. They may not have the express purpose of filtering of any group of people identifiable on the basis of race, national or ethnic origin, colour, religion, sex, age, mental or physical disability, sexual orientation, or other personal characteristic.
    2. Where disproportionate filtering of an identifiable group is likely to result from the implementation of a label, the scope of the label must be crafted to minimize this.
  5. We acknowledge that any system of warning labels will be inherently non-neutral and arbitrary, reflecting majority values while being over-inclusive and under-inclusive for others, as individuals have widely different expectations as to which, if any, groups of images should be filtered, and what images would fall within each group.
  6. We acknowledge that introducing warning labels, despite being for the express purpose of allowing personal choice, empowers third-party censors to make use of them.
  7. Categories are not warning labels. Because the task of labeling images that contain any controversial content is fundamentally different from the classification process of describing what an image is about, the warning label scheme used for filtration will be kept separate from the category scheme used to organize and describe images for retrieval.

I think the above principles could lead to the establishment of workable warning labels. It acknowledges this as an intellectual freedom limitation, which places the discourse with the right area of caution, and acknowledges the cost of every label we add. It also seeks to minimize the worst effects of warning labels in terms of prejudicing the user, namely, targeting or implicitly disadvantaging certain classes of people.

So what labels could potentially meet this criteria? Well, I think the following might:

  1. Nudity, including any depictions of buttocks and genitalia. If we include nipples, we include both male and female (i.e. Providing a filter that applies to Topfreedom and not to Barechested would not be minimally discriminatory.) We also would not distinguish between artistic, educational, or pornogrpahic depcitions of nudity, as such distinctions are not objectively definable.
  2. Wounds, medical procedures, and dead bodies.
  3. Sacred religious depictions, such as depictions of Mohammed or temple garments. But not including practices of one group of people which another group feels to be blasphemous, sacrilegious or otherwise offensive.

The major con of the above principles is that they will lead to categories which are perhaps not the best match we could develop to meet user expectations (e.g. a lot of people would probably prefer to filter female nipples but not male nipples, or entire minority topics like homosexuality.) This is by design, as it flows inherently from valuing the principles of objectivity and minimal discrimination above user expectation. It's also likely to be not all that culturally neutral (Though I confess that I don't understand how any set of warning labels could be neutral at all.)-Trystan 23:31, 26 August 2011 (UTC)

How do you reconcile "never on the basis of physical disability" with "it's okay to suppress images of wounds"? Do the victims of acid-throwing attacks or landmines not count as having a physical disability? WhatamIdoing 17:40, 27 August 2011 (UTC)
That's exactly the sort of discussion that I'd like to see the community address. My thinking of "wounds" was fresh wounds, like battlefield images, as opposed to a filter that hides images of amputees or burn victims. I have real difficulty when we start identifying classes of people as controversial.--Trystan 18:14, 27 August 2011 (UTC)
The unpleasant truth is that some diseases and disabilities really are incredibly disfiguring. There are regular complaints about the images at en:Smallpox. It was a horrifying, disfiguring, disgusting, deadly disease, and the pictures of victims are seriously disturbing to some people, just like the real thing was often seriously disturbing to both its victims and their caregivers. We want to educate, but we have people who are so upset that they close the page to get away from those images, and that means we aren't educating those people. (The images don't happen to bother me.)
I've read that the same was true in real life. Badly marked survivors were discriminated against. They had trouble making friends, forming relationships, and getting jobs. People who caught sight of them frequently reacted with disgust. (Human reactions to skin diseases may have evolutionary roots: our disgust might protect us from contagious diseases.[7])
So, yes, they're real people, and they deserve respect. But, yes, other people, who are equally real but unfortunately squeamish, may be unable to tolerate some of these disgusting images long enough to read the article. I don't believe that we will find an easy solution to this problem. Since we have announced that we don't want special categories solely for filtration (e.g., "Category:Images that are so disgusting they should be filtered") this may be one of those subject in which the proposed filter do far less than what some users want. WhatamIdoing 21:24, 29 August 2011 (UTC)
Please consider how well the alternative implementation discussed here would work in such a context. A reader with the filter enabled would visit the article and not encounter any disturbing images without specifically opting to load them. As in the category-based setup, he/she would have the ability to view only some of the images (depending on their descriptions). —David Levy 04:12, 30 August 2011 (UTC)
Especially in this case it would hide the knowledge from the user. Yes it doesn't look nice, but we want to educate. This images are a big part of it. They allow the reader to get a good understanding which crucial effects this disease had/has. I would find it unacceptable to hide the images in this context, out of respect to the people that fought against it. --Niabot 05:55, 30 August 2011 (UTC)
WhatamIdoing makes a good point about readers avoiding these articles entirely. Surely, we'd prefer that they avail themselves of the prose.
I oppose a setup in which we define such images (or any others) as "objectionable," but one in which all images are treated identically (and individual users are empowered to decide for themselves) seems appropriate. —David Levy 06:51, 30 August 2011 (UTC)
How would they know what to expect? They read the description, read the text and have no clue what they will be facing. Out of curiosity they will open such an image, "be shocked" and learning that they shouldn't open images, even so most of them (the other images hidden) wouldn't disturb them. Regarding this article i have to say: Anyone that truly wants to understand the article has look at the images. Otherwise i doubt that he had an interest for this topic to begin with. --Niabot 06:57, 30 August 2011 (UTC)
I agree that sighted persons can best gain an understanding of the subject by availing themselves of both the text and the images (however uncomfortable this might make them), but I disagree with the idea that we should seek to impose this as a mandatory condition (already a technical impossibility). I also disagree with your "doubt that [someone blocking the images] had an interest for this topic to begin with," which projects your thought process onto others. —David Levy 07:44, 30 August 2011 (UTC)
Would you really study medicine if you can't see blood? You would either understand only the half of it, or you should not study it. Learning only half the truth is mostly worse then to never hear about it. --Niabot 08:46, 30 August 2011 (UTC)
Obviously, a person unable to stand the sight of blood cannot realistically enter a profession requiring him/her to routinely encounter blood. The same isn't true of someone who merely wishes to read about such a subject. Relevant imagery can aid in one's understanding, but your assertion that its omission results in "learning only half the truth" simply doesn't make sense. I wonder whether you'd apply this bleak assessment to a blind reader's experience. —David Levy 16:56, 30 August 2011 (UTC)
A blind reader will always have this problem, that he can't actually see it and has to rely on sources that go in the very detail. He needs a description by someone who expresses his feelings about such an topic. Something we can't do, due to principles of an encyclopedia: Short to the point, neutral judgement, no emotions, ...
That we lack such detailed descriptions is true for many articles, and can be only improved with better more detailed articles. A new task like sorting images after new categories will bind manpower and would have the opposite effect. It doesn't help a blind man or woman. Instead the new buttons/links could confuse screen readers. Another issue.
You should simply believe me, that someone that has an detailed interest in such an article and would also be able to accept the images. It is as it is, its part of the knowledge. --Niabot 17:14, 30 August 2011 (UTC)
1. I strongly oppose the introduction of a filter system based upon "sorting images after new categories." I support the alternative implementation discussed here, which would enable a reader to block all images and unblock whichever ones he/she wished to view.
2. I agree that it's very important that the interface not interfere with screen readers (or otherwise reduce accessibility). Presumably, the additional buttons/links won't appear unless the user enables them (which a blind person obviously wouldn't do).
3. Again, you're applying your thought process to others. It isn't our place to pass judgement on readers. I personally believe that it's best to view the images, but people have the right to decide not to. In such a circumstance, reading the available prose is far better than avoiding the article completely. —David Levy 18:00, 30 August 2011 (UTC)
I agree to the first two points and i partially agree with point three. That someone is able to hide images manually isn't a problem. That he can hide all images as default is also no problem. The problem arises if we start to select the pictures for others, that are hidden by default. This results in non neutral judgment (everyone has at least some other opinions), discrimination of the content (majority wins), manipulation (majority is widely spread, some minority groups can push content in/out [local majority]), endless debates (wasted time) and many more problems (accessibility could be one of it).
You said, that we should give people the freedom to hide what they don't want to see. I say, that we should not decide for people what they shouldn't see. That is not their freedom; that is our personal judgment about what is acceptable for the public and what not. --Niabot 18:47, 30 August 2011 (UTC)
I agree 100% with all of the above. I strongly oppose any implementation requiring the community to take an active role in determining what content is "potentially objectionable." —David Levy 19:07, 30 August 2011 (UTC)
I guess we came to the agreement that
  • a feature to personally hide/show images is acceptable,
  • a feature to hide all images as the default is acceptable,
  • a feature with predefined categories is unacceptable,
if we want to keep our goals intact. We came to this conclusion after a relatively short discussion. The ALA did so, sixty years ago. They still think that this is the foundation for free knowledge, that should be kept under any circumstances, if possible. Wikipedia was founded under the same premise 10 years ago.
Now we may ask:
  • Why wasn't the Foundation able to figure this out for itself?
  • Why is a report written by an non expert and his family the valid and only source for such a critical project?
  • Why it was decided to implement a filter, even before asking publicly for other opinions?
  • Why must such an critical decision be the first test case for a "global" poll?
What makes me most curious at the moment is how they want to come to a conclusion after the poll. Will the results be split by language or does the English speaking majority count? It's important, since one of the goals is: "cultural neutrality". --Niabot 19:46, 30 August 2011 (UTC)
Niabot, you say that people with "detailed interest" in smallpox won't be put off by the images. Shall we educate only those people with a "detailed interest"? Should only medical professionals learn about it? Personally, I would rather educate every single reader, including those squeamish people who currently close the page out of sincere disgust. WhatamIdoing 19:54, 30 August 2011 (UTC)
If they close the page out of sincere disgust, then they already have seen enough. Reading a good written article should cause the same effect, otherwise it isn't a good article. Blind man example: You can't change the attitude by hiding facts without making it a lie.
PS: I did not say, that hiding images it out of question. But we shouldn't be the chosen ones, that decide what to hide and what not. --Niabot 20:54, 30 August 2011 (UTC)
No, that's not enough information. Closing the page because you are sincerely disgusted by the photo at the top of the article, which is a close-up, full-color head shot of the last-ever person to catch smallpox, does not teach you, for example, that the disease was completely eradicated through vaccination. It does not teach you that people died from the disease. It does not teach you that the scars are permanent. That disgust only teaches you that at some point in the course of the disease, it looks disgusting.
Different people react differently to words and to images. Some people can see images without any effect, but are disgusted by the description. More people accept the description without a murmur and are disgusted by the pictures. Most people don't mind either very much. We need to educate all of these people, not just some of them. WhatamIdoing 16:30, 31 August 2011 (UTC)
Then give them the option to hide any image or no image, but don't play the judge for what might disturb them. It's a very simple solution for the problem. We ensure that the feature is neutral in any way, that it can't be exploited and would have the effect you desire. Any problem with that? --Niabot 16:37, 31 August 2011 (UTC)
That feature already exists.
Users have repeatedly said that the existing image-blocking feature is not good enough.
Users say specifically that it is not good enough precisely because they want someone else to make a guess at what is likely to disturb them, and to filter out such images—and only such images, not 100% of images—before they have the opportunity to be disgusted by them. WhatamIdoing 16:43, 31 August 2011 (UTC)
"It doesn't exist", because most users don't know this feature hidden deep inside the configuration of their browsers.
That is your personal opinion on this matter. There is no given source that proves your claims at this point. But it can be proven that letting others judge for yourself will harm you at the end. The history books are full with this insight and this story repeats itself again and again. Today it is our turn to make or to make not the same error again. Turning on a filter is not the same as to decide which content will be filtered. Filtering for others and not for yourself is not an option.
Additionally you ignored so far any problems that will arise with category based filtering. It's already mentioned some paragraphs above. So i won't repeat myself at this point. --Niabot 17:04, 31 August 2011 (UTC)
No, the existing image-blocking options do not require the user to do anything "deep inside the configuration of their browsers". There are several purely on-wiki options. (Please go read en:Wikipedia:Options to not see an image instead of guessing what the existing options are.) WhatamIdoing 19:30, 1 September 2011 (UTC)
  • How many percent of the readers actually find this help page or are searching for it: < 0,0001%?
  • How many people find the option to hide images inside their browser: 5%?
  • How many of the people that found the function actually use it to block images: 0,1%?
  • How many browsers allow to block images only from a single domain without plugins: 0?
  • How many people that found a way to block all images from Wikipedia are complaining? 0,00000001?
I know this page and i know that it is visited by 12 People a day in average [8]. A big number right? How many of the 2.200 daily Visitors [9] of Futanari complain about the image? < 1 in a month?
The numbers look a bit funny, right? --Niabot 22:01, 1 September 2011 (UTC)
The setup discussed here (which would be far easier to implement than one based on categories/community judgement) would provide a solution significantly more robust than simply disabling/enabling images via one's browser. I don't know how advanced the various add-on scripts are, but very few users even know that such a thing exists (let alone how to use one). So it's unreasonable to equate this with anything available now.
I don't doubt that some readers would prefer to automatically filter only the images that they deem "objectionable" (and many probably have similar feelings regarding text). For the reasons discussed, this is neither feasible nor compatible with most Wikimedia Foundation projects' core principles. —David Levy 19:48, 31 August 2011 (UTC)

Categories for muslims[edit]

my proposition[edit]

  • photos and images (and videos' and gif/flash/silverlight/html5 animations' thumbnails and ascii art) of women for men
  1. do not hide any women
  2. hide all women except women covered at least from knee to shoulder and neck
  3. hide all women except women who has only face or hands or feet open
  4. hide all women, including images where their part of body is visible

explanation of 1st 2 degrees: it can be considered/counted by somebody that first glance is allowed and then that images should be hidden manually. and may be general naked men filters can be used instead of them, and this degrees may be removed from islamic filter levels.

explanation of children images: women and men mean all women and men including childs and babies.

explanation of image where it is not clear whether it is woman or man - they should be rounded to the direction to be hidden.

some people may think it is allowed for men to look at little girls who are not in "hijab" ("hijab" means only face, hands, feet visible) but "moderately" clothed, i mean, closed from knees to neck and shoulders, including knees and shoulders, or some people may even think looking at more naked girl children is ok. i have found that that idea is based on a weak hadith, and i do not want to make the categorisation more complex. and that hadith even do not say anything about little girls. probably, if it is true hadith, it just means someyhing like that children should not be scolded much...

explanation of being covered: clothing should cover form of body part as cloth can do, clothing should not stay too close to body, if it is so, let it be considered as uncovered by this my degrees, for example, arm with skinny sleeve should be accounted as uncovered arm; and clothing should not be transparent, for example, arm should not be visible through clothing.

  • photos and images of men for women
  1. do not hide any men
  2. hide all men except men covered at least from knee to navel
  3. hide all men except men covered at least from knee to elbows and neck
  4. hide all men, including images where their part of body is visible

(see explanations in the section of images of women for men) explanation of levels with covering more than from knee to elbow: it is not said clearly in quran, to what women should not look, so, that levels are also possible.

  • photos and images of men for men
  1. do not hide any men
  2. hide all men except men covered at least from knee to navel

(see explanations in the section of images of women for men)

  • photos and images of women for women
  1. do not hide any women
  2. hide all women except women covered at least from knee to navel
  3. hide all women except women covered at least from knee to shoulder and neck

(see explanations in the section of images of women for men)

  • photos and images of thigh of any men/women or animal
  1. do not hide any thigh (except if they are hidden already by other filters)
  2. hide thighs of any animals and/including people, not including thighs of insects

# hide thighs of any animals and/including people, also thighs of insects, also pistils and stamens ie reproductive organs of plants, also reproductive organs of other living creatures

explanation of level of hiding reproductive organs of all living creatures. i think hadith about not looking to thigh also shyly refers to reproductive organs. and hadith allowing to look at thigh of locust is a separate hadith, so, if it is not true, then such policy is also possible. i deleted it, but now i bring that back and strike that out. strong argument to not use that level: if that be so, there would be hadith about that. though, i am not sure, maybe in time and place of muhammad there just were not culture of giving flowers nor looking at them, nor artificial growing of them, nor was photos, and so there was no enough need to say about that.

  • photos and images of any animals (and people)
  1. do not hide any animal (nor people)
  2. hide animals that are big, clear visible, some criterion should be selected, for example, size in any direction more than 10 pixels
  3. hide all images where is any animal or people or their parts are visible

--Qdinar (talk) 11:15, 10 January 2015 (UTC)