User:Jason Quinn/NPOV is a problem for images

Photos that stray far from neutrality and violate our neutrality policy are becoming a problem at Wikipedia. Such photos can be misleading, distort reality, or promote a non-neutral visual message. As editors, we need to learn to identify such photos and eliminate them. My experience is that editors are severely under-prepared to tackle this issue. Editors tend to just "want photos" and that bias often trumps good judgement over a photo's inclusion in an article, even in the most egregious cases of deliberately propagandist photos.

Photographs greatly increase the value of an article on Wikipedia and are crucial to its success. No amount of text, however well-written, can precisely describe a person's face, a landscape, the subtle aspects of a salamander's body, or the texture of some tropical tree's bark. Photos convey in an instant such intrinsically visual information. The old saying "A photo is worth a thousand words" is an understatement. Yet, photos are never really "neutral". For most photos this concern is neglibible and they are worth including in our articles. For example, typical biases in photos are merely related to photographic aesthetics and technical aspects of the photo. These things can be, but usually aren't, a problem.

The problem is at its worst when photos are used to send a political message or for self-promotion. Wikipedia is no longer a fringe website. Governments, political organizations, and public relation companies are now editing Wikipedia to change and form public opinion. Given the power of photographs, replacing unflattering photos with flattering ones (or vice versa) is becoming a tactic they are employing.

How do we go about spotting and stopping this? Well, it has to start with recognition of the problem itself which many editors do not see.

Gentle explanation
Every photographer knows that a photograph captures a subject in a particular way and at a particular time. The variables (intentionally set or not) that went into producing a photograph end up portraying the subject in a particular manner. To human interpretation, a particular photograph can be "good" or "bad" based on a very small differences.

Most familiar to people is how we take multiple photos until we get "one we like". A professional photographer's career is usually based upon a talent for achieving such "good" photos. Sometimes it is the photographer who is the judge of what constitutes "good". And sometimes it is the subject (for example when the subject selects which portraits to purchase from a photographer). And, of course, with "selfies" we get to be judge as both photographer and subject!

Here are some things to consider about the technical aspects of getting a "good" photo:


 * Portrait photographers choose angles and lighting that flatter the subject and a background that suits the subject and their clothes while capturing a pleasing expression on the subject's face.
 * A nature photographer often chooses the best-looking specimens to photograph while avoiding ugly ones such as birds with missing feathers or mammals with fur problems. They may angle their shot to avoid man-made objects to evoke a more natural environment. They often blur the background to isolate the subject.
 * Sport photographers use huge telephoto lenses because they cannot be close to the subject and because they must stop the action. Very often a good photograph means to capture a precise moment or a look that belies emotions players felt during the game.

Photography is often practiced as an aesthetic art. Many photographers aim to achieve a distinctive "look and feel" to their work that differentiates them from previous photographers. For example, a photographer may decide to specialize in taking long-exposure nighttime scenes of the homeless or children playing games using polarized filters. The actual possibilities are infinite which is partly why photography has found such broad appeal.

No matter what the reason why somebody is taking photographs, photographs are loaded with bias in how they were taken and how they were chosen. This is at odds with an encyclopedia's goal of presenting things factually and "as they are". This idea has received virtually no discussion and has been overlooked by editors. On the occasions where I have mentioned it, my point was either ignored or not understood.

This point becomes particularly important when there are monetary or political reasons for presenting subjects in a particular way.

Choice-of-subject bias
Bias in a photo does not require any manipulation of the photo and may actually capture a real subject in a real situation. For example American colleges often want to produce brochures that celebrate the diversity of their campus. This diversity may be rather limited. Sometimes the photographer will simply ask a select group of diverse students to pose together for a photo-shoot for the brochure pictures. That's clearly a biased view. But a photographer could also actively seek-out and find a diverse group of students to picture, passing hundreds of non-diverse groups along the way. Either way the finished brochure misrepresents the diversity of the campus.

Distortion and perspective bias
All photographs have distortion in them. This can be used to great effect to make things look bigger or smaller than they really are. In movies and television and media, sometimes blocks are used to make actors or politicians appear to be the same height. Or a photo might be taken to emphasis the height difference to disparage somebody.

Big Mac/Whopper case study
Consider for a moment the articles about McDonald's Big Mac and Burger King's Whopper. Certainly adding a picture of these items to each article would be a good idea. How should this be done? Probably, if asked, each company would be willing to provide an image for the article with appropriate licensing. But the mouth-watering and delicious-looking photograph they deliver would not really reflect accurately on the product received by most patrons. (Those of us old enough may remember a false advertising lawsuit in the late 80's against McDonald's because of this exact disparity between the ads and the actual sandwich. It resulted in McDonald's adding preservatives to the buns that helps keep them from looking soggy.) It is simple truth that advertisers want to show a subject so it appeals to potential customers. The hamburger shown in marketing photographs will have been very carefully specially prepared to be a prime specimen. The lettuce will be crisp. The sesame seeds will be ideally placed (perhaps literally with tweezers). The buns will have no flaws. The meat will look thick and juicy (probably "squished" so the meat on the camera-facing gives the impression the whole patty is thicker than it is). The sandwich will be lit with professional lighting. The composition and angle of the shot will be carefully considered. Every controllable aspect will be designed to influence you, the potential customer, into wanting to buy and eat the portrayed sandwich. (That is, assuming that the Big Mac or Whooper shown in the marketer's photograph is real, edible food. Food photography often involves the substitution of real items with mimics, such as Crisco for ice cream and so on. Trickery and fakery in food marketing is a large, fascinating &mdash; and sometimes revolting &mdash; subject itself.)

I think most of us have the common sense to know that using corporately-produced photos on articles is generally a bad idea. The scenario above is a situation where bias has run amok and it is instructional for that reason. Unfortunately, the issue is not solved by simply allowing editors without a personal interest to take the photo instead of the company.

Suppose now that some helpful editor decides to take a photo of a Big Mac or Whopper to add to their respective article. They order the sandwich and just photograph it, right? Wrong. A lot of choices go into even the simplest photograph by even a novice photographer. The equipment itself may introduce a bias. If the technical aspects of the image, such as resolution, are of too low quality, it may reflect poorly on the subject. Then there are many variables to the picture such as angle and magnification. Why and how were they chosen? Almost always the answer is that the angle was used to image a side of the sandwich that looks the most photogenic, clearly a non-objective bias. What about the ambient lighting, exposure, and white balance? These can easily present the subject "better than it is" or "worse than it is" according to human aesthetic judgments. Which do you think would look more appetizing, a sandwich brightly lit so that the colors on the bright red tomatoes and bright green lettuce "pop" relative to the beef, or a poorly lit sandwich whose image has a grayish cast to it? What about background? Was the sandwich set aside and imaged alone or presented along with the tray and napkins? Why? Was a softbox used? Was the background digitally edited? For good or for bad, these could be viewed as manipulation of the subject. And then there are the statistical questions: was this particular sandwich typical or atypical? Our hypothetical Wiki-photographer might even photograph several sandwiches, perhaps even from different restaurants, to see. Once the final set of images is made, clearly some thought will be put into choosing the "best" one. How will that be done? Will "best" be interpreted as the "most documentary" or will it be "best by the human eye's aesthetic judgment"? Almost always people operate by the latter and this "pursuit of a good picture" ends up biasing the subject.

Conflict of interests
Our Wiki-photographer, who may have even strove for neutrality, encountered a number of challenging issues relating to photographing even a sandwich in an unbiased, neutral manner. The Wikipedia faces an even more insidious problem: editors with a conflict of interest. Wikipedia's success has meant that some articles are viewed by companies as potential vehicles to advertise their products. The high number of "copy-n-paste" edits of PR (public relations) text into Wikipedia articles attests to that. A new industry called reputation management has grown around managing a company's image on the Internet. There is a clear financial motivation for photographs to portray their products in a positive light and their competitors' products in a negative way. An intelligent marketer with a little bit of know-how of Wikipedia's functioning could easily game the system and successfully pass intentionally non-neutral, biased images off as "just another editor's upload". A clever marketer of this type would attempt to use the best images they could without making their intentions obvious. The copy-n-paste edits of PR material are easily noticed but it is unreasonable to suspect that all marketers who edit Wikipedia take such a simplistic, easily-detected approach. One reputation management firm discusses their approach as trying to "emphasizing positive content" while "correcting mistakes" and "soft[ing] or remov[ing] negative content", all of which could mean by providing, removing, or editing photographs.

Portraits
Portraits offer a wide avenue of ways to create a non-neural image. This usually falls into two categories: portraying the subject in a more negative way or portraying the subject in a more positive way.

Making the subject look worse
Fox News has a history of airing altered photos (see Fox News controversies). Particularity egregious examples of negatively edited photos were aired by Fox News in 2008. As shown on the right, these photos were digitally edited to portray subjects negatively by exaggerating features or introducing repugnant aesthetic aspects, likely while catering to anti-Semitic stereotypes.

Clearly such photos should be unacceptable for use on Wikipedia. Photos with this kind of manipulation must be rejected by our content policies and there should be strong repercussions for editors who upload such images. The danger here is that editor groupthink will decide otherwise. For example, for a subject with no alternative free image, it is easy to imagine that editors might decide to keep such a distorted image under the idea that some image is better than no image. This line of thinking must be rejected.

That was the most obvious kind of bias in a photo. But a photo does not have to be altered to have a negative biased image. Using unedited but poor photos may fall under bad faith, or at least unwise. For example, images captured at an awkward or embarrassing instant can be quite unfair to the subject if used, regardless of the presence or absence of malice by the photographer. Assuming good faith, if no other uploaded photos exist, it must be asked if such a photo makes sense to add to the article.

Making the subject look better
Portraits are another area where professionalism leads to biased results. A good portrait, by definition, is one that flatters a person, not one that gives a candid impression. Portrait photographers know how to convey a particular mood, personality, or attitude onto the subject through posing, backdrops, lighting, and so on. (See music video Sparkles and Wine by Nacho Guzman.) As many shots as necessary may be taken until their vision is captured. If it takes 25 or 50 shots to get an acceptable headshot, how is that not an example of a biased representation? Should professional "headshots" of actresses, politicians, and CEOs even be used on Wikipedia when sometimes these headshots distort the subject so much that the real person is not even recognizable, as many casting directors and modeling agencies could surely attest. Wikipedia's articles on actors, singers, models, and politicians should avoid overly professional photos intended to give the best impression possible of the subject. It is a growing aspect of public relations work that agents offer to edit Wikipedia pages to a client's advantage. This is unacceptable. An aggressive stance must be taken against users suspected of editing Wikipedia in a professional capacity.

Using "too good a photo" adds bias. Similar to how using an image captured at an awkward moment that make the subject look bad is unfair, using rare images that make the subject look good is unfair. For example, if it takes a person 100 shots to get one "good one", then that one shot is non-representative of the subject and falls into a "too good" category. If other photos do exist, editors are in the position of asking what photos are better but don't cross over into the "too good" territory. Nobody said that being neutral is simple!

Cityscapes, landmarks, etc.
Cityscapes, landmarks, and tourist attractions are also subject to bias. It is clearly noticed in Wikipedia that article on these items tend to have main photos that were taken on cloudless sunny days even when such conditions are highly atypical for the location.

Beijing, for example, which is one of the cities with the worst air pollution in the world, currently has as its main image a montage of famous sites on clear, smog-free days. Which is less biased for a smog-ridden city, the photo with or without the smog?

Landmarks are often surrounded by many tourists. A photographer may wait a long time to get a tourist-free shot. If a photographer had to wait an hour for a five second interval with no tourists in the field-of-view, does that tourist-free image give a proper impression of the landmark? If they didn't get a tourist free shot, is it okay to digitally remove them?

Certainly travel agents, airlines, and other ticket vendors would prefer locations shown uncrowded and with clear blue skies but the purpose of Wikipedia isn't to advertise locations but discuss them objectively.

Editing and manipulation
On top of all these concerns, there's an even more difficult problem which is now ubiquitous: photo editing. The digital age has made manipulation of photographs easier than ever before. Virtually all photos are manipulated prior to being published. This can be as simple as just cropping the image, changing the brightness, sharpening or softening. It can be as subtle and technically challenging as color balancing. Or it can be as overt as complete and total manipulation wherein objects are added or erased entirely. There were several high-profile instances of this during the second Iraq War where extra smoke was added to images or extra rockets being launched. Clearly such techniques have been and will be used for propaganda purposes. Editing can turn truth into lies and lies into "truth".

Wikipedia must draw up guidelines to determine how much editing a photograph should have. This will be very difficult. Even for the "simple" act of cropping, the crop may alter the context of the photograph completely. Access to the camera original image and transparency about what editing has occurred to the original helps solve some of this problem. No obvious solution will likely be found to this issue except perhaps by identifying a group of individuals who have good judgment determining when photographs have been "manipulated too much".

The best solution is to come up with a list of things that are explicitly allowed and a list of things that are explicitly not allowed. For example, the follow should generally be permissible:

All these are permissible so long as they are not relevant to the context of the photo. On the other hand, the following things should generally not be allowed:

These changes alter the subject in a non-objective or non-factual way. Wikipedia is an encyclopedia. It is meant to inform, not to be used for marketing, managing public relations, or as a propaganda outlet.

Just as with the textual NPOV guidelines, there will never be a perfect (or even consistent set of rules) to determine what is kosher and what isn't. It must be realized that this task will always require common sense and editorial judgement that considers notions like balance, impartial tone, undue weight, and so on.

Conclusion
Given all these concerns, what should be done? Wikipedia needs to discuss what constitutes a neutral photograph to protect the content of Wikipedia from editors with an agenda and to maintain verisimilitude. The neutral point of view policy needs to discuss this aspect in more detail, perhaps with a new sub-policy. Just as with the textual policies at Neutral point of view, it is unreasonable to expect that any policy for images will be perfect but it can give useful advice on what constitutes a neutral photograph. Of course, the policy will be enforced with common sense as with all policies and guidelines. Existing standards of photojournalism ethics, such as those by the National Press Photographers Association, should be studied as a model.

The nature of this problem is complex and should be handled in stages. The "post-image" digital manipulation of photographs is an easier topic to handle than the "pre-image" aesthetics choices and should be the first area to have guidelines.