#amazonfail and metadata

April 14, 2009

As a web developer I’ve been keenly following what has been termed #amazonfail using twitter parlance.  After giving it some thought I feel its instructive to talk about what happened as it directly affects photographers who use digital media.

The core issue behind Amazon Fail is that a large number of items in the amazon.com catalog were removed from their product ranking system and partially obscured from their search engine.  These products appeared to have all been tagged using amazon’s internal metadata as belonging to any number of queer categories.  These works were marked in the system as “adult” which removed them from amazon’s ranking system (which tracks the popularity of any title) and partially obscures them from amazon’s search engine.   Products associated with gay and lesbian catagories were being blacklisted by amazon’s own internal mechanisms.   These books included Virgina Woolf’s Orlando (oddly enough not all editions) and a popular piece of non-fiction on gays serving in the military.   The blacklisting was limited to marking queer content catagories as adult themed works such as Playboy collections or the biography of porn star Ron Jeremy were not removed from rankings as they lacked any gay or lesbian catagory tags.

(NB: I tend to use the term queer as an inclusive term as Gay/Lesbian/Bisexual/Transgendered/Intrasexual is a bit of a mouthful.)

Without getting into the back and forth speculation of how this situation came to be, especially as amazon.com has yet to release an official statement, let’s talk about metadata.

Without getting into a technical discussion metadata is like sticking a post-it on a piece of information so you have an idea of what it is.   This post has a title.  It has tags and a catagory associated with it.  These pieces of metadata help describe what this specific piece of content is.  Its extraordinarily difficult to parse meaning out of human language algorithmically so we have to rely on this additional information to inform any interested parties what it is I’m talking about.  I have metadata associated with the blog in general as well.  It helps search engines and other automated systems describe what I’m talking about here so that people can find this blog if it meets their needs.

Now let’s consider a photograph

Woman on Motorcycle, NYC

This is a photograph of a woman on a motorcycle.  It was taken on 5th Ave around 50-something street facing North using a Mamiya RB67 using a 90mm K/L lens on Fuji Acros film rated at 200 ISO which was developed using Diafine.  This is all valid metadata which I can attach to the photo along with my name, contact information, and that I retain all copyright on the image.   This gives you a better appreciation for the photograph and its one of the advantages of making a digital copy of your photos.   Its quite a bit of information to write down on the back of a print so its pretty handy to scan in your images, tag them up as a group and be able to come back and refresh your memory.   If you’re shooting digital most of this has already been done for you as the camera will record a large amount of exposure information onto each image for you.

Now let’s say that this photo is of a woman who’s a member of Dykes On Bikes and it was taken at the 2008 Queer Pride Parade in New York City.

If this was a book Amazon would have removed it from the site rankings as I’ve just used metadata to associated the photo with a queer cultural and political event.  As useful as the exposure information is I consider it secondary to providing context for the photo.   Despite what people may say pictures do not speak for themselves.  Without providing the information that this woman is participating in a queer pride event there’s no sense of place or purpose.  It would just be a photograph of a woman on a Harley with some dude behind her.   In five or fifty years all cultural based context clues would be lost without providing an explict account of what the photograph is.

It could be argued that by using metadata we are opening our work up to being swept up in any spot of trouble like the one that occured at amazon where some person or persons decided to inhibit the ability of the general public to find and purchase material related to queer issues.   If I were to limit my tagging of photos to only the name, rank, and serial number so to speak I could avoid being swept up in any momentary political or social kerfuffle.  My work could continue to be viewed and (in theory) purchased.

I feel that we have a responsibility to describe our work as fully and accurately as possible.  We cannot be responsible for how people react to that information or what they choose to do with it but without it the work looses its context and the reason it was taken in the first place.  I went out to the pride parade to take pictures of a queer cultural event.

Photographs don’t only exist to be considered for their compositional elements.  They can form records of events and what was happening with our culture.  This is true of paintings as well although we tend to think of them as pure aestetic objects.  Looking at a portrait of Louis XIV without understanding the symbolism of elements and objects placed in the painting misses the point of the portait.  Louis XIV very consciously embarked on a public relations campaign which included portraiture, sculture, and other media to establish his political authority.  You can certainly look at his portraits in terms of light, color, and composition but without reference to his symbolism the reason the portrait was made in the first place would be lost.

In the end we can’t be responsible for how our information is used.  We can only be responsible for its accuracy and completeness.  Without that information we wouldn’t have positive use for the material.  It would just be a snapshot.

Posted in News