There's an inherent contradiction in the intentions of this project. On one hand, we want to break down stereotypes and outdated notions of representation, to be more inclusive and less judgmental. On the other hand, entourage people are inherently stereotypical because we're forced to judge them solely on appearance. This gives rise to the first problem we've encountered: do you overtag or undertag?
Tagging too much can start to feel pretty uncomfortable. How fine grained are the distinctions? "Asian" or "Chinese"? What about "Laotian"? Can you be sure based on one image? Even if you're right, will other people know? What would you search for? Should you tag someone as "fat" or "plus-sized"? Or is there an even better term we can use? Can we start shaping people's judgments based on the tags we're using and the search terms we're popularizing?
The alternative is to tag too little, perhaps only by activity. This is probably the most equitable solution for the future, but it also perpetuates the latent racism of those who claim to be race blind. Somehow when no one thinks about race in renderings, everyone winds up white. And if you're doing a project in, say, the South Bronx, which is 60% Latino and 39% Black, the exclusion of people of color from project renderings amounts to projections of mass displacement and gentrification. Whether or not you think that's likely, do you think it's right?
For NONSCANDINAVIA we're tentatively going with the approach of overtagging, because we believe these issues have to be faced consciously and head on. The goal is essentially to tag people in a hierarchy of specificity, beginning with the most general and honing in with as much detail as we know to be true. Here's our basic tag template:
- Activity: Walking, Sitting, Talking...
- Mood: Happy, Angry, Sad...
- Accessories: Phone, Bag, Shovel...
- Gender: Woman, Man, Non-Binary...
- Race/Ethnicity: Asian, Indonesian, Balinese...
- Age: Child, Middle Age, Older...
- Season: Summer, Winter...
- Miscellaneous: Anything else...
Some of these things are are visually apparent in a given image, and will be tagged accordingly. Others - like nationality, religion, or sexuality - aren't always apparent, and are often prone to conjecture. We'll only be tagging these non-visual traits based on known information, such as narrative text that accompanies certain images in their original context, known individuals, etc. The issue really is how specific do we want to get, and how comfortable are we resorting to appearance-based stereotypes as metrics for classification?
We don't have all the answers, but our general approach is to tag what seems relevant using terms that are respectful. The tag template is an attempt to break down those criteria in a reproducible way that will create a searchable archive. The tag cloud page lets people see all of the terms we're using, sized based on frequency of use, which gives us a good way of maintaining balance and transparency.
This process is ongoing and open-source, so if you've got ideas, let's hear them. Comment below to make it public or email nonscandinavia@gmail.com to express your opinion privately.