Venn Metatopia

Stumbled across an interesting article in this months' Wired - about the importance and difficulty of adding sufficient human readable metadata to digital photos.

A few things strike me about this. First, we're already dealing with this problem in the analogue world. My family is full of really rather good amateur photographers. I've inherited boxes and boxes of slides from aunts, grandmother and parents - and they're full of pictures about which I know no contextual information at all.

One of the projects I must do before my great aunt dies (the last of her generation in my family - currently somewhere in her high nineties, and daft as a brush, in a lovely way) I need to sit with her and do some archiving work: look at objects and photographs in her house, and preserve their stories and histories. For instance, I have an old copy of the Book of Mormon on my shelf - it belonged to my great grandmother, who nearly converted to Mormonism around the turn of the century. She was a nurse, on night duties, and used to fill her time by making the most beautiful, detailed copies of religious postcards in a huge sketchbook. Now, think of the stories behind that - and think of what will be lost from my family's oral history if no-one bothers to record this before she goes...

I sort of want to have some kind of automatic metadata generator for _people_.

Secondly... And back to the photos after that digression; the article only very tangentially mentions Flickr, which I think is a revolutionary service. Why? Because playing with their service, and the way they present 'tags' makes it very quickly clear what the benefits of tagging your pictures can be. OK, tagging is a classification system that is personal, and not based on any kind of agreed vocab... So it might not translate. But when you get a sufficient volume of people adding a few simple tags to photos, you get much more meaningful data.

Add to that tagging the ability for anyone to go and add further tags to refine the classification of a picture according to their personal system, and what you get is effectively a subjective map of the position of the picture in a semantic framework. I think this gets round a whole number of problems one sees with databases, metadata and so on - it allows everyone to sort it in their personalised way, but doesn't favour one system over another, so you get a broad spectrum of meaning around the object being classified. It's the kind of thing that would send a librarian completely potty. But here's why this doesn't matter.

The example given in the wired article, of looking for president Coolidge and some Indians, involves hunting through index cards, then walking into another room, and physically sorting through files. This is the bit that is eliminated by a digital system, and it's also the section of the work that requires absolute pinpoint accuracy - when you're doing that amount of physical labour, you maybe want to check 3 or 4 images before you get your positive 'hit'. But think about using google's image search - often based on the scantiest metadata available. You'll happily flick through four or five pages of 50 images to find the one you're after - because the work just involves clicking a mouse.

There is more room for ambiguity in digital classification systems.

The recursive nature of search - refine- search refine is closer to the way information is organised inside your head - it doesn't have a good analogue in the analog world.

Finally, the article doesn't mention the ESP game - a neat way of using bored five minutes, and people's competitive spirit - to add tags to images. Treat adding metadata as something that is of obvious use, or is a fun activity, and people will be much more inclined to do it. Couple this way of capturing the 'subjective' data with the diary synchronisation, GPS and EXIF data mentioned in the article, and that feeling of bewilderment at a box of physical objects with no context outside the life of the (mortal) creator will be a thing of the past.