I was playing around with Matlab thinking about vector clustering algorithms over the weekend. Something out of my depth really. While thinking about complex data sets and vectors, I remembered PhotoSynth from microsoft.
Then I thought about facial biometrics which usually use a few bio points to reduce a face to a few key vectors. Companies like Animetrics do this.
That got me thinking about a social application mash-up. In the human animal the gaze is important as we are highly socially evolved. Our eye have small irises relative to the "white" of the eye compared to the Gorilla or Chimpanzee. One theory behind this adaptation is that it allows us to track the gaze or focal point of others. Think about how finally tuned we are to gauge trust based on eye tracking, contact etc.
Social gaze is our desire to watch others behaviours and in times of uncertainty to mimic it. This occurs not just during the teenage years but throughout life. Look at any market. Charlie Munger writes about this.
Social gaze is a powerful tool, it is the basis of marketing. Facebook and Twitter allow for social broadcast & gaze at an information but not physical level with people continuously answering the question, "what are you doing right now?"
One can easily imagine the following applications being developed. Variants are probably already in use at Darpa etc or being nurtured in silicon valley.
PeopleHistoryWhere: combine Photosynth for creating location tags and vecotors. Then use similar technology for clustering social vectors extracted from the images, overlap the data sets and from a large pile of public photographs, you could probably determine location, time and social network (relationships). My guess is that the vector clustering could even indicate passage of time using standardized clues to evolve crude social spatial chronology.
This would make a very interesting form of historical journalism, genealogy from public photos of who with whom, where and when. The immortal why of history would be up to the historians. Imagine getting an e-mail from a photograph asking, "what were you thinking at this moment?" The moment and actors in any space could theoretically be recreated. Disneyland pictures 2.0.
Given the human animals social instincts and relatively tight clustering behaviour you could probably infer who was holding the camera where, when and most likely during what time of day and day of year if the picture was outdoors and there are enough shadows a latitude could be approximated. This could be done using photographs of well photographed landmarks from any period. This would be very interesting for historians.
The Flitter application mentioned in the title needs less technology overhead, simply register your camera phone or account and start-uploading photos. Each time a photo is updated an approximate time could be guessed at as to when it was taken and where assuming it was outdoors or indoors near a well photographed (tagged or untagged) landmark.
Most photos of people are taken next to physical or social landmarks. A social landmark is a birthday party filled with people for example. given the small quick nearest neighbors set that a social network allows, it wouldn't be to tough to identify and tag the photo participants automatically. The software could even prompt the user for confirmation.
Upon uploading to your flitter account, the photo is automatically indexed and placed in either a physical or social context. This would most likely be a popular social gaze application building on top of existing social software and allowing for interesting auto-tagging.
Like all technologies, it is ethically agnostic. It is more a matter of when rather than if this happens in the public sphere. This technology will of course manifest itself through our cultural filters of greed, lust, fear etc. Lets hope it provides a better understanding and appreciation of our relationships with each other rather than stalking and finger pointing.