I’ve always admired John Watson’s ever growing menagerie of Flickr hacks. Having recently revisited the site, I enjoyed his overview of what happens over time to one day page of Flickr’s Explore. A given day page is not frozen in time for Explore’s purposes. The only thing that’s fixed is the population that can appear on that day page: photos shot on that day. Otherwise, only the Flickr boffins know what determines which photos are displayed at any given moment.
Once upon a time I used to be really interested in Flickr’s interestingness algorithm. I was also curious about the statistics and properties of the photos that did have high interestingness. The Flickr API has to be one of the nicer web service api’s out there and there was a nice Python library to access the service. I implemented some code for decent sized longitudinal studies similar to Watson’s captures.
Looking at the video Watson created, the thought came to me that there’s actually a lot more potential for using those interesting photos as a gateway into Flickr. As his automatically built movie demonstrates, one can easily build new media artifacts with those photos as a seed. Attached to each of those photos is a Flickr user, an entry point into the Flickr social network. I’m going to start Flickr hacking again to see what I can come up with.