What happens if you mix the geo-data embedded in photos with some data about where our listed buildings are in London?
Higher graded buildings were more likely to have photographs taken near them: 88% of Grade Is had at least one photograph falling within 25m of their centre (as defined by the coordinates given in their list description) as opposed to 76% of Grade II*s and 61% of Grade IIs).
The average number of photographs for those that had photographs within this distance was also highest for Grade Is (168 photographs on average), followed by Grade II*s (58) and Grade IIs (42).
Some interesting work from John Davies at Nesta, using the Flickr API. Just proving what's intuiting, perhaps, but a nice illustration of how you can use a big public dataset like Flickr to test assumptions.