They never sleep, and you never know when they might turn and bite you. Machines Like Sharks explores alternative means for browsing the vast body of information lurking on peoples' hard drives and sloshing around the Internet. We treat this data as a huge repository of found objects and continuous nocturnal murmurings. These prototypes use photos from personal web sites, ad banners, porn animations, copy from annual reports and private confessionals and whatever else the machines can be coerced into divulging to create dynamic collages of images and text. The results are pieces which are constantly on the edge of emergence.

This work has its practical underbelly as well. All of the ingredients of the collages retain links back to their source files, whether local or long-distance. By clicking on them you get taken to the original text or web page from whence they came. The user can also parameterize the search paths in order to mine a particular sector of the information landscape.

Future experiments will plunder sound and image files, make use of semantic parsing of text streams and generally be more clever and better behaved.