Entropy Bound

December 2018 | Developer

Entropy Bound is a Google-supported Future Storytelling research project written by Jeff Burke and Jared J. Stein which weaves together dramaturgy and code.  After suffering a head injury, a woman (Bek) who is no longer capable of forming memories begins using an AI brain implant device. The story documents the evolution of the relationship between implant and implantee as Bek simultaneously trains and learns from her device.  

The main actress wears a POV camera which is the input for various machine learning processes, used to identify faces and props.  Annotations are stored in a repo which serve as the system memory- the ‘implant’ can then recall past scenes that match what it is currently observing.  The main actress can also comment on what is happening to her using emoji.  These are also stored for potential future recall.

Major responsibilities and contributions throughout the workshop include: management of wearables for the main actress, design and implementation of overhead projections (using TouchDesigner) which provided (1) dynamic & interactive lines for improvisational rehearsal, and (2) live system feedback data concerning object/prop/face recognition & scene segmentation, a mobile web interface for the main actress to provide live reactions to her surroundings, and general tech provisioning/maintenance.

Outside of the workshop, I prototyped an Add-In for Microsoft Word to establish bidirectional communication with an external database which contains show-specific lines from the script. The play’s script is in constant motion, with improvisation and change performing key roles. This tool can be used across rehearsals/shows/performances to enable playwrights to properly write for the show.