Data Epics
Sept 2019–Ongoing
In the Data Epics, fiction writers use data from home IoT devices to create short stories for their home dwellers to read. The project investigates how home dwellers might encounter and make sense of their home IoT data and these short stories capitalize on the ways people understand complex situations and concepts through narrative structures. We argue that for the IoT to actualize its promise of moving dwellings closer to a ‘smart home’, it is imperative that home dwellers become empowered to understand what data is collected as part of IoT services, what assumptions can be inferred from the data and the implications of these assumptions.
Collaborators:
Audrey Desjardins
Heidi Biggs
Gabrielle Benabdallah
Stephanie Tang Waldrop
Data Epics takes a radically different approach to presenting home data: by turning data into short fiction stories created by fiction writers. It is the start of a journey into more nuanced, messy, lively and situated ways of seeing data.
Process
Step 1. A smart home device collects data over a month. This data may be simple on/off from a smart plug, or the commands and questions asked to a voice assistant, for example.
Step 2. Data are downloaded from smart device systems. For example, here we see data from a voice assistant. The log includes the utterance said, a timestamp, the source, the name of the associated audio file, etc.
Step 3. The data are cleaned to keep only the actions or phrases, and timestamps, removing other meta data. For example, here are data from a voice assistant. We see how the participant asked the voice assistant to translate phrases between English and French, they asked the temperature in Paris, change the volume, and set and stopped a timer.
Step 4. If the data are numerical, we create a visual graph of the data over the period of a month. Here, thermostat data shows temperature ranging from 64 F to 84 F during the month of March.
Step 5. The data are sent to a fiction writer, who reads and interprets the data to write a short fiction story. The writer can choose to integrate the data however they want in the story. With voice data, they may include them verbatim. With numerical data, they may look for trends or outliers as inspiring narrative points.
Step 6. We typeset the story into a small booklet.
Step 5. We print, fold and hand bind the booklets.
Step 8. We send the booklet back to the household who provided the data in Step 1. In our second study, we are working with 5 writers and 5 households, creating 5 stories per volume.
Publications
Desjardins, A., Biggs, Heidi. (2021). Data Epics: Embarking on Literary Journeys of Home Internet of Things Data. CHI'21 New York, ACM Press, in press. (acceptance rate: 23.8%)
Next Project
Soft Fading