What Counts? Narrating the Backstories of Open Datasets


Every dataset has a backstory. What ultimately comes to “count” in a dataset often depends on a series of judgments and assumptions held by data creators, the configuration of infrastructures for sorting observations, and political struggles over what should be included or excluded. To interpret the values recorded in datasets responsibly, it is important to understand the social and political contexts that gave rise to their production. This workshop will provide a framework for narrating dataset backstories - studying ‘what counts’ by analyzing the historical underpinnings of open datasets.

The session will open by way of an example - with a presentation of a backstory of NYC’s 311 dataset. From here, the facilitators will lead an interactive session introducing a methodology for piecing together various components of a dataset backstory. Following this presentation, in breakouts, groups will have an opportunity to apply these techniques to narrate “what counts” in a series of NYC open datasets.

The session will emphasize the importance of rich metadata practices and provide resources to recent literature on best practices for data documentation. It will also provide insights as to how biases emerge in datasets and how they can be accounted for in data practice.