Over the course of two days — Friday and Saturday, March 10 – 11, 2023 — CIVIC X Brown University will offer the Theory to Practice: Context-Aware Systems Symposium in the Patrick Ma Digital Scholarship Lab at the Rockefeller Library. Additionally, the keynote will be made available to the public via live stream. Registration is required for both (see program below for registration links).
Context-Aware Systems is an active practice that considers how infrastructure can contain and evolve situated knowledge, and seeks to more deeply understand and clarify embedded assumptions which can distort the structures, interpretations, and impacts of data.
The purpose of the Theory to Practice Symposium is to reach beyond discourse and criticism of the current landscape of data and ethics to offer tangible principles, methodologies, and frameworks while building a multidisciplinary collaboration environment for participants to experience what more equitable approaches to technology creation feels like in action.
The Symposium will span two days, comprising four in-person sessions of three hours each. Sessions combine lecture presentations and applied lab activities which ground theories of contextual technology development through curated learning examples. Examples of learning material may include case studies, real datasets, dataset imaginaries, schema samples, simulated project environment elements, and hypothetical or gamified scenarios.
No prerequisites are needed to participate in any workshop, and they are designed to create the most value for multidisciplinary faculty and graduate students interested in both conceptual and practical approaches to centering equity in data and technology initiatives.
Friday, March 10 – Register for this day [link coming soon]
- 8:30 to 9:15 a.m. – Continental Breakfast
- 9:15 – 9:30 a.m. – Welcome and Opening Remarks
- 9:30 – 10:15 a.m. – Keynote – Register for the online keynote [link coming soon]
- 10:15 – 10:30 a.m. – Break
- 10:30 a.m. -12 p.m. Session #1:
- Lecture: Architectures of Friction and Flattening
- Lab: “Les Deliverables”
- 12 – 1:30 p.m. – Lunch
- 1:30 – 3 p.m. Session #2:
- Lecture: Data Constituent Engagement
- Lab: “Data-Driven Gaslighting”
- 3 to 3:15 p.m. – Break
- 3:15-3:30 p.m. – Closing Remarks
- 5 – 7 p.m. – Dinner offsite
Saturday, March 11 – Register for this day [link coming soon]
- 9:30 to 10:15 a.m. – Continental Breakfast
- 10:15 – 10:30 a.m. – Opening Remarks
- 10:30 a.m. – 12 p.m. Session #3:
- Lecture: Beyond Performative Dashboards
- Lab: “Dashboard Glow Up”
- 12 – 1:30 p.m. – Lunch
- 1:30 – 3 p.m. Session #4:
- Lecture: Remediating Bias With Contextual Metadata
- Lab: “Hansel and Gretel Bias”
- 3 to 3:15 p.m. – Break
- 3:15-3:30 p.m. – Closing Remarks
- 3:30 – 4:30 p.m. – Closing Reception
Full descriptions of the sessions:
Session #1: Architectures of Friction and Flattening
March 10 // Friday Morning (Kick-off)
This workshop challenges assumptions of techno-solutionist narratives while presenting alternate ways of thinking about technology that embrace intersectional representation and lived experience. Shifting from the mindset that data are objective and neutral, we recognize that cultural and environmental factors can have profound impacts on how we perceive information.
Framing “context” as a learned discipline and set of core competencies connected to every stage of the data lifecycle, this session will present strategies for leaning into constraints and limitations of data and bring attention to what is lost in the gap between reality and what can be captured by information structures. We will invite participants to imagine (and experience) how processes that center equity and impact not only improve industry-standard models, but are intrinsically necessary to break barriers and achieve the next stage of modern innovation.
Interactive Lab Component: “Les Deliverables”
An experiential game which simulates the experience of building an equity-based software product. The scenarios (dramatic, thrilling, and sometimes treacherous) are based on composites of real case studies and projects, inviting players to make choices and allocate resources at pivotal moments along a technology development lifecycle while balancing conflicting priorities and emerging developments. Participants will work in teams to “win the game” by completing their project on time, on budget, with positive community impact while avoiding catastrophic mistakes that can make a deliverable fail.
The goal of the game is for participants to:
- Notice how seemingly abstract principles show up tangibly in action
- Ground an understanding of how equity and ethics are bedrock practices, not optional or last-minute add-ons
- Have fun while covering a wide variety of examples and strategies in a storytelling format
Participants who are new or skeptical to conversations of an equity-bedrock technology approach:
- Acknowledge that data is inherently limited, and understand that examining blindspots and documenting constraints makes data more valuable
- Consider how identity and representation can be fractured and flattened by datasets and underlying structures
- Perceive how purposeful or unintentional decisions within technology infrastructure can contribute to surveillance and/or conditions of selective oppression
- Critically examine the “build fast and break things” model and expand perspective on the impact of technology beyond its creator
Advanced participants with experience and expertise related to the subject areas:
- Gain new vocabulary and concepts that directly apply to their work and way of thinking
- Be able to tap into context-competencies to make strides toward equitable outcomes
- Feel validated and inspired that success is achievable with the friction required to do this work in the wild
Session #2: Data Constituent Engagement
March 10 // Friday Afternoon
This workshop expands on concepts from traditional human-centered design to include the role of “data constituents,” defined as people represented in, or impacted by datasets or their lifecycle. This approach differentiates data constituent roles and value propositions from that of the “user” or “business stakeholder” and offers new participatory strategies that strengthen the integrity of technical systems and ethical impacts.
This session presents examples and case studies which illustrate how strategic engagement with data constituents is not a marketing or PR strategy, and can measurably inform data structures, validation cycles, and interpretation of analytics.
Interactive Lab Component: “Data-Driven Gaslighting”
Participants will critically examine the human side of data pipelines and power structures, noticing who is included and excluded. Focusing on questions of information provenance, we investigate the culture of confidence in data-driven decision making with special emphasis on elusive ways that confidence may be misplaced when key constituents are left out of the process and the structures of accountability are not properly in place.
Apply data constituent strategies to a curated set of data sources with realistic scenario prompts, to be distributed in breakout groups. This workshop does not produce a real-world production strategy but offers exposure to a range of critical thinking examples.
For participants who are new to business strategy exercises and/or have had low exposure to data environments:
- Learn to identify constituents of various forms of data
- Become familiar with how to differentiate data constituents from users and stakeholders
- Develop constituent personas, with consideration toward intersectional identities and relevant relationships to power
- Facilitate reflection on the potential for harm, impact, benefits, and opportunities across individual and collective constituent types
- Drive discourse on extractive v. respectful behaviors of engagement
- Prepare for concepts of data governance and data team workflows, which will support their learning in subsequent Context-Aware Symposium sessions
For participants who have experience working with data stakeholders and are comfortable applying business strategy frameworks:
- Build upon best practices from human-centered design workflows to understand where data constituent engagement can be adapted in technology life cycles
- Gain insight on how to structure feedback from data constituents in order to improve the quality and accountability of data sources
- Consider value propositions for how investing time and resources in constituent engagement can lead to higher quality outcomes
Session #3: Beyond Performative Dashboards
March 11 // Saturday Morning
This workshop illuminates reductionist behaviors in data visualization that can unintentionally contribute to mischaracterizing analysis or reinforce marginalization. We’ll showcase why metadata is necessary as a component of information design and lean into the controversy of how the effort and investment required to create context may be inherently destabilizing to the essential nature of what people tend to want from data — predictability, certainty, and objective groundtruth.
There is a direct connection between the popularity of dashboards and a global tech economy optimized to celebrate founders and reward products which promise greater scale, decreased friction, and promote acquisition of information as a means of omniscient advantage. In this way, it’s become normalized to overprescribe data as a source of power in and of itself.
In this session, we position data visualization as the fruiting body of a larger complex entity and tap into its roots to deconstruct various decision processes influencing what gets prioritized, obscured, or made hyper-visible. Placing special emphasis on “equity” dashboards, we see how cookie-cutter practices and constraints of the genre can perpetuate harmful distortions without proactive awareness.
Interactive Lab Component: ”Dashboard Glow Up”
Participants will break out into teams and be assigned a dashboard visualization to analyze and redesign. The sample will be a real screenshot from an open data dashboard available online, printed from the internet, and enlarged onto a posterboard like a piece of art. Along with the visualization, groups will receive a package of all metadata publicly available to describe the data used in the dashboard. Following CIVIC’s Context Data Communication guide (not yet publicly available), teams will use paper, scissors, tape, and Sharpies to reimagine and reform components of the visualization interface. Importantly, they will not be able to change the data visualization itself, but they can completely alter everything around the visualization container to build clarity and reduce bias in the information display, including reauthoring labels, changing colors, creating visibility for key metadata, etc.
Teams will come together at the end to present their changes “before and after” style, and share how their decisions impact the narrative and perceptions of the data.
For participants who are new to design-strategy frameworks or have had low exposure to data communication environments:
- Recognize misleading information constructs
- Expand literacy skills to assess and understand data visualizations in the wild
- Gain experience in accessibility design practices
- Learn how the use of color, word choice, and other presentation features can significantly influence impressions of data
- Increase confidence working with metadata and begin to appreciate documentation in a new way
For participants who have high data literacy and experience building visualization frameworks:
- Permanently alter the perceptions of industry-standard data visualization dashboards
- Build competencies to recognize and avoid damaging practices that can cause harm to communities
- Integrate Context Communication methodology in their own work to improve equitable and ethical data communication
Session #4: Remediating Bias with Contextual Metadata
March 11 // Saturday Afternoon
This workshop introduces methods to uncover and critically assess inferences and unexamined judgements which can present measurable distortion within datasets. Often the conditions which create bias can be sneakily embedded, nearly invisible, and start small. Unnoticed or unaddressed, misplaced assumptions can compound into big problems — rendering your data unusable or actively inflicting harm to a constituency. Particularly when those assumptions are deeply rooted in a legacy system and amplify race, gender, or historic marginalization factors, action can feel unclear and overwhelming. While it may be difficult, it’s important to find ways to shift this issue from abstract to tangible.
In the session, we’ll crank up the magnification on metadata structures and demonstrate how it’s possible to gain traction through incremental operations. Providing an overview of the CIVIC Contextual Metadata schema, the presentation provides a granular approach to reviewing data lineage and methodology with an emphasis on discovering bias. Going further, the session will explore archivist-inspired techniques to annotate datasets to increase provenance and integrity of future use cases.
Interactive Lab Component: “Hansel and Gretel Bias”
Participants will use an abridged version of CIVIC’s Structured Context Schema (not available to the public) as they navigate a series of questions and prompts in order to author inputs for metadata fields. Working in teams, each group will receive a curated data imaginary which includes a scenario, sample data, and workshop facilitators role-playing data custodians, stakeholders, and/or constituents.
Data imaginaries are selected to highlight unique aspects of embedded bias which should be discoverable by “following breadcrumbs” through a guided investigation of the material and responses from facilitators. As teams generate documentation, they will recognize opportunities for data to be remediated through additional fields and make strategic decisions about how to handle gaps in knowledge, problematic methodology, and ethical governance questions.
For participants who have not been actively engaged in discussions of bias and/or have little hands-on experience with metadata schemas and documentation:
- Shift thinking bias from an abstract issue to a tangible workflow
- Learn an inquiry framework which can help reveal where bias may exist in data
- See how bias shows up in a variety of forms in realistic settings
- Reduce intimidation of data (especially if they don’t consider themselves a “data person”) by working with metadata as a gateway entry point
- Value collaboration with multidisciplinary team members
For participants who have are immersed in discussions of structural bias, and/or are familiar with metadata and schema documentation:
- Expand understanding of what metadata is and the simple but powerful role it can play
- Receive tactical training on CIVIC Structured Context Metadata Schema to support data provenance
- Learn practical strategies to improve data equity through structured remediation practices