Mockup derived from image by Graphictwister on Freepik
Known and unknown accessibility issues within the award-winning Jupyter notebook block disabled students, educators, scientists, data scientists, journalists, and others from tools that define whole contemporary tech fields. To better understand the specifics of notebook issues disabled users have and ways we might fix them, Space Telescope Science Institute (STScI), Iota School, Quansight Labs, and the Jupyter community joined forces to run cycles of user experience research with the support of disabled scientists and technologists, document discoveries and potential solutions, and produce proof-of-concept code improvements to the notebook to pave the path for fixes.
With a user base in the millions, the impact of the code, narrative, and multimedia collage that makes up the Jupyter notebook is major. Notebooks For All is working towards a future where Jupyter software empowers disabled people in their education and careers—an experience greater than plain legal conformance.
I led the accessibility-focused usability testing efforts with Jenn Kotler. We collaborated with other team members Tony Fast, Patrick Smyth, and Erik Tollerud throughout for technical feasibility of tests, reporting results, evaluating fixes, and curating event content. All of this work is public and open source, just like the notebook file itself.
You can watch me discuss our recommendations for notebook file authors (with no captions yet, my apologies) at SciPy 2023 or review any of our other media mentions on the project’s media page.
Skills: User experience research, accessibility testing, usability testing, semi-structured interviews, slide design, writing, public speaking, grant writing
Motivation
Notebooks are experientially known to lack accessibility support. Exactly how and why the issues behaved as well as what the disabled scientists impacted would prefer, however, was not as clear. This systematic exclusion, like many others, has ripple effects on the lives of disabled people and the communities (and opportunities) they are excluded from.
Like many other institutions, STScI uses notebooks. In fact, they use notebooks quite a lot. From behind-the-scenes calculations and in-progress efforts, to project documentation and tutorials, to resources that showcase the fruits of space exploration, it’s easy to spend days lost in the notebook files STScI produces. As a part of their mission to the public, the team at STScI wanted to understand and improve the accessibility of a file type they rely on while supporting the open source ecosystem it hails from.
Strategy
Focusing on a concrete and actionable strategy to this effort, we decided to prioritize the following.
• Scope our investigation to the notebook reading experience. Users must read before they can edit, so we considered the reading experience a blocker to further notebook use.
• Test STScI notebooks with disabled scientists and technologists who use it, want to use it, or work with equivalent tools. Gather feedback on the current state of the notebook and on their expectations are of a properly accessible interface.
• Make code fixes, even if as prototypes, to the notebook and export templates. Fixes build on the feedback from tests and provide us further opportunity to test in increments.
• Contribute wherever we can. While the whole project was open and public, we also gave our work back to the community to support the efforts from the source.
• Share it out. Have public events for input and education.
• Do it again! This strategy was intentionally designed to be cyclical so that we could compound the accumulated knowledge of disabled scientists’ lived experience, user research, code development, and open-source community we organized.
• Test STScI notebooks with disabled scientists and technologists who use it, want to use it, or work with equivalent tools. Gather feedback on the current state of the notebook and on their expectations are of a properly accessible interface.
• Make code fixes, even if as prototypes, to the notebook and export templates. Fixes build on the feedback from tests and provide us further opportunity to test in increments.
• Contribute wherever we can. While the whole project was open and public, we also gave our work back to the community to support the efforts from the source.
• Share it out. Have public events for input and education.
• Do it again! This strategy was intentionally designed to be cyclical so that we could compound the accumulated knowledge of disabled scientists’ lived experience, user research, code development, and open-source community we organized.
Notebooks for All followed a pattern of overlapping cycles that fueled one another. The accessibility-focused usability testing identified and documented the software’s strengths, weaknesses, and potential for exemplary accessibility support. User research findings were adopted by the development team to explore references fixes to be tested in future cycles. The user research and development findings also culminated in two interactive events. While different people led each of the efforts, each of us worked on all aspects of the project throughout.
This cyclical project style allowed each team to prioritize their own work while directly supporting one another.
Methods
On the user experience research side, I was responsible for the following.
• Designed, scoped, recruited, and led tens of user testing sessions from the ground up. Sessions combined usability testing and semi-structured interviews in STScI notebooks. They were designed to emphasize accessibility feedback and complete common Jupyter notebook tasks
• Authored and presented our public issue tracking and aggregate results written per testing cycle.
• Led internal team and open-to-the public discussions and summaries of results throughout project.
• Contributed testing materials and results upstream to community-maintained projects for discoverability and sustainability.
• Presented work and solicited community feedback at Notebooks for All and broader community events like conferences.
• Reviewed development directions and fixes per cycle based on testing.
• Designed, scoped, recruited, and led tens of user testing sessions from the ground up. Sessions combined usability testing and semi-structured interviews in STScI notebooks. They were designed to emphasize accessibility feedback and complete common Jupyter notebook tasks
• Authored and presented our public issue tracking and aggregate results written per testing cycle.
• Led internal team and open-to-the public discussions and summaries of results throughout project.
• Contributed testing materials and results upstream to community-maintained projects for discoverability and sustainability.
• Presented work and solicited community feedback at Notebooks for All and broader community events like conferences.
• Reviewed development directions and fixes per cycle based on testing.
Work in use
Work from this project can be found collected in the Notebooks for All project repository.
You may also find work from this project in the
• Jupyter Surveys project (the project-wide collection for user research)
• The Jupyter accessibility meeting notes
• The Notebooks for All virtual event and STScI Day of Accessibility
• JupyterCon 2023
• SciPy 2023
• Jupyter Surveys project (the project-wide collection for user research)
• The Jupyter accessibility meeting notes
• The Notebooks for All virtual event and STScI Day of Accessibility
• JupyterCon 2023
• SciPy 2023
