Notably Inaccessible -- Data Driven Understanding of Jupyter Notebook Inaccessibility

Jupyter notebooks are versatile tools that open up rich storytelling possibilities by providing the means to interleave code, data and text together. Notebooks are widely adopted in a variety of expert domains such as data science, machine learning, and astronomy making the ability to create and consume them a critical requirement to participate in these professional domains. Notebooks and notebook software (tools and infrastructure to create and share notebooks) however are inaccessible resulting in barriers preventing blind or visually impaired (BVI) experts from participating in these professions.

In this talk, I will draw from our systematic large scale analysis on 100000 Jupyter notebooks and provide a broad understanding of why these notebooks are inaccessible, and exactly how inaccessible these notebooks really are. I will close with recommendations to infrastructures that drive the notebook pipeline and authoring practices to create and share truly accessible notebooks. It is imperative that we contribute to accessible notebook artifacts and software to make high-skilled work accessible!

About Venkatesh Potluri

Venkatesh is a graduating PhD candidate at the University of Washington. He understands accessibility barriers experienced by blind or visually impaired developers participating in high-skilled programming domains such as data science, physical computing, and user interface design. His research efforts contribute new interaction techniques to accessible programming, and real-world systems that improve developer tools.