About Bill

I was born with a degenerative visual impairment and a coordination disorder. I have never been able to read standard print. In first grade I started typing my schoolwork on an electric typewriter because my handwriting was never legible.

My dad was a mathematics professor at the University of Georgia. When I was seven he arranged for one of his former students to be my programming tutor, and my parents got me a Commodore 64. We studied BASIC and Logo. My dad told me that computers would eventually be powerful enough to remove my barriers. He wanted me to understand the tools well enough to build those solutions for myself rather than having to depend on other people to build them.

In fourth grade he told me about research into talking computers and said that I would eventually have a computer powerful enough to read my books to me. I remember thinking then that I wanted to be the one to build that.

When I was a student at UGA, Recording for the Blind and Dyslexic sent me many of my textbooks on tape. This was an outstanding service, but they did not have all of my books. My mom recorded many books for me herself, all the way through graduate school at UGA. I learned that good readers do not read figure captions and move on. They describe images and photos with as much relevant detail as they can. When reading a table, they describe rows, columns, and cells based on information and not just position. That practice is the direct origin of what I now call aggressively accessible design.

At UGA I was a math major planning a research career. In 1999 I tried to build a pipeline to scan my math textbooks and have the computer read them to me. The technology was not there, but the problem was genuinely interesting as a research problem. I started studying linguistics. As a graduate student in computer science at UGA I wrote a programmable text-to-speech engine from scratch in Prolog, which I called Cynthia. That work earned me a Gates Cambridge Scholarship for an M.Phil. in theoretical linguistics at Cambridge, focused on acoustic phonetics.

At Cambridge I found that natural-sounding speech was not the central problem. In my first year of the Ph.D. program in computer science, I needed to read widely across the literature to find an unsolved problem worth working on. I was scanning papers and listening to them from the beginning, with no way to know whether a paper was relevant without hearing all of it. What I was missing was the ability to skim. I started studying linguistic patterns in text so I could build a system to scan a library of papers and produce topical outlines, letting me identify relevant work without listening to everything. The system was called Skimcast. It turned out that building it required solving a problem that had not been solved: skimming text without a dictionary. That became my Ph.D. in computer science at Cambridge.

When I joined the UGA faculty I was building slides for my first lecture on finite state machines. After an hour I had one slide, drawn by pointing and clicking on things I could not reliably see. Existing tools were not built for how I needed to work, so I built Grafstate. It is a mathematical language and interactive platform that I have designed and developed since 2013. Students write their work in Grafstate notation, which is the same formal mathematical notation used in the course, and the platform renders diagrams, simulates computation, and shows exactly how a machine processes input, including all the branching paths that nondeterminism can take. Students who construct a working Grafstate simulation have not just answered a question; they have produced executable evidence that their mathematical object behaves as claimed. It is the primary instructional tool in Theory of Computing at UGA.

Assistivox is where all of it leads. It is a spoken-language-first accessibility platform built on the principle that genuine access means delivering the information, not a physical inventory of where content sits on a page. Every barrier in this story produced a tool that has helped me: Cynthia, Skimcast, Grafstate, Assistivox. My dad’s instinct was right.