Brian Pytlik Zillig was trying to get rid of a rubber stamp.
Nearly 30 years ago, while managing the University Libraries’ Interlibrary Loan, Pytlik Zillig spent part of his day processing invoices the old-fashioned way — armed with a large rubber stamp, ink pad and hope that the result would be legible.
“Our way of sending out invoices involved using this great big rubber stamp. It had payment information and little tick boxes, and we would very carefully ink this up on an ink pad, and then we would very carefully place it on top of our copy of the ILL request, press down, and hope it was legible once we lifted up the stamp,” Pytlik Zillig said.
He wasn’t looking to change the future of digital humanities. He just wanted to make billing faster.
“I did not have a grand idea. I was only trying to make billing faster,” Pytlik Zillig said.
Instead, the experiment introduced him to a technology that would shape the next three decades of his career.
The technology was an early version of Extensible Stylesheet Language Transformations, or XSLT. Pytlik Zillig discovered that he could use it to transform invoice information from the computer system into formatted files that staff could print.
Within hours, a tedious manual task had become an experiment in what computers could do.
“I had this kind of profound and abiding insight about what computation could be, and how much it could help an otherwise sort of boring, soul-destroying task,” Pytlik Zillig said.
That insight would become the foundation of a career spent experimenting with technology and finding new ways to connect scholars with texts.
In the 1990s, libraries were beginning to replace paper records with computerized systems. The technology could store information electronically, but compared with today’s systems, its capabilities were limited.
As Pytlik Zillig managed inter-library loans, libraries and universities were experimenting with ways to turn text and image files into websites that could be used by students and researchers. XML-based technologies, including the Text Encoding Initiative standards and XSLT, were emerging as important tools for organizing and displaying digital texts.
University Libraries needed someone who understood how those technologies worked.
When a faculty position for a digital initiatives librarian opened in 2000, Pytlik Zillig applied and was hired.
His job placed him at the intersection of scholarship and emerging technology. He helped establish standards and practices for building digital humanities projects, creating new ways for researchers to access, examine and interpret texts.
But the work was still largely experimental.
XSLT was in its Working Draft stage, meaning there was no mature ecosystem of tutorials, tools and online communities to provide ready-made solutions. Much of what Pytlik Zillig needed to accomplish had to be figured out from scratch.
“It was a very modest sort of thing. We wanted to take the texts that we had, and my task was to find a way to make these visible using XSLT,” he said.
He describes the process less as hacking than as experimentation.
“Some people call it hacking. It’s kind of like tinkering in your garage, only with code. I don’t really call myself a hacker anymore. I call myself a committed experimenter.”
That willingness to experiment became particularly important when he took on one of his most challenging projects: the Journals of the Lewis and Clark Expedition.
The University of Nebraska Press wanted the digital edition to reproduce the visual appearance of the published journals as closely as possible. That required Pytlik Zillig to find a way to transform complex tables from TEI markup into a format that could be displayed on the web.
“The University of Nebraska Press wanted something a little beyond what we knew how to do technologically. They wanted to recreate the visual layout of the published journals as closely as possible,” Pytlik Zillig said.
At the time, many of the tools needed to accomplish that kind of work were not yet commonplace.
“This was before Stack Overflow and other websites that would help you, give code examples, and things like that,” Pytlik Zillig said. “It all sounds like exactly the kind of common things that you see everywhere on the web today, but we had to figure it out.”
As the technology matured, Pytlik Zillig began looking beyond simply displaying digital texts. He began exploring what computers could reveal about them.
He developed and published TokenX, software that allowed researchers to analyze TEI documents and extract information from large collections of text.
“You could upload a TEI document into TokenX and gather all kinds of data about it,” he said.
Researchers could use the software to create concordances and examine word frequencies, recurring themes and patterns that might otherwise be difficult to identify.
Pytlik Zillig then pushed the idea further with Indigo, a data visualization tool that transformed text into animation.
“What I started doing was creating these short animations — I think the longest one I’ve done is about 12 minutes — and submitting these to various digital humanities conferences and things like that,” he said.
Indigo sits somewhere between data analysis and art. The software uses XSLT to transform text into visual animations, allowing researchers and audiences to experience information in a different way.
“It’s fun to talk about because it really straddles a line between what is art that you experience visually and what is art that you experience intellectually,” Pytlik Zillig said.
Now, nearly three decades after that rubber stamp first sent him down this path, Pytlik Zillig is experimenting with a technology that would have been difficult to imagine when he began: artificial intelligence.
One of his latest projects involves developing a Retrieval-Augmented Generation, or RAG, index for the Willa Cather Archive in collaboration with Emily Rau, assistant professor of digital humanities and editor of the Willa Cather Archive.
The goal is to create an AI-powered search tool that can understand relationships and concepts within texts rather than simply matching the words a researcher enters.
“If you wanted to do a search involving trees, it was good to know the names of all the trees. But in this new tool that we’re working on now, the RAG index tool needs to have an understanding of where trees are referred to, even if they’re not called any of the names that you think they are,” Pytlik Zillig said.
That represents a fundamental shift in how researchers might interact with digital collections.
“So, I think in terms of technology, we are really pushing against what users know and what users can expect,” he said.
The technology has changed dramatically since Pytlik Zillig first experimented with XSLT. The question driving his work, however, has remained remarkably consistent: What new questions become possible when technology gives people a new way to interact with information?
For nearly three decades, Pytlik Zillig has explored that question through digital humanities projects that have changed how students, scholars and researchers access and analyze literary, historical and cultural texts.
For him, the point has never been technology for its own sake.
“I like experimenting. I like creating new forms of access that allow scholars, students and researchers to ask new questions. People always want to ask new questions. I mean, I know I do.”