Bell Jar Legacy - Need for data through time and space
Remembering Gordon Wells: A Life Observed
A Presentation by Suzanne A Pierce
Manager and Research Scientist, Texas Advanced Computing Center
The University of Texas at Austin
Bell Jar Legacy – The Need for Data Through Time and Space
Today, I want to talk about Gordon Wells as a friend, colleague, and a fan! If you received random text messages about Japanese high school marching bands or river otters from Gordon... you were among the fortunate few who experienced one of his greatest gifts: unbridled enthusiasm for the world's delightful peculiarities.
The Kyoto Tachibana High School Marching Band
For example, Gordon was a consummate fan of Kyoto Tachibana High School's marching band—not merely because they are exceptional musicians—but also because he marveled at their multi-generational connections and their ability to execute structured, disciplined performances in unwieldy conditions.
I watched a lot of YouTube videos and participated in well narrated discussions of the band's prowess with Gordon.
Kotaro and Hana
And then there were Kotaro and Hana (also Aty & Ui). The Japanese urban otters. Two adorable creatures living in Japan whose adventures Gordon followed with the dedication most people reserve for their favorite sports teams.
I miss those random texts. I suspect many of you do too.
This wasn't mere distraction or quirky hobby collecting. Gordon's fascination with a marching band that could maintain formation through chaos, with animals whose behaviors he could observe through carefully positioned cameras, as phenomena revealed themselves through patient observation - this was Gordon being Gordon.
Because at his core, Gordon Wells was an observer. His sensors were always on.
Nature Monitoring and Professional Observation
Those sensors extended to East Texas where he shared a weekly 'javalina count' and included updates from his property where he'd installed monitoring equipment that generated additional delightful texts with live feeds of foxes, woodpeckers, robins, and other wildlife plus one adorable donkey. But these weren't just charming nature videos. They represented something fundamental about how Gordon moved through the world. He understood—intuitively and intellectually—that observation creates understanding, that persistent watching reveals patterns, that data tells stories if you're patient and disciplined enough to listen.
This same observational bent positioned Gordon uniquely between academia and applied emergency response, between theoretical understanding and providing enough information to support life-saving action.
To Gordon, the collection and analysis of data such as hurricane landfall, storm surge forecasts, and wind impact assessments during natural hazard events or rallying medical professionals statewide to share daily updates on the number of available beds and other key performance measures during the pandemic weren't abstract academic exercises. They were tools used during actual emergency activations and critical events to inform decisions while storms raged and communities hung in the balance.
Debriefs and Loyalty – The Gordon Nod
Gordon possessed what can only be described as a fierce intelligence coupled with an even fiercer loyalty. He was an effective mentor who would fight to help you improve your performance under any conditions. Much like that Japanese marching band navigating crowded streets, Gordon pushed his colleagues to maintain formation and execute with precision even when circumstances seemed impossible. He championed his people with the same intensity he brought to tracking hurricane paths.
Gordon held everyone—including himself—to extraordinarily high expectations. He believed in battle-tested methods that were hard-earned and useful, while simultaneously challenging the team to innovate and consider new ideas. This could make him... a rather demanding collaborator.
There was little more entertaining than watching a group discussion veer into territory where Gordon had strong opinions.
And there was little more sobering than seeing Gordon shake his head during a meeting, knowing you'd mis-stepped and could anticipate a "spicy debrief" afterward where he would set you straight.
Those spicy debriefs, though? They made you better. Gordon drove results beyond the bounds of our own expectations while expanding our orbit in both technical performance and social connections. The work his teams produced was work everyone could be proud of, precisely because Gordon refused to accept anything less than excellence.
The Data Bell Jar Concept
This brings me to Gordon's most enduring contribution—an idea born from his keen observer's mind in 2018. As Brent Porter recalls, "Gordon began talking about being able to look at data throughout an event—not just before, not just after, but during, as phenomena unfolded in real time."
He called this idea the Bell Jar.
A bell jar is one of science's most elegant instruments—a transparent vessel that permits continuous observation while maintaining the integrity of what's enclosed. For Gordon, this became a powerful metaphor for data collection during dynamic events.
Just as the physical bell jar creates a bounded space where phenomena can be observed without contamination, a computational bell jar establishes a structured interface between chaotic reality and systematic understanding. It represents capturing not just the event itself, but its context, its progression, its evolution as an integrated whole.
Gordon understood that during event-driven phenomena—hurricanes, wildfires, disasters that erupt suddenly and demand immediate response—traditional research methods fall catastrophically short. You can't study a tornado at your convenience. You can't ask Hurricane Harvey to pause while you perfect your protocols.
As he would say with characteristic pragmatism: "Beggars can't be choosers. Best available and timely are more important than chasing a perfect dataset that might never exist."
Bell Jar Metrics
The bell jar framework demands more than conceptual elegance—it requires rigorous implementation. Gordon and his team developed specific metrics for evaluating data within this framework:
- Completeness
- Currentness
- Constraints
- Attribution
- Accessibility
- Adaptability
These weren't abstract academic categories. They were practical tools for decision-makers who needed reliable information.
What fascinated Gordon was the temporal dimension. He wanted timestamps that were clear and specific, metadata that could track datasets step by step with precision, automated processes that could detect change throughout an event's evolution—not just at beginning and end, but continuous observation as understanding deepens and consensus forms.
The bell jar adapts while maintaining the integrity of observation.
Implementation and Legacy
Prototyping End-to-End Workflows
Today, Gordon's bell jar vision isn't just theory. The decision support office at TACC continues developing concrete implementations.
Example workflows from recent publications analyze unstructured stories or reports from the field with natural language processing and LLMs to then map the decision structure in those texts onto scientific variable objects that we can then use to automate matching with registered datasets and physics-based models. It's an early attempt to implement intelligent automation processes.
Computational Infrastructure
We're exploring how AI tools can support the analysis and curation of event-driven collections—digital twins, knowledge graphs, methods to communicate risks reflecting lived experiences from prior events.
The bell jar concept represents: - Rigorous observation - Structured thinking - Innovation in service of practical action - Holistic understanding over reductionist segmentation - Unified standards for data quality across all modalities - Enhanced reliability of analysis enabling confident conclusions
Creating Gordon's Bell Jar
Today I'd like to invite you to contribute to a memorial collection in a way Gordon would have appreciated—by creating a bell jar for Gordon.
Each of us holds observations of Gordon—time windows that overlap but remain incomplete. By bringing our observations together, we're applying Gordon's own methodology to understanding the phenomenon that was his life: - Collecting multi-modal data - Synchronizing our timestamps - Filling gaps in each other's knowledge
We're building a digital collection to preserve your stories, anecdotes, datasets, and artifacts in Gordon's honor. Near the entry to this room, you'll find physical bell jars where you can drop cards with your memories or you can email me after this event to coordinate. Your contribution—your unique window into Gordon's life—becomes part of a comprehensive understanding that no single observer could achieve alone.
Please help us build a bell jar collection that preserves the stories of collaboration and impact we all want to remember.
Conclusion
Gordon taught us to be better observers. To watch with discipline and enthusiasm. To maintain high standards while championing others. To find structure within chaos. To understand that persistent, careful attention—whether to Japanese marching bands or hurricanes impacting the Texas coast—reveals patterns that can save lives and deepen understanding.
Gordon Wells was a steward of knowledge, and a fierce champion for Texans.
His sensors were always on, always watching, always seeking to understand. And in doing so, he expanded all of our orbits, pushed us beyond our own expectations, and left us better equipped to observe and respond to a world that demands both precision and compassion.
Thank you, Gordon, for the random texts, the spicy debriefs, the impossible standards, and the elegant frameworks. Thank you for teaching us to look at data through the context of events, to capture the progression of phenomena, to build bell jars that preserve what matters most.
Your sensors may be off now, but ours remain on... sharpened by your example, still watching, still learning, still grateful.
Memorial Presentation for Gordon Wells
Texas Advanced Computing Center
Tilleggsinformasjon
| Felt | Verdi |
|---|---|
| Data sist oppdatert | 23. oktober 2025 |
| Metadata sist oppdatert | 23. oktober 2025 |
| Opprettet | 23. oktober 2025 |
| Format | |
| Lisens | License not specified |
| MINT Standard Variables | corpus_nlp |
| Has views | True |
| Id | cf84ccc8-dce8-4391-bc64-f89757f238cc |
| Mimetype | application/pdf |
| Mint standard variables | corpus_nlp |
| Package id | ff9673d1-c9af-4434-aa09-39c55e97317f |
| Position | 0 |
| Size | 2,9 MiB |
| State | active |
| Url type | upload |