We discussed the year in review to some extend but looked more to the future. We decided that we will start next year with a dialog around Knowledge Objects for XR (copy paste!), for a lack of a better term.
AI: Summary
This session focused on the final stages of publishing the Future Text book while grappling with its fundamental unresolved question: what should a knowledge object look like in spatial environments? Dene Grigar discussed editing work including grammar standardization, metadata enrichment, and accessibility compliance, while the group debated whether WebXR, Unity, or Unreal Engine should be used for creating the VR version of the publication. The conversation revealed a tension between open web standards and proprietary engines, exploring how to preserve digital literature in immersive spaces while maintaining discoverability and indexability. The group also discussed AI’s role in knowledge preservation, force graph visualizations for navigating virtual worlds, and the concept of timestamped LLMs for historical context.
AI: Insights
The group identified a critical architectural tension between the privacy-preserving design philosophy of WebXR and the need for collaborative, discoverable spatial knowledge systems. Brandel Zachernuk articulated that WebXR reduces markup and browser functionality to a generic runtime, making it impossible to index spatial pages the way traditional web pages are indexed, because everything is generated by JavaScript rather than being declarative. This creates an irreconcilable conflict with the web’s success model based on discoverable information rather than executable programs.
Ken Perlin and Brandel Zachernuk found common ground that while WebXR is imperfect, it represents a viable path toward spatial web standards, especially compared to proprietary game engines. This suggests a pragmatic approach where immediate needs can be met with current tools while working toward better long-term solutions.
Peter Wasilko articulated frustration with paternalistic privacy systems that prevent users from making their own informed choices about data sharing, highlighting a broader tension between protection and agency in spatial computing environments.
Dene Grigar’s vision for the Electronic Literature Lab museum revealed an interesting hybrid approach: maintaining the museum in WebXR for accessibility and preservation while creating specialized Unity versions for Apple Vision Pro to enable features like metahuman avatars of authors. This suggests that different platforms may serve different purposes rather than requiring a single universal solution.
The concept of “timestamped LLMs” emerged as a fascinating insight about AI and historical context. As LLMs evolve, there may be value in preserving not just documents but the AI interpretations of those documents from specific time periods, creating a kind of meta-archive of machine understanding alongside human knowledge.
The discussion about force graphs and semantic knowledge representation revealed that spatial navigation systems require careful consideration of what creates meaningful “clumps” of information. Peter Dimitrios noted that semantic weights matter more than traditional PageRank link-based rankings, suggesting XR requires fundamentally different organizational principles than flat web architectures.
Tom Haymes observed that “humans are only capable of lossy compression” and “it gets more lossy as you get older,” which sparked reflection on how both human memory and AI systems compress and lose information over time. This connects to the larger question of how knowledge objects should be structured to maintain fidelity across transformations and platforms.
The phrase “XR Markdown” appeared in the chat as a potential conceptual framework, suggesting the need for a lightweight, declarative format for spatial knowledge objects that mirrors Markdown’s success in making structured text accessible and portable.
AI: Resources Mentioned
Virtual Worlds Museum Teleportal force graph visualization: https://virtualworlds.museum/teleportal
Mentioned by Evo Heyning, built by Steven Van Loon and Julian Reyes
Brandel Zachernuk’s timeline VR experiment: https://zachernuk.neocities.org/timeline-vr/
Brandel Zachernuk’s related video: https://www.youtube.com/watch?v=DeKdDatIItU
Arrival.space web tool for testing splats in games and educational experiences: https://arrival.space/realitycraft
Mentioned by Evo Heyning
Evo Heyning’s website: https://evo.ist
Richard Holton’s Figurski at Findhorn on Acid hypertext novel
Referenced by Dene Grigar as being preserved in the Electronic Literature Lab
Story Space proprietary software
Original platform for Richard Holton’s 2001 hypertext work
Supernatural VR boxing and kickboxing application
Mentioned by Dene Grigar as an example of interactive avatars
WebXR, Unity, Unreal Engine, Godot
Game engines and platforms discussed throughout
Apple Vision Pro and Quest headsets
Mentioned as target platforms
Song
This track is an AI orchestrated piece inspired by the transcript of this meeting, meant as a fun provocation to further thought. (suno.com)