ORAgen Fables: Advancing the Design and Management of Content Attribution

Paper Title

ORAgen Fables: Advancing the Design and Management of Content Attribution

Publication Info

• Topic area: Attribution and provenance in digital content creation and sharing.
• Keywords: Attribution, digital content, generative AI, media provenance, tokenised licensing, ORA framework, remix culture, dynamic attribution, relational attribution, user experience.

Background and Problem

• Problem / challenge: Attribution in digital content sharing is complicated by remix culture, generative AI, and the lack of systems to manage dynamic, relational, and granular attribution. Current approaches often fail to address the needs of everyday users and the complexities of modern content creation and reuse.
• Significance: Proper attribution is critical for acknowledging creators, establishing reputation, and enabling fair compensation, especially in the context of AI-generated content and widespread remixing on social media.
• Motivation and related work: Prior research has explored attribution in remix communities, copyright challenges, and the impact of generative AI. However, gaps remain in understanding how everyday users perceive and manage attribution, and how emerging technologies like C2PA and tokenised licensing can address these challenges.

Solution

• Proposed approach: ORAgen Fables, a collaborative storytelling tool, demonstrates the potential of the ORA framework, which combines media provenance (C2PA) and tokenised licensing to enable nuanced attribution management.
• Novelty:
  1. Introduces a spectrum-based, relational, and dynamic approach to attribution.
  2. Explores how everyday users perceive and interact with attribution systems.
  3. Demonstrates the application of media provenance and tokenised licensing in a public, interactive setting.
• Procedure and key techniques:
  • Designed and deployed ORAgen Fables, where users contribute to collaborative stories and apply bespoke licenses to their contributions.
  • Used C2PA to embed metadata for tracking content provenance and tokenised licensing for managing reuse conditions.
  • Conducted iterative deployments and follow-up interviews to gather insights on user attitudes toward attribution.

Results

• Concrete findings:
  • Attribution is perceived as a spectrum, influenced by the degree of contribution, medium, and dissemination context.
  • Users value attribution for its relational and future-oriented potential but also express concerns about privacy, stress, and unwanted recognition.
  • Dynamic attribution systems are needed to accommodate changing user preferences and contexts over time.
• Advantage over baselines:
  • ORAgen Fables demonstrates how media provenance and tokenised licensing can provide granular, relational, and dynamic attribution, addressing gaps in existing systems.
  • Offers a user-centered perspective on attribution, moving beyond professional contexts to include everyday content creators.
• Experiments / evaluation:
  • Four public deployments in Edinburgh, UK, with 128 contributions and 12 follow-up interviews.
  • Deployments included both professional storytellers and generative AI to create final stories, enabling comparisons of attribution preferences.
• Limitations and future work:
  • Limited participant diversity due to the study’s location in an affluent city.
  • Future work could explore more automated attribution features and refine the interaction design to better demonstrate tokenisation and licensing.

Summary

This paper explores the complexities of attribution in digital content creation and sharing, focusing on everyday users and mundane content. Through the ORAgen Fables tool, it demonstrates how media provenance (C2PA) and tokenised licensing can enable nuanced attribution that is spectrum-based, relational, and dynamic. Findings highlight the need for systems that accommodate user preferences, privacy concerns, and the evolving nature of content reuse. The research offers design opportunities for creating more equitable and user-centered attribution systems for the future internet.