W3C Workshop on Annotations: Hypermedia Notebooks and User-centered Publishing

Randall Leeds, Developer, Hypothes.is Project


The Hypothes.is Project leverages Web technologies to produce tools for authors and engaged readers. Our products enable users to capture hypermedia citations and deploy these in the production of literate Web content. These citations form the basis of conversations that posses provenance information of greater granularity and stability of reference than is typical of Web based commenting systems.

For developers, Hypothes.is produces software libraries that aim to increase interoperability of user interface components and applications dealing with references. The resulting tools provide new opportunities for to author richly linked documents and data. Through this lens we view our work as pertaining to annotation, or the management and production of reference metadata.

As part of our broad mission to annotate the world’s knowledge, we have organized a yearly conference, I Annotate, which brings together stakeholders from industry and academia in order to collaborate on future directions for research and development of related tools and publishing best practices. We bring to the workshop our perspective, as practitioners actively engaged in the production of Open Source Web applications and scripting libraries, on the evolving conversation about the future of Web annotations.


The Web provides unprecedented capabilities for publishing by offering world wide distribution of multimedia content. Hypermedia, which enables linking together distributed content, allows for navigation between related material not possible with physical media. However, as the Web Platform has evolved to accommodate the needs of publishers, the casual user has been left behind, lacking convenient tools for authoring linked documents. Browsers often provide only the ability to bookmark documents. Meanwhile, support for highlights and marginalia is becoming increasingly common in e-reader devices, yet interoperability between systems is nonexistent. The common unit of reference, the Universal Resource Indicator (URI) [1] , is limited to identifiers chosen by publishers, leaving the question of how to describe arbitrary references, created post publication, unanswered. Efforts to standardize new syntax for URI fragments, such as the Media Fragments URI specification[2] , cannot possibly keep pace with the diversity of domain-specific requirements.

Recently, browser vendors have begun to incorporate features related to identity, social networking and data synchronization. It is increasingly common for users to sign in to their browser in order to gain access to cloud storage and cross-device synchronization of bookmarks, history and other user data associated with a persistent social identity. Some vendors simultaneously offer document editing and publishing, yet these activities are not fully integrated into the browsing experience. Significant value can be captured by bringing reading and writing activities closer together. The browser of the future should function as a social notebook, allowing users to capture references to media of interest, author their own documents that embed and cite these resources, share frictionlessly across networks and publish with a diverse set of providers. Giving users a choice of provider for identity, synchronization, social networking and publishing is a core concern of the Open Web and user-centered publishing.

To achieve this vision, developers need support from the Web platform that makes hypermedia authoring a first class citizen of the user experience. The platform should support format, device and view independence for developers building applications that engage users in the creation of hypermedia. In concrete terms, this requires rethinking the notions of selection and focus. Current development around technologies like the Shadow DOM[3]  and Web Components [4]  do not adequately address this facet of script interoperability. The Open Annotation data model provides a framework for creating flexible references, but finding ways to surface this model within the scripting capabilities offered by the Web platform opens the door to modular user interface components that loosely cooperate, decoupling presentational markup from the structured description of a reference.

We posit that a consensus on fundamental models and vocabularies for annotation, alongside platform enhancements that ease third-party proofs of interface concepts, can precipitate renewed interest in raising the baseline for browser capabilities.


  1. Iterate on the Open Annotation data model and produce a standard vocabulary for annotation resources and their relations to other resources on the Web.
  2. Provide scripting support for components to export structured selections, possibly via a data field on the select event or refinement of the DOM Selection object.
  3. Develop a method for handling navigation within documents that provides device, format, and component independence for “scroll into view” functionality.
  4. Standardize a method for programmatic searching of text within documents, exposing the “find on page” functionality common across browser implementations.
  5. Describe media types and link relations for discovery of feeds intended to be consumed as supplemental material.
  6. Describe media types and link relations that provide hints about preferred providers for publishing user generated content directly from the browser.


  1. http://tools.ietf.org/html/rfc3986
  2. http://www.w3.org/TR/media-frags/
  3. http://www.w3.org/TR/shadow-dom/
  4. http://www.w3.org/TR/components-intro/