WearWrite: Orchestrating the Crowd to Complete Complex Tasks from Wearables (We Wrote This Paper on a Watch)

Michael Nebeling, Anhong Guo, Kyle I. Murray, Annika Tostengard, Angelos Giannopoulos, Martin Mihajlov, Steven Dow, Jaime Teevan, Jeffrey P. BighamarXiv 2015 · OpenReview

Introduction

Wearable devices, particularly smartwatches, have revolutionized how we receive notifications and track health metrics. However, they remain notoriously poor for content-heavy, complex tasks like academic writing. The primary constraints are physical—the "fat finger" problem on tiny screens—and behavioral, as users typically engage with wearables in "micro-moments" lasting only a few seconds.

WearWrite is a collaborative authoring system designed to bridge this gap. It allows a user to write a complex academic paper from their wrist by "orchestrating" a remote crowd of workers. By offloading high-bandwidth prose composition to desktop-based workers, the author can focus on providing high-level expertise and direction during brief intervals of attention.

System Architecture

The WearWrite system synchronizes a smartwatch interface with a collaborative editing environment, specifically Google Docs. The architecture is composed of four primary modules that manage the flow of information between the mobile author and the desktop-based crowd.

The WearWrite system architecture, detailing the components for the Android app, Google Docs extensions, the Node.js server, and the crowdsourcing task management.
  • The Author Interface: An Android Wear application where the author dictates "seed" content (outlines or key points) via speech recognition and reviews crowd edits.
  • The Crowd Interface: A custom task queue that provides workers with structured instructions, such as "Turn these bullets into a paragraph."
  • The Middleware: A Node.js server and Chrome extensions that facilitate real-time synchronization, capturing document states and extracting edits to be sent back to the author’s watch.

The Orchestration Loop

The core of WearWrite is the interaction loop between the author and the crowd. The author acts as the "expert-in-the-loop," providing the essential domain knowledge that the crowd lacks. The author provides instructions, and the system provides a window into the document's progress directly on the watch face.

The WearWrite interaction loop, illustrating how an author gives instructions to the crowd and reviews their responses via the smartwatch interface.

This loop allows the author to maintain control over the document's direction without needing to type a single word of the final prose. During the study, authors used these "interaction moments" to approve or reject edits, respond to worker comments, and refine the document's structure.

Experimental Results and Insights

To test the system, the authors used WearWrite to write the very research paper describing the system. The results demonstrated both the feasibility and the current limitations of crowd-orchestrated writing:

  • Feasibility: A 6-page academic paper was successfully drafted within three days.
  • Cost and Labor: The draft required 36 hours of crowd labor, costing approximately $810 USD. While high, this represents a significant compression of the traditional writing timeline.
  • The Role of Expertise: A key finding was that while the crowd is excellent at prose generation, they struggle with sections requiring deep domain context, such as "Related Work." This highlights that while labor can be outsourced, high-level synthesis remains a non-transferable expert task.
  • Worker Agency: Crowd workers often took ownership of the text, suggesting better terminology and acting as informal peer reviewers rather than just passive typists.

Limitations and Future Directions

Despite its success, the system faced challenges with speech recognition errors, which required authors to spend extra time correcting their own dictated instructions. Furthermore, maintaining "global context" from a small screen remains difficult; if an author becomes too disconnected, their feedback loses effectiveness.

Future work will explore integrating voice-to-comment features and expanding this orchestration model to other constrained input devices, such as Augmented Reality (AR) headsets, where the crowd can continue to act as a human-powered interface for complex productivity.

Summary updated

Abstract

In this paper we introduce a paradigm for completing complex tasks from wearable devices by leveraging crowdsourcing, and demonstrate its validity for academic writing. We explore this paradigm using a collaborative authoring system, called WearWrite, which is designed to enable authors and crowd workers to work together using an Android smartwatch and Google Docs to produce academic papers, including this one. WearWrite allows expert authors who do not have access to large devices to contribute bits of expertise and big picture direction from their watch, while freeing them of the obligation of integrating their contributions into the overall document. Crowd workers on desktop computers actually write the document. We used this approach to write several simple papers, and found it was effective at producing reasonable drafts. However, the workers often needed more structure and the authors more context. WearWrite addresses these issues by focusing workers on specific tasks and providing select context to authors on the watch. We demonstrate the system's feasibility by writing this paper using it.
BibTeX
@misc{nebeling2015wearwrite,
  author = {Michael Nebeling and Anhong Guo and Kyle I. Murray and Annika Tostengard and Angelos Giannopoulos and Martin Mihajlov and Steven Dow and Jaime Teevan and Jeffrey P. Bigham},
  title = {{WearWrite: Orchestrating the Crowd to Complete Complex Tasks from Wearables (We Wrote This Paper on a Watch)}},
  year = {2015},
  eprint = {1508.02982},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1508.02982},
}
Figure