Better Support for Macrotasking

[mbernst]

Problem

At least two of our largest requesters in the last few months had extremely large tasks (e.g., asking Impossible Questions, transcribing a long video/audio clip). Each task took up to a couple of hours to do. The tasks both stalled, and workers complained about the task being too risky to try --- nobody wants to spend three hours on a task only to have to redo it. I suspect we will see more of this kind of work on the platform going forward.

Proposal

While I would not suggest this be our top priority, this proposal articulates a commitment on our part to investigating whether we can better support such large, long tasks. The Collective has previously held many discussions about this under the term "macrotasking", borrowed from Adam Marcus and team's Argonaut paper. Those brainstorms included, for example, the idea of task milestones. (IIRC, I believe that eventually evolved in part into Prototype Tasks.)

Concretely, we will continue to investigate mechanisms that might support such macrotasks. We have already collectively developed support for several of these. Examples might include:

• Milestones (initial practice tasks) that gate people into the full task
• Mechanisms to retain workers who do well on a macrotask
• Designs that might make the task feel less risky for workers

Implications

Short-term: this proposal implies a commitment on our part to monitoring large tasks as they are posted on the marketplace, and think about ways to improve support for them.

Long-term: I think this will be an absolutely critical step to lifting crowd work out of the microtasking swamp and towards work that is more fulfilling, better paid, and engages with expertise.

Contact

@michael Bernstein on Slack

---

To officially join in, add yourself as an assignee to the proposal. To break consensus, comment using this template. To find out more about this process, read the how-to.

[qwertyone]

Lessening perceived uncertainty through the interface: a feedforward notification system for requestors to communicate current response times for payment that can be expected by workers. I began coding this a long time ago and the effort never got off the ground. The idea targets the uncertainty in response times from requesters to workers.
https://github.com/qwertyone/a-little-green-square
Proposal: http://crowdresearch.stanford.edu/w/index.php?title=Winter_Milestone_4@System_Proposal-A_feedforward_visualization_system