At RJI, we’ve been working to improve how we share information with our readers.

Subscribe

Photo: Siriuswerks/FlickrMany moons ago, I started exploring the idea of a "credibility layer" that could help remind readers to think more carefully. The more I dug, the more I realized that accomplishing something like this would involve taking on three impossible tasks:

  • I would need a user interface that people were willing to use, even if it challenged their world views.
  • I would need to find a good way to collect credible information and add it to the tool.
  • I would need a good algorithm that could detect other phrases that were similar to the ones stored in the system.

While working on the Truth Goggles prototype, I focused a lot on that first challenge, but I wasn't able to go any further since graduate schools like it when you graduate. Thankfully, RJI wants to help me make some progress on the other two fronts!

Staying realistic

During this new era of Truth Goggles development I'm keeping a keen eye towards two grim realities; the computational challenges around language are huge, and I lack the specialization to solve them. That might sound like really bad news, but actually it just means that I need to leverage tools made by others and make interfaces that allow humans to contribute their brains to the system.

For example, the first part of my fellowship is being dedicated to building a tool that will make it easy for journalists to contribute notes and articles to the Truth Goggles knowledge base. This will end up being quite stupid from an algorithmic perspective; it is just a simple submission system that sucks up documents and provides a nice interface for the authors to quickly highlight their key points and submit them to the database.

The benefit of this approach is that the challenges become human, not computational. What kinds of interfaces would reporters be willing to use? How much time would they be willing to spend to do this (hint: not much) and what do they need to get out of it to be willing to participate? Over the next month and a half, I am finishing the first version of this system which I will use to run tests with real reporters in real newsrooms. From there, it will be back to the drawing board with a slew of great feedback.

Daniel Schultz  
 
Nonresidential fellow



Share

Recommended for You

Related Stories

comments powered by Disqus