For researchers: try to break it
Transparency News invites outside researchers to test whether its published rules produce skewed results: rankings that favor one side, words that are treated differently depending on who uses them, or cases where the score misleads.
What is available
- Every rule, word list, severity and constant: Method, with the rule book's fingerprint.
- The code that scores every report: Source code.
- Every measurement since the site began, by month, plus a daily snapshot of every outlet's score and every front-page snapshot: Archive and downloads (free to reuse with credit, CC BY 4.0).
- Each event's comparison as a spreadsheet or data file, linked from every event page.
- The cases the method already gets wrong: the challenge set on the audit page.
How to submit
Send your method and results through the corrections page with the type "Research challenge". Every submission gets a published outcome. A result that shows skew is published with the fix and a new method version.
What the site cannot share
Article text. Since method 3.1.0 the site can keep the text of pages it read with the outlet's permission, privately, only to measure reports again after a rule change and to settle disputes. Private storage is not switched on yet, so no text is kept today. That text is not published or shared. Studies that need the full text should use the Internet Archive links recorded with each measurement.