The Corrections: Can a Third-Party App Stop the Spread of Inaccurate Tweets?

If lack of a correction tool is Twitter’s Great White Fail Whale then Web developer Adrian Holovaty may be its Ahab. The whale raised its head in January when NPR incorrectly reported Rep. Gabriel Gifford’s death after she was shot in Tuscan, beginning a series of serious discussions about how Twitter corrections should be handled.

Holovaty, a developer, journalist and the founder of EveryBlock, led a forum along with editor Craig Anderson Tuesday on addressing the subject of how exactly a correction tool for Twitter might work, with the goal of arriving at something of a “requirements document.”

One possibility for the future may lie in an “Annotations” tool currently under development at Twitter, which could allow notes and other metadata to be attached to tweets. There’s no timetable for release of the feature and sources at Twitter had no comment on where it stands.

Generally change in the Twitterverse comes from without, not from within Twitter HQ. As Holovatypointed out grassroots-driven features are in Twitter’s DNA; even hashtags were started by the community. With its own developers frying proverbially bigger fish, we can’t expect the tool to come from Twitter, but evolve as a result of usage and innovation from developers.

“Apps like TweetDeck and HootSuite can exist because Twitter has an API,” said Anderson, “They have a bit of an approach and philosophy that says, ‘Let other people innovate on our platform.’ So it’s up to us.”

Working within the framework of what could currently be implemented and starting with a plan for best-practices, Holovaty and Anderson (based partly on thinking on the matter already done by Anderson) presented a clear plan to move forward with a tool and system, and Holovaty is now hard at work on coding an automated solution.

For the time being, at least, the third-party app would function as more of a correction notification system. As outlined by Holovaty here are the bones of that system:

  • A Web site (plus an API, so third-party apps can build native functionality in their apps) where a user would enter the URL of the tweet they’d like to correct (verifying they’re the person who sent it)
  • Behind the scenes, the app makes a note of everybody who has retweeted the erroneous message
  • On the Web site, the user would type in a correction: “No, that dude didn’t actually die”
  • Then it would give you a “Delete the original tweet?” checkbox, (leaving that highly controversial question to individual users)
  • Finally the app would send an at-reply correction to all of the people who retweeted the original (though there can be rate limit issues there that needs to be addressed)

The tool would have the goal of quickly diffusing an incorrect post and stopping the spread of inaccurate information. A hashtag such as #correx or #cx would help people to quickly check if a tweet has been corrected and would, said Holovaty, “a way for people to aggregate corrections and do other interesting things with them.”

Everyone involved is quick to stress that the system will only be as good as the way people use it, and hope that, with community feedback, a working model will be chasing the Great White Fail within weeks.

How do you think corrections should be handled?