Analysis Reveals Major Problems in Twitter’s BirdWatch Misinformation Checking Program

Last month Twitter introduced a new feature on their application. The feature is called the Bird Watch Fact Checking Program. The program is currently in pilot mode and the company is testing its basic only with a grand total of one thousand users having the accessibility to it. But what is the purpose if this new feature that is introduced by Twitter? Well, the Bird Watch Fact Checking Program allows people who are under the Bird Watching community to flag tweets that they think are misleading or falsely interpreted and under the notes can convey what the correct and authentic information regarding the Tweet is. Other birdwatch users can rate the response in the notes on the basis of how helpful it is. The community of the one thousand birdwatchers is very diverse and from all parts of the world in order to ensure that most helpful notes can be elevated for the public and to minimize the spread of false information.

However, less than month after the program being introduced, a study revealed some troubling information for Twitter related to crowded source initiative. An origination Poynter Institute thoroughly searched through almost 2700 contributions from different Birdwatchers and revealed that only less than the half of the contributions to the program submitted reliable and authentic information in response to Tweets they had flagged. The institution also enlightened us with the fact that some of the tweets had a rather biased response to it, meaning that people flagged or rated them based on only one sided information of the story which was wrong and diminished the meaning of the newly introduced feature. The institution said that among the tweets that they analyzed this week only seven percent of them were rated helpful, while others tweet notes just had different tweet links or some just flagged the tweet with no response as to why they think it is not authentic.

After the responses on the Tweet of Tim Pool who had tweeted about the US presidential elections being rigged. The Tweet was a mixture of responses because some Birdwatch users disagreed and it backed up Pool’s assertion, with others rating such notes as helpful. After this Twitter removed the helpful labelling from the tweets and changed a bit of Twitter Birdwatch algorithm.

The new settings of the algorithm said that the notes will not only be labelled as helpful if they have at least three to five ratings on the cited note and the helpfulness rating threshold was increased from 0.5 to 0.84 by the company.

The leading fact checkers of Twitter, suggested Poynter that Twitter should give proper training to Birdwatch users on what basis should they flag a tweet and also should provide a solid evidence or content which supports their stance claiming that the tweet is misinterpreted. Also, only professional employees should have access to high ranking notes. Upon which Twitters vice president of product, Keith Coleman said that the company right now does not have the time to roll out a much broader version of Birdwatch and is trying to learn from the program on what changes could be brought about in the near future while the community of Birdwatch is still small.

Read next: Viral Tweets Are Not as Valuable as You Might Think