About 30,000 notes, which is 96% of the total, are not displayed in the ``community note'' function that performs fact checking with the power of Twitter users



From December 11, 2022, the deployment of a new function ``

Community Note '' that allows users to fact-check tweets posted on Twitter has begun. The data in this community note is open to the public to increase transparency, but Bloomberg, an economic newspaper that evaluated this data, pointed out that ``misinformation that creates political conflict tends to be difficult to disclose.'' Did.

Twitter CEO Elon Musk Pushes Community Notes Fact-Checking System Despite Its Major Blind Spot
https://www.bloomberg.com/graphics/2022-twitter-birdwatch-community-notes-misinformation-politics/?leadSource=uverify%20wall

Community notes are a function that allows users who participate in community notes to leave 'notes' that add more detailed information to misleading tweets. Notes will be evaluated by users of community notes, and only notes that are judged to be the most 'useful' by the algorithm will be released to the public.

``Community notes'' that make tweet fact checks with the power of all Twitter users are now available worldwide including Japan - GIGAZINE



Anyone who meets a few simple conditions can participate in community notes, so it is hoped that this feature will prevent the spread of various false rumors by many users. However, Twitter employees and Bloomberg who reviewed this feature point out that 'the lie that causes pros and cons is not fully addressed.' This is because the algorithm needs input from people with different points of view in order to decide that a note is 'useful.'

In the first place, community notes work like this: Bloomberg cites Twitter CEO Elon Musk's tweet. Mr. Mask once posted an image of CNN, an American television station, but you can see that the 'background information' enclosed in a red frame has been added to the tweet. Here are the 'notes' made and rated by the community note participants. The CNN image is a fake image created by a parody site, and the note quoted an article pointing out that CNN was fake.



According to Bloomberg's analysis, this note was evaluated as 'useful' by the majority of participants. People who evaluated the blue tiles as 'useful', those who evaluated the yellow tiles as 'slightly useful', and those who evaluated the pink tiles as 'useless'. You can see that it was evaluated as useful for people of. However, notes are not ultimately judged to be 'useful' simply by the number of evaluations, but behind the scenes various calculations are performed by algorithms.



The algorithm analyzes the tendencies of the participants who rated the notes and roughly determines what kind of thoughts the person has. Based on the results, the participants are classified as follows. Those who are judged to have politically left-leaning thoughts are on the left side of the graph, and those who are judged to have right-leaning thoughts are on the left side of the graph. Tiles are placed to the right of the chart. Looking at it this way, you can see that the notes attached to Mr. Mask's tweet above evaluated that the left side was 'useful' and the right side was 'useless'.



However, it can be judged that this notebook has a certain 'useful' evaluation regardless of whether it is left-leaning or right-leaning. The note was released to the public because it received similar reviews from people with such diverse perspectives. Mr. Musk responded, ``If the community note can correct my tweet, of course it can correct anyone. I rated it.




The advent of community notes can be expected to suppress a lot of misinformation, but according to Bloomberg's analysis, many notes have not been released to the public, accounting for about 96% of all notes posted so far3. It is said that more than 10,000 notes are unpublished.

An example of an unpublished note is the note attached to the tweet below. The tweet was posted by the Family Research Council, a non-profit organization dedicated to clarifying and promoting the ideals of family-centered public life from a biblical worldview, stating, 'Abortion saves mother's life.' It's never medically necessary,' but community note participants should add a note explaining, 'That tweet is wrong,' citing sources such as the Associated Press and Reuters . It was made.




However, since the note has not received a certain number of evaluations from those who are judged to be 'right-leaning' among the community note participants, it has not been released due to an algorithm that prioritizes evaluation from a large number of viewpoints. . Such notes exist regardless of whether they are on the right side or the left side, and it seems that notes that point out incorrect information about abortion, new coronavirus infections (COVID-19), elections, etc. are rarely released.



It can be seen that the published notes will be less politically biased, as Musk intended, but community notes are not an absolute deterrent to misinformation. . Regarding this, Bloomberg said, ``Even if hundreds of users find the note useful, if users with opposing political views choose not to vote, the chances of the note being displayed are extremely low. I will,” he insisted. 'While thousands of notes are waiting for enough votes from different perspectives, misinformation can spread further.'

In addition, according to research by the Massachusetts Institute of Technology's Sloan School of Management, users are more likely to be interested in notes written by people who share the same political views as themselves, and when they share political views with the note author, they are more likely to ' It turned out that there was a much stronger tendency to evaluate that it was useful. From this, Bloomberg analyzed, ``One of Twitter's challenges may be that users don't want to be involved in notes that go against their political ideas.''

Related Posts:

in Web Service, Posted by log1p_kr