How to Spote Fake News
We live in a murky era of misinformation. Alex Kasprak of Snopes says the volume of really shoddy stuff out there is mind-numbing. There’s an old saying that lies have traveled half-way around the world before the truth has time to put her shoes on.
Nothing beats researching information we find questionable, particularly prior to acting on it or sharing with others. We all bear responsibility for not passing on misinformation because of serious consequences.
But, what if fake news could be detected before it reaches us? Programmers are working on automated systems that can red- flag information. They can correctly identify false articles 71% of the time, but automatic lie-finding tools are “still in their infancy,” according to computer scientist Ciampaglia—Indiana University.
Programmers first studied characteristics of false versus true information. They discovered that the less related the subject and the object were, the more likely the information is incorrect. Relatedness is determined by a huge web network of nouns. Think of the meta law that any two people are connected within seven degrees of separation. If the subject and object are only related via seven steps, the sentence is more likely to be false.
This test alone is limited, but adding more characteristics improves the process. A careful analysis of 75 true and 75 false stories revealed that compared with real news, false articles tended to be shorter, more repetitive, contain more adverbs, and fewer quotes, technical words, and nouns. Other studies indicate that false news expresses higher certainty. Real news contains more work-related words, negations, comparisons, and quantifiers. Fake news contain more social words, positive emotions, thought processes, and references to the future.
Another method lumped like articles together, and verified a small percentage to predict the credibility of the lumps. The accuracy was also about 70%.
A computer scientist in China examined news-related tweets by comparing backgrounds of those who believed the story versus those who didn’t and gave more weight to the opinion of individuals who were in a better postion to know. Even the patterns of retweets of a real story versus a fake story were different enough to predict veracity.
Facebook is using some means not to eliminate suspect posts, but to show them to fewer customers.
I thought I was a good detector of fake news and joked that growing up on a farm, I was familiar with the smell of bullshit. I laughed at the poor presentation and ridiculous nature of the lies passed in emails. However, one online post caught me. It fit my belief system, cited a reporter, and resembled a previous true story.
Ala Hill Street Blues: We all need to be careful out there.
I highly recommend the long article from Science News on which this post is based.