Reading and watching the news has always been an important way for people to educate themselves about the world. However, as it becomes easier to spread information and opinions through the internet, people are increasingly getting their news from untrustworthy sources, which leads to misinformation and confusion about which facts are true.
Between 2016 and 2017, the number of people who have gotten news from Twitter has gone up 15%. For Snapchat, it’s 12%. And from 2013–2017, the number of people getting news from Facebook has gone up 21% (Shearer and Gottfried). While it’s impossible to know the exact ratio between fake and real news on social media, anything can be shared, and there’s no differentiation between fact, opinion, and lies. Watching mainstream TV often isn’t any better. According to Politifact, 60% of claims made on Fox are untrue, 44% for MSNBC, and 21% for CNN (Sharockman). These claims only include those on news channels and can include claims from the anchors or pundits. The phenomenon of untrue news has come to be known as fake news. The term fake news first came into public attention during the 2016 election, when a Buzzfeed journalist noticed a large amount of false news coming from Macedonia. Soon afterwards, both candidates started using fake news to attack each other and Trump used the term fake news to deflect news that he did not like. According to a Stanford study, the spread of fake news has caused 84% of American voters to face difficulty in differentiating between real and fake news. This problem also affects young people. 44% of teens and tweens say that they cannot tell the difference between real and fake news.
There are many proposed solutions to fake news including media literacy classes and improved algorithms. However, these solutions don’t address the underlying cause of fake news. And, as technology and AI improves, new forms of fake news make these solutions ineffective.
Watch Jordan Peele use AI to make Barack Obama deliver a PSA about fake news_What does the future of fake news look like? No one really knows, but here's a little sampler from Jordan Peele and…_www.theverge.com
In order to truly get rid of fake news, we need to attack the incentive structure behind it. The main cause of fake news is the misalignment of incentives between news publishers and readers. As Harvard Business Report writes, a true solution needs to “solve the media’s structural challenges [and] change its incentives”. News publishers make money solely off of writing interesting news and getting clicks (there are subscription based news sites that are marginally better, but most people aren’t going to subscribe to news publishers that don’t write somewhat interesting news). The incentives of readers should be to inform ourselves, but our brains are wired to prioritize “popularity over quality”. As a result, news publishers are rewarded for writing sensationalized (and often completely fake) news. This has always been the case (ex. Yellow journalism) and the internet has drastically increased the amount of people who are able to make money from writing sensational news.
In order to fix fake news, we need to disincentivize news publishers from publishing fake news (making it unprofitable or a waste of time to write fake news) and incentivize people to do research and think critically when they read the news.
My proposed solution comes in the form of a prediction market powered news site. On this system, news publishers would have to stake a certain amount of money in order to publish a story. If the story is confirmed to be true, the news publishers would get their stake back. If the story is proven false, the publishers would lose their stake. This disincentivizes publishers to reward structure for fake news because their news story gets deleted and they lose money. This completely changes the reward structure for news publishers. There’s is absolutely no reward (there’s actually a loss) for writing fake news. However, the potential reward for publishers is much bigger than traditional publishing. News readers would obviously like being able to earn money. This will lead to more ad money for the news publisher. The hard part about this is proving whether a story is real or fake. This is where crowd wisdom comes in. The readers of the story can be asked whether or not they believe the story is real or fake. These readers must wager money on their prediction. In addition, the readers are asked how they think other people will vote. Their answers are put through the Bayesian Truth Serum, which finds the surprisingly popular answer. Studies show that the surprisingly popular answer (and not the majority vote) is the correct one. An MIT study showed that the Bayesian Truth Serum “reduced errors by 21.3 percent compared to simple majority votes” (Knight). If the vote says that the news story is true, the story stays on the site and can earn ad money, subscription money, part of the money that the false voters wagered, etc. If the vote says that the news story is false, the news publisher loses their stake to the yes voters and the story is deleted. The people that voted true also lose their money. The wagering system (also known as a prediction market) incentivizes news readers to do research and more importantly, it incentivizes them to share their research to try to convince other people to their side. In addition, this system forces people to change their frame of mind. Instead of seeing news articles from a lens of agreeing or disagreeing with it, they need to look at it from a purely objective point of view in order to make a vote. The effectiveness of this solution depends on two things. The first one is it’s popularity. If no one reads stories on this site, then it becomes useless. The second one is the accuracy of the prediction market. Unless the prediction market is accurate the majority of the time , then this solution would actually be harmful because fake news would be promoted as real. However, “prediction market is closer to the actual outcome than are polls 74% of the time” (Cornell).
Also, it could be more feasible to fact check specific statistics within a story, assign a “truthfulness” score, and remove stories that have too low of a score. This would ensure that only empirical claims were being analyzed.
It would make sense to utilize a service like Augur (which comes with the additional benefits that the Augur prediction market provides). In addition, the news site could be hosted using the Blockstack system, ensuring decentralization. Using Blockstack’s p2p payment system will also ensure that publishers are fairly paid.
I don’t personally have enough developer experience to solo this project, but I think a small team of people could build a simple MVP over a weekend or two. If you’re interested in doing this, please reach out to me either via Twitter (@Legionof7) or Discord, where I’m Legionof7 here (his is the best option).
I’m excited to see what this could turn into, please reach out!