Navigating Big Data’s Fake News Problem
In 2013, hackers managed to access the AP twitter account and posted a fake tweet implying that there was an explosion in White House and the then President Obama had been injured.
Following the Tweet, the stock market went into panic mode with the Dow plunging over 150 points and the S&P 500 shedding $135.6 billion in a matter of seconds. The flash crash was associated with high-frequency trading algorithms interpreting the tweet as true and executing huge sell orders.
Fake News and Big Data
Fast forward in 2018, and fake news is no longer limited to occasional incidents of hacking. Social media has seen explosive growth in the last few years and with it has come a massive influx of fake news, often used purposely as a tool for business and manipulation. Even worse, as media technology continues to get more sophisticated and available, fake news is becoming hard to detect.
While fake news is a problem for those who seek information online, it is a bigger problem when viewed from a big data perspective. This is because in this age, big data is the core of decision making for governments and businesses alike and when elements of bad data can no longer be sieved from big data, poor decision making is inevitable because the decision is based on faulty facts.
Even when not rampant, fake news has the power to make big data unreliable. Falsehoods and clickbait tend to spread faster than truth and a single piece of misinformation has the power spread like wildfire, hurting the integrity of big data.
If big data is to be trusted for decision making, a solution must be found to deal with the fake news menace conclusively. Following the exposure of the part social media played in alleged Russian interference in the 2016 US elections, Facebook and Twitter have introduced a mechanism that enables people to flag news they suspect to be fake. When a good number of users tag a story as fake, it appears less in people’s news feeds and carries a warning that it may contain false information.
While this is a good start, there are many issues with this system, such as people flagging content because they don’t like it, limiting them from seeing news which will contradict their worldview and hurting publications needlessly. This solution also doesn’t address the big data problem because people are likely to identify fake news based on other people feedback not necessarily based on accurate fact checking. The approach of identifying fake news on the basis of users’ feedback may not be very reliable given that people can manipulate the feedback to punish those that they do not like.
With blockchain technology, it is possible to trace news directly from the original source and therefore determine if it can be trusted. The distributed ledger technology makes it possible for all parties in a network to identify the chain of an article’s distribution of information from its origin. Apart from exposing the original source, blockchain protects the information from manipulation by ensuring that each party in the network can follow the entire dissemination process.
Another way that blockchain is helping tackle fake news is by incentivizing good content and punishing bad content. Social media platforms such as Ask.fm are leveraging the token economy to reward content publishers based on how their content is received by users. Content that is popular with readers is rewarded with tokens and shown to many readers while unpopular content is hidden.
The more a publisher gets positive feedback on the Ask.fm platform, the higher they rank in earnings and discoverability. This gives them an opportunity to attract brands that need exposure and are paid to mention them in their posts. Given that the publishers’ earnings are dependent on the overall rating, they have to ensure that they only publish high quality and truthful content.
However, while the incentive model may help eliminate fake news to some extent, it cannot be relied on fully either. As mentioned earlier, a well crafted fake story is likely to gain popularity more than the truth given the human innate tendency to share rumors. Also, just like the Facebook approach, the incentive model can be gamed to meet a certain group interest. The incentive model also doesn’t offer a solution for the big data issue given that it does not provide a way for the algorithms to determine the legitimacy of a piece of information.
With the rate at which the blockchain is advancing with, there is hope that better solutions to fake news and big data issues are yet to come. The few available options have already set the roadmap and it is just a matter of time before lasting solutions are found, In the meantime, exercising healthy doses of skepticism and caution is a good idea.