Tackle “Fake News” with Blockchain
Fake news has become a high stake problem to societies & businesses — Permissionless blockchain protocol could be a major part of the solution
Through his frequent usage or giving out peculiar awards, President Trump has “tremendously” popularized the phrase “fake news”. While lies and deceit are nothing new, the widespread use of social media, especially as a source of news, has greatly enhanced their “virality”. In fact, content virality can be highly profitable — Facebook’s profitability is at least partially driven by this factor. Amid public concerns that forced its business model to focus less on virality, Facebook saw its share price fall by 23%, eliminating US$ 119 billion from their market cap — the biggest one day drop in US market history.
Tackle falsehood, not virality
From impacting election result to rubbing salt to the wound of victims of politicized tragedies, the virality of fake news has caught the eyes of many in recent years. However, take a step back, is virality really the main issue? No one was really complaining about viral baby or animal videos after all. In reality, the true cause for concern was the viral content also being false and provocative. Virality magnifies the impact of fake news exponentially but is hardly the issue per se. The fake news problem could have been tackled more directly and effectively if a scalable, trustworthy, and low latency solution exists to detect fake news.
Further narrowing down to what is realistically solvable, any technical solution to detect news falsehood is unlikely to dissuade people who currently believe in well known untrustworthy sources, e.g. InfoWars (if you do not know what this is, good for you — but check out this and this for context — the later of which is my personal favorite). On the other end of the spectrum, news directly distributed from the websites of reputable news outlets (e.g. Financial Times, BBC, etc), while potentially politically biased, likely does not require too much fact checking. A technical solution to detect falsehood is most useful when readers read a piece of news from
- Lessor known news outlets or independent journalists — the “Fake at source” problem
- Known media outlets that appears on a third party distribution channel — the “Fake content” problem
Reputation on the blockchain
Enter permissionless blockchain protocol (or “blockchain” in short) that can be molded into various solutions against “Fake News”. Browser-based ad block product, Ad Block Plus, has released a beta extension called “Trusted News”. This browser extension leverages fact-checking websites like PolitiFact, Wikipedia, etc to rate the trustworthiness of websites. They plan to move the URL rating database to the Ethereum blockchain so that it will be open to user feedback and also robust to attacks.
Redpan, a news aggregator startup, aspires to rate the authors instead of the websites. It plans to first apply data mining and machine learning techniques on news, social conversations, etc in deriving the initial reputation of authors. This reputation can then be further verified by readers of the news aggregator. According to their white paper, they plan to put all the user feedback about authors’ reputation on a blockchain ledger to keep the reputation generation process transparent.
Don’t believe your eyes
While remained to be resolved, news that are “fake at source” (media outlets or authors) is a age-old and well known problem. A much less talked about problem is “fake content”, especially content in the form of videos. The widespread use of photo editing software already makes photos much less convincing as a source of truth. This caution, however, does not apply to videos. With the assistance of neural networks (a type of machine learning algorithm) and the availability of enormous amount of video data online, especially for celebrities, it is now possible to create near-perfect fake videos that make a person in the video to speak the words written by the video creator. This kind of fake video is called “Deepfake”. Researchers from University of Washington has created an example Deepfake video for President Obama.
Imagine when a Facebook or YouTube user sees a video footage of a celebrity delivering a speech. A couple of years ago, even without much verification, such a video is almost certainly real because faking a video well enough to be unnoticeable is very difficult and probably exclusive to experts with access to expensive technology. Rapidly improving and fairly inexpensive DeepFake techniques will overthrow that logic soon, just as photo editing software has taken away the assumed authenticity of a photo.
Fingerprint for content
If widely adopted, Quadrant Protocol could help distributors to verify the content of publishers. The protocol is created to store the “DNA stamp” (i.e. a unique fingerprint) of all kinds of data on the Quadrant blockchain. The process works as follows,
- Media outlets (Pioneer) e.g. BBC, hashes a video and pushes the resulting hash (the DNA stamp) to the Quadrant Blockchain
- When a distributor (Data Consumer), e.g. Facebook, sees the video uploaded to its platform, it can hash the video and check,
a) if the “DNA stamp” is recorded on the Quadrant Blockchain, and
b) If the creator of the video, also stored in the DNA stamp, is BBC
Since any change to the video, DeepFake included, would result in a completely different DNA stamp, the Quadrant protocol can help to easily identify DeepFake videos. This is obviously under the assumption that the publisher of the video is following the Quadrant Protocol.
Why permissionless blockchain protocol?
For the reputation of a website, journalist, or the fingerprint of a video to be trustworthy, it is crucial that the database keeping such records is immutable, i.e. once recorded, it cannot be changed by any central party. Reputation can improve or deteriorate based on new user feedback but the whole history of reputation changes should remain intact. Data immutability is the backbone of a a trustworthy verification solution. Also, if such a verification solution is built through cooperation of the major media players, it could be difficult for any smaller outlet to participate, effectively creating additional entry barrier to the media industry. Comprising both immutability and decentralization of control, a permissionless blockchain protocol is well positioned to take up the “Fake News” challenge.
