Using a blockchain-based reputation and reward system to fix fact-checking

In this blog post I will describe why we have been working on 4facts.org, a new approach for motivating fact-checking efforts and fighting fake news.

Fact-checking has a problem. Even though Internet is overflowing with content of questionable integrity — news articles, blog posts, tweets — we are no where near an appropriate collective response.

While initiatives like PolitiFact and FactCheck.org are doing an amazing job, their capacity is limited due to a limited pool of resources. Their effort is mostly funded with public grants, private donations or cross-subsidies from media organisations. All of these sources are volatile, and fact-checking does not have a sustainable business model.

More importantly, this approach to fact-checking is not scalable. Currently, new potential fact-checkers have almost no economic incentive to start serious work. Of course, there are many online enthusiasts, but there are far too few of them, and it is naive to think that this might change without some sort of compensation. Enthusiast work has another problem — currently there are few reliable metrics for understanding when someone is a competent fact-checker and is doing a good job. Reddit upvotes or FB likes unfortunately are not a useful signal. This lack of reliable metrics is one of the reasons why social media platforms like Facebook or Twitter are struggling with automatically decreasing the visibility of low quality content.

This leaves us with a situation where we have an unfathomable amount of new online content created every day, and a very limited capacity for evaluating the integrity of this content. This imbalance is a major threat to democracy and a free, informed way of life.

So how do we fix the fake news problem at scale?

One obvious response would be to make the fact-checking process automatic by training models to recognise and flag low quality content (people already working on this include FactMata and Fullfact). While there has been some progress in this direction, achieving a usable solution that is not focused on a very narrow domain (or only on large languages) will require notable advances in Natural Language Processing and AI. We are not there yet, and it is hard to predict when we will get there.

That is why we have come up with 4Facts.org, a blockchain-based system that connects fact checkers, content creators, aggregators and consumers in a single peer-to-peer network:

We think that there are several reasons why 4Facts.org can be an excellent way for enabling a scalable, well-coordinated effort for fact-checking.

Firstly, it creates a set of incentives for fact-checking at a much wider scale:

• Fact-checkers are interested in validating content, since they can get financial rewards and also establish reputation for high quality work. The amount of rewards received increases as fact-checker improves his reputation. Ideally, this should become a viable business model for fact-checking that will attract new organisations and freelancers to this effort
• Content creators will be interested to submit their content for validation, since they want to build their reputation. This will be a way of demonstrating their credibility to users (think of something like a “verified” badge on Twitter or Facebook) and attracting a larger audience. Very importantly, these reputation metrics can be used to automatically prioritise high quality content on social media sites at scale — content creators are extremely interested in their visibility on social networks
• Social media and news aggregators are interested in automatically promoting high integrity content — both to ensure better user experience and also to address compliance and CSR concerns. Germany has already passed a bill that will fine social media outlets for not removing fake news, and other countries like Austria are likely to follow suit. Decreasing the visibility of false claims is a step that should be taken by all responsible players in this space

Secondly, the distributed ledger of checked content will be freely available to the general public. This will allow quickly checking the validity of encountered content, as well as reputation of different content creators and fact-checkers. The ledger should also be a great source of input data (accessible through an API) for app developers.

Thirdly, for fact-checkers, a distributed ledger system will help achieve more coordination between different parties in the fact-checking community. Currently there is a lot of unnecessary work duplication in cases where several high-reputation fact-checkers have already validated or invalidated some content. Smart coordination of effort will help achieving more with a limited pool of fact-checking resources. Furthermore, this ledger of fact-checked articles will be a valuable corpus of labelled data for developers of automatic fact-checking algorithms.

We are currently finalising a prototype of this platform and are talking with all involved parties. We are especially interested in engaging with the fact-checking community — established organisations, as well as passionate journalists or activists. Let us improve the integrity of online content together — join us in this fight!

Interested? Sign up to be notified when we launch.