Fake news: What is it, and how can we tackle it?
www.nesta.org.uk/blog/fake-news-what-is-it-and-how-can-we-tackle-it/
www.nesta.org.uk/blog/fake-news-what-is-it-and-how-can-we-tackle-it/
Although a technological quick fix for fake news seems unlikely, the global tech community has begun to use some of these methods to fight fake news.
Reports that ‘Fake news’ may have affected the outcome of the US election have directed the world’s attention to this rapidly growing phenomenon. As three of the EU’s founding nations - Germany, France and the Netherlands - head to the polls in 2017, many are worried that Europe may suffer a similar deluge of misleading, or brazenly untrue, information.
As part of our work looking at digital social innovation (DSI) across Europe, we have been exploring the various ways that people are using the power of digital platforms, crowdsourcing and other technologies to develop tools to tackle social problems. Although a technological ‘quick fix’ for fake news seems unlikely, the global tech community has begun to use some of these methods to fight fake news. Below, we explore if and how these tools can help us.
Fake news is not a new concept. It has a long and brutal history - from 15th century anti-semitism and a mysterious 18th Century earthquake through to ‘pizzagate’ and the Pope’s apparent support of Donald Trump.
However, as the internet has given us access to a world of instant information, it’s also enabled the spread of disinformation on a hitherto unseen scale. It’s this new form of digital fake news that has become so pervasive. We can identify two broad categories:
stories that may have some truth in them, but are biased or based on unreliable sources
deliberately false stories, designed to mislead and cause confusion.
As more of us move to online sources for news, we have become more exposed to unreliable news sources but remain ill-equipped to spot them.
Most worryingly of all, it may be about to get much worse. So far, fake news has, on the whole, been limited to written text but we can expect to see video formats being used more in the future. For example, Adobe's new ‘Photoshop for audio’ enables voices to be replicated and form new words, while the Face2Face project's manipulation of video footage in real-time gives an indication of the direction technology could lead us.
Human-led fact-checking is the most obvious (and longstanding) weapon against misinformation. Often attached to a news organisation or university, fact-checking organisations like FactCheck.org, PolitiFact, or Channel 4’s Fact Checking Blog have assessed political claims and debunked myths for years.
Although an effective method for assessing a more manageable amount of information - claims made by politicians in parliament for example - this labour-intensive approach cannot keep up with the current pace at which fake news is emerging. (A recent investigation found that a single ‘fake news farm’ was responsible for 78,349 YouTube videos across 19 channels with some channels uploading a new video every three to four minutes.)
As traditional fact-checking becomes impractical, technology is enabling two other approaches to combat fake news: harnessing the crowd and using artificial intelligence.
Digital platforms enable large numbers of people to work together remotely. For example, myth-busting platforms like Snopes rely on users to crowdsource ‘rumours’ or fake news which can then be verified or debunked by the site. Although the level of accuracy of Snopes is very high, like many fake news tools, its information struggles to reach those who aren’t looking for it. For example, even after rumours that Barack Obama was a Muslim had been debunked, a CNN/ORC poll found 43 per cent of Republican supporters still thought the former president was Muslim. If tools are to have a meaningful impact, they must find ways of being more closely linked to the news-reading process, or to the readers themselves.
FactCheckEU.org went one step further, enabling the crowd to participate in the verification process itself. Users were able to submit information for review, vote on claims, translate sources into other languages, and verify or debunk claims by citing sources. Initially launched by Pagella Politica, an Italian political fact-checking website, as a pilot during the 2014 European Parliamentary Elections, the site closed in June 2016 after funding was cut.
If fact-checking initiatives aim to categorise individual claims as true or false, other tools try to asses news outlets as a whole. OpenSources relies on the crowd to submit news sources for review which are then categorised by the site’s experts. There are many potential categories, from ‘fake news’, ‘conspiracy theory’ and ‘junk science’ to ‘hate news’, ‘political’ and ‘credible’.
Artificial intelligence (AI) allows for the automation of human tasks and, with machine learning algorithms, can train itself to become more efficient over time. This technology may offer another route to success. In the UK, fact-checking organisations like FullFact and Factmata have begun to experiment with automated fact checking. Robocheck, FullFact’s new tool under development, should be able to check the truth of claims against verified data sources within seconds. Of course, all fact-checking, digital or not, relies on access to reliable sources of open data. As AI projects develop, the importance of open data initiatives such as EUROSTAT, the Open Government Partnership and the World Bank DataBank will increase.
Rather than offer an absolute judgement on truth, other AI initiatives are hoping to give warnings to readers about the credibility of news, based on a range of signifiers. For example, the Fake News Challenge has issued an open call for AI solutions which can assess the stance of an article on a certain issue. This stance can then be compared with the stance of trusted news sources, meaning that outliers could be flagged as potentially untrustworthy. Other signifiers might include a young website domain (trusted news sources are generally well-established), or claims based on data which do not cite sources.
As is so often the case, our best chance of success may come through trying a little of everything. A number of fake news initiatives have committed to licensing their work under Creative Commons Licenses or sharing it via GitHub, allowing others to contribute, collaborate and cherry-pick aspects of individual projects.
For example, the crowd-led OpenSources project is open source and freely available on GitHub. This has enabled the team behind B.S.Detector, a browser plugin that detects the sources of news stories, to integrate OpenSources’ classifications within their sensing application. Scrolling through a Facebook newsfeed with the B.S. Detector plugin will classify newslinks according to the listings on OpenSources.
Tackling fake news is a major technical challenge. However, it also presents a number of opportunities to bring about positive social change.
We cannot expect fake news tools to work with 100 per cent accuracy. And their efficacy will be further hindered if the trend to dismiss any source that contradicts a person’s strongly held beliefs continues - as witnessed in the recent US election. Therefore, a push for greater digital literacy and education will be key. It will be particularly important to reach internet users who might not automatically doubt the truth of claims on the internet. We can follow the example of the Cibervoluntarios Foundation in Spain, which sets out to teach digital skills to groups of people who are at risk of digital exclusion.
As we challenge fake news we should also think about how data can be used to show the truth more effectively. There are a number of exciting initiatives using open data to inform the public about important issues. Examples include the OpenSpending platform and the FunkyCitizens initiative which track government spending, or Wikirate and OpenCorporates which shine a light on the practices of large companies.
Clearly, fake news presents a major challenge for society, but technology is both a cause of and potential solution to that problem. As this article has shown, there are a range of tools being developed with very different uses: some are designed for journalists, others for concerned citizens and a number for those who turn to social media for their daily news. It’s in all our interests that they succeed. Yet the technology alone will never be enough. Rather, it must be hoped that we use the fake news crisis as an opportunity to become a more open, discerning and media savvy society.
This blog was originally posted on digitalsocial.eu, a platform for digital social innovation in Europe. Read the original.
