A collaboration between NGOs First Draft News and Full Fact to 1) monitor and identify news items and sources, and 2) verify those news items and provide the public and news media with up-to-the-minute analysis and reports of their veracity.
Problems and Purpose
First Draft News and Full Fact are two non-profit organizations dedicated to the verification of news and self-described 'factual' content posted on social media. The two organizations collaborated on the "UK Election Watch" project to combat the dissemination of fake news during the British general election in 2017.
Fake news has gained particular political importance in the wake of Donald Trump’s election to US President in 2016. Not only did Trump use of the term “fake news” to discredit legitimate news sources, but his victory has been partially accounted for by the dissemination of actual false information about his political opposition by members of his campaign, his supporters, and Russian intelligence operatives.
That 'fake news' would similarly impact the 2017 general election in Britain was a real fear among UK citizens. Thus, the collaborative initiative lead by First Draft News and Full Fact was an attempt to keep democracy 'pure' and to ensure voters were not swayed by false or misleading information claiming to be true.
According to First Draft News and Full Fact, the UK Election Watch collaboration was designed with two aims:
- "To provide the public with the necessary information to form their own conclusions about the information they receive. By working collaboratively, we [First Draft News and Full Fact] hope to more quickly ascertain what is not factual or reliable, and to give that information to newsrooms and the public.
- At the end of the project, [collate and send] our protocols and findings to help other journalism organizations in collaboration"(Moschella & Watts 2017).
Background History and Context
A Brief History of Fake News
Propaganda and misinformation have existed in all aspects of human life – and particularly the political sphere - from the earliest times of civilization. According to Brendan Nagle and Stanley Burstein (2009), many historians point to the description of Darius I’s rise to the Persian kingship in 515 BC as one of the earliest examples of political propaganda. (Nagle, Burstein, 2009). Others have pointed out that Rameses the Great spread lies after the Battle of Kadesh in Egypt (Canduci, 2009) even earlier, in the 13th century BC. Of course, with the invention of the printing press in the 15th century and the information it permitted to disseminate, the use of propaganda grew, with modern examples in the 20th century including during the Russian Revolution, both World Wars, and the Cold War.
Most recently, the author Kevin Young published in late 2017 the book “Bunk: The Rise of Hoaxes, Humbug, Plagiarists, Phonies, Post-Facts, and Fake News”, which aims to document some of the most notable hoaxes throughout history. Thus the “post-Truth Era” term coined by Ralph Keyes in 2004 in his book of the same name should not have been such a surprise. Nor should it have been a surprise when the Oxford English Dictionary chose “post-truth” as its word of the year for 2016. That same year, Donald J. Trump adopted the term “fake news” as a tool in his daily interactions with the press during his presidential campaign – a tactic he has continued to use since being successfully elected US President.
The major difference between now and the 13th century BC is the speed and breadth of dissemination of propaganda – and what henceforward in this article, we’ll call “fake news”. Technology – the Internet broadly and social media specifically – have permitted an almost infinite amount of information and misinformation to be disseminated. Fake news has grown in its presence alongside the growth of the Internet and social media. Fake news can now take many different forms including:
- False or misleading articles or headlines in traditional media
- Articles of news stories planted in social media feeds like Facebook and Twitter
- Internet “trolls”: individuals who engage on different Internet-based fora (chat rooms, blogs, websites, etc) to create antagonistic dialogue
- Fake news websites dedicated to false stories, often creating sensational stories to garner web traffic
What is notable about these forms is how highly customizable they can be given current technology. An individual based in South Africa who has an interest in Canadian elections (as an example) can be very specifically targeted by the generators of fake news, often through automated bots and increasingly through artificial intelligence (Greenemeier, 2017).
Fake news has gained particular importance in the wake of Donald Trump’s election to US President in 2016 – not only for his regular use of the term “fake news” but also for some postulations that the use of fake news was an important factor in his victory. According to Statista, the engagement of news stories on Facebook regarding the US presidential campaign had a ratio of 9:7 of fake news stories to mainstream, “true” news stories from August 2016 to election day. In the final three months of the US election campaign, Facebook users interacted with a materially greater number of “fake news” stories than with true news stories regarding the presidential election.*
A Brief History of Fact-Checking
With the rise of fake news - especially using new technologies and social media - has come a growing community focused on fact checking. While new technologies have been used to spread information, they can also be used for fact-checking. Historically, fact checking was the domain of journalists – with many fact-checking activities gaining prominence in the US in response to some of the yellow journalism of the 1890s (not coincidentally followed by the truth-obsessed “muckraking” journalism a decade later) (Fabry, 2017).
However, according to one of the global leaders and standard-bearers in journalism, the Poynter Institute, the development of a dedicated, professional class of fact-checkers began in 2003, with the launch of FactCheck.org (Monerat, 2017). This led to a fact-checking cohort that subsequently grew into its current existence, with over 100 non-partisan organizations globally dedicated to fact-checking (Stencel, Adair, 2016). The Reporters Lab at Duke University tracks these fact-checking groups and has identified some notable ones such as Africa Check in Africa; SMHoaxSlayer and Boom in India; GoHoo in Japan; Les Decodeurs in France; Pagella Politica in Italy; Aos Fatos in Brazil; PolitiFact.com and TruthOrFicton.com in the US. Most of these fact-checking groups have also been brought into an umbrella of collaboration by the Poynter Institute, under the title of the International Fact Checking Networking (IFCN).
The impact of fact-checking has been debated substantially, and it is not my intention to enter that debate. As it relates to our discussion here, however, it is worth pointing out that a 2014 study by Brendan Nyhan and Jason Reifler documented that legislators can be substantially less likely to disseminate false information when they believe that they are being fact-checked. Additionally, while the so-called “backfire effect” was often cited as a negative impact of fact-checking, Thomas Wood and Ethan Porter demonstrated that this effect is generally unsubstantiated in a research piece in 2016.
However – and importantly – the one group that has been most markedly affected by fact-checking appears to be journalists themselves, a conclusion reached by Michelle Amazeen in a study published in 2013.
Concurrent with this rise in fact-checking has of course been the steady decline of trust in traditional, mass media. According to a Gallup News Poll in September 2016 (notably before the election of Donald Trump), Americans trust and confidence in the mass media “to report the news fully, accurately, and fairly” dropped to an all-time low, with just 32% saying they have a great or fair amount of trust in the media that year. Gallup has been asking this question since 1972, and the “great or fair amount of trust” rating hit its highest level in 1976 at 72%. The “trust” mark stayed in the 50s for most of the 1990s and early 2000s, before dropping to 40% in 2015 and the most recent 32% in 2016.
Interestingly, however, many traditional media organizations believe that the rise of fake news will accrue to their benefit, presumably because the need to go to “trusted” sources of news will grow. According to a poll conducted by Statista in 2017, 70% of editors, CEOs, and digital leaders of global online news organizations believe that the distribution of fake news would, in fact, strengthen their organizations.
It is into this environment, then, that the British general election in 2017 occurred. The growing rise of “fake news” in the US election the fall prior had seen a concurrant escalation in “fact checking”, especially by 'outside' or non-profit organizations as there was also a growing distrust of traditional media.
Organizing, Supporting, and Funding Entities
For the general election in the UK in 2017, Full Fact and First Draft News joined forces to counter the dissemination of fake news. As Matteo Moschella and Ryan Watts of First Draft News described after the election, fact-checking and verification had historically been viewed as different specialties within journalism, but they decided that that distinction no longer made sense (Moschella, Watts, 2017). Importantly, First Draft News and Full Fact also partnered with Facebook and Google News Lab for the UK election focus.
First Draft News
In 2015, an organization in the US called First Draft News formed a nonprofit coalition to verify content in the social media world. The effort was based out of the Shorenstein Center on Media, Politics, and Public Policy at the Harvard Kennedy School, and it enlisted assistance from partners in the media, academic, and technology universes.
Media partners included NBC News, CNN, TrinityMirror, Gannett, Reuters, the New York Times, PBS NewsHour, Bloomberg, MSN, AlJazeera, BBC News, Sky News, The Guardian, Fast Company, AFP, and ABC News. Academic partners included the USC Annenberg School for Communication and Journalism, the City University of London, the Fletcher School at Tufts, George Mason University, Hamburg Media School, Temple University, and Cardiff University. Technology partners included FaceBook, YouTube, and Dataminr.
Since its creation, First Draft News has worked on news verification around events such as the elections in France, Germany, and the UK; the hurricanes in North America in 2017; as well as specific pieces on Chinese social networks, and efforts in South Korea to address fake news (firstdraftnews.com, 2017)
A similar nonprofit organization was created in the UK in 2009. Called Full Fact, it had a similar mandate of verification of news stories, although predominately with a UK and EU focus. Full Fact was and is backed by funding from the Omidyar Network (seeded by eBay founder Pierre Omidyar) and Open Society Foundations (seeded by hedge fund manager and philanthropist George Soros).
Another of Full Fact’s missions is to develop automated fact-checking tools. One of these is Live which, in real-time, verifies statements against a database of verified facts. Another tool is called Trends which tracks and displays false news information (fullfact.org)
First Draft News/Full Fact were not the only major group to fact-check the British general election in 2017. As Dr. Jennifer Birks of the University of Nottingham has demonstrated, Channel 4 News’ FactCheck and BBC Reality Check were also very focused on fact checking (Birks, 2017).** As many are partnered with First Draft News and Full Fact, it is likely that the organizations supported one another's work through citation, mentions, web traffic direction, etc.
Participant Recruitment and Selection
The collaboration was run by a team of 25 individuals selected for their expertise in social media trend analysis and fact-checking. Both organizations hire their own fact-checkers and everything is done 'in-house'. For their part, Full Fact utilizes a methodology that requires two separate individuals to review every claim, with many of their fact-checkers on three-month secondments from the Government Statistical Service. Similarly, reviewers for First Draft include academics, economists, professional fact-checkers, legal experts, etc. who are often recruited to volunteer their time with the organization. Some of these individuals were specifically asked to lend their expertise during the month-long General Election project run by First Draft and Full Fact.
Methods and Tools Used
First Draft and Full Fact's collaborative initiative had two inter-related goals: 1) monitor and identify news items and sources, and 2) verify those news items and provide the public and news media with up-to-the-minute analysis and reports of their veracity. The first goal was accomplished using various technological tools such as Trendolizer, CrowdTangle, and Newswhip which use algorithms to determine what content is gaining popularity among users across all social media platforms.
The second goal - fact checking - was performed by a team of 25 individuals most of whom came from the organizations' employees or were brought on board specifically for the month-long General Election project. Two newsletters were sent to members of the public and all the major media partners: the morning piece highlighted "trending" items and stories and the afternoon piece contained specific claims that had been fact-checked by the team (Watts, Ma, Dias, 2017).
What Went On: Process, Interaction, and Participation
For the 33 days leading up to the UK election on June 8th, First Draft and Full Fact collaborated. This collaboration took the form mostly of monitoring online conversations, a collection that grew, by June 8th, to 402 sources and topics; 34 million tweets; and 220,000 Facebook posts (Watts, Ma, Dias, 2017).
One tool they used was Trendolizer, which allowed them to monitor any content across all social media platforms that were gaining traction at any given time. Facebook’s CrowdTangle helped them identify any content that was “over-performing” on a given page (e.g. receiving more interaction than normal). Newswhip Spike spotlighted top performing articles across all platforms, and Trendsmap logged the growth and geographic dispersion of hashtags and keywords on Twitter (Watts, Ma, Dias, 2017).
There was a combined fact-checking team of 25 professionals, which included both full-time First Draft / Full Fact employees and other respected professionals who were brought in to the project for this 33 day period.
The group settled on producing two newsletters per day, which went out to all the major media, with the morning piece addressing broader trends that were receiving the greatest attention, and the afternoon piece addressing specific claims that had been fact-checked by them (Watts, Ma, Dias, 2017).
The group’s greatest focus was on topics related to the British general election, including Brexit, employment, climate change, immigration, education, pensions, and the NHS (Watts, Ma, Dias, 2017).
Interestingly, the group highlights that there were many articles, memes, photographs or other content that they didn’t want to address. They would have given “oxygen” to this content that otherwise wouldn’t have garnered much attention (Watts, Ma, Dias, 2017).
Influence, Outcomes, and Effects
It is difficult to quantify the success of a fact-checking campaign against fake news. What can be identified are the trends in fake news and the trends in the responses to fake news.
First Draft News/Full Fact, in a piece after the election, broke down the types of fake news in the 2017 British general election into five categories:
(a) Misleading headlines in mainstream newspapers
(b) Exaggeration and misinformation by the hyper-partisan press
(c) Hyper-partisan websites attacking the mainstream media
(d) Official political party pages bending the facts
(e) False rumours that got traction before being flagged by us or others (Busby, Khan, Watling, 2017)
In a separate piece, Professors Susan Banducci, Dan Stevens, and Dr. Travis Coan documented the use of fake news during the 2017 British general election. According to their research, the three most commonly heard misstatements during the election campaign were:
- due to immigration pressures, the NHS had an additional 1.5 million patients in the prior three years
- An increase of “thousands to millions” of the users of foodbanks
- That employment was at a historical high due to the insecurity of many of those jobs
According to their analysis, over 50% of respondents found the first two claims to be believable, and over 40% found the third claim to be believable (Banducci, Stevens, Coan, 2017).
First Draft News has identified seven types of mis- and dis-information, or what we have chosen here to call “fake news”:
(a) Satire or parody: No intention to cause harm but has potential to fool
(b) Misleading content: Misleading use of information to frame an issue or individual
(c) Imposter content: When genuine sources are impersonated
(d) Fabricated content: New content is 100% false, designed to deceive and do harm
(e) False connection: When headlines, visuals or captions don’t support the content
(f) False context: When genuine content is shared with false contextual information
(g) Manipulated content: When genuine information or imagery is manipulated to deceive (Wardle, 2017)
By these standards, however, the British general election of 2017, was materially different from the US election in 2016. As has been well documented, in the 2016 US election, “fake news” was led by bot- and troll-driven plants of news stories and by false websites. By contrast, as Mattha Busby, Isaan Khan, and Eve Watling of First Draft News describe, “unlike the US election, the most misleading content [in the British election] didn’t come from newly created websites or automated accounts created to push disinformation.
Instead, misinformation in the UK election came from misleading headlines, graphics, and statistics from the mainstream press, political parties, and hyper-partisan websites.” In fact, they go on to advise that those trying to combat fake news stop focusing on fabricated news sites and focus more on the current technology of memes, in which vivid images have brief amounts of text overlain on them (Busby, Khan, Watling, 2017).
Analysis and Lessons Learned
The collaboration between First Draft and Full Fact highlights two inter-related lessons for effective news verification: first, that the style of the content and the means by which it is translated or disseminated are constantly changing and, second, that it is therefore imperative to use “real time” technology for the sourcing and subsequent fact-checking of fake news. In their post-election pieces, the team from First Draft News acknowledged the importance of their technology tools – like Trendolizer and CrowdTangle – in being able to immediately identify all news stories that were garnering attention. From there, their team of fact-checkers was able to respond relatively quickly with verdicts regarding the veracity of those stories (Moschella, Watts, 2017).
Another lessons to be learned from this initiative is that citizens are the front-line in the war against fake-news. Since they are the primary targets, they are the best situated to stop the spread of fake news. In his book “The Truth Matters: A Citizen’s Guide to Separating Facts from Lies and Stopping Fake News in Its Tracks”, Bruce Bartlett recommends citizen education as an important defence against fake news. Citizens - including politicians - must be educated on the critical analysis of content, the use of fact checking sites, and the identification of reliable and/or impartial news sources.
"Fake news” – in all its forms of misinformation – has been around for virtually as long as humankind has been documenting events. As we have also seen, technology has made the spread of “fake news” ever more substantial than it was in the past. It seems that the best way to address this spread is a combination of work – (1) the use of technology by organizations such as First Draft News and Full Fact; (2) even greater citizen engagement in the analysis of the news and of their news sources.
Adair, B. and Stencel, M. (2017) “How we Identify Fact-Checkers – Duke Reporters Lab” https://reporterslab.org/how-we-identify-fact-checkers/
Allcott, H. and Gentzkow, M. (2017) “Social Media and Fake News in the 2016 Election”. Journal of Economic Perspectives, Volume 31, Number 2, Spring 2017, 211-236
Amazeen, M. (2013) “Making a Difference? A Critical Assessment of Fact-checking in 2012”. New American Foundation
Banducci, S., Stevens, D., Coan, T. (2017) “Should we worry about fake news?” eletionanalysis.uk. http://www.electionanalysis.uk/uk-election-analysis-2017/section-3-news-and-journalism/should-we-worry-about-fake-news/
Bartlett, B. (2017). The Truth Matters: A Citizen’s Guide to Separating Facts from Lies and Stopping Fake News in Its Tracks Berkeley: Ten Speed Press
Birks, J. (2017) “Fact-checking the election”. Electionanalysis.uk http://www.electionanalysis.uk/uk-election-analysis-2017/section-3-news-and-journalism/fact-checking-the-election/
Busby, M., Khan, I. Watling, E. (2017) “Types of Misinformation during the UK election” https://firstdraftnews.com/misinfo-types-uk-election/
Canduci, A. (2009) History’s Greatest Lies: The Startling Truths Behind World Events Our History Books Got Wrong Boston: Fair Winds Press
Fabry, M. (2017) “Here’s How the First Fact-Checkers Were Able to Do Their Jobs Before the Internet”, time.com http://time.com/4858683/fact-checking-history/
Greenemeier, L. (2017), “Could AI Be The Future of Fake News and Product Reviews”, ScientificAmerican.com https://www.scientificamerican.com/article/could-ai-be-the-future-of-fake-news-and-product-reviews/
Keyes, R. (2004) The Post-Truth Era: Dishonesty and Deception in Contemporary Life New York: St. Martin’s Press
Monerat, A. (2017) “The Bet on Fact Checking: Journalists create more initiatives to verify public discourse and reveal fake news” Knight Center for Journalism in the Americas. https://knightcenter.utexas.edu/blog/00-18252-bet-fact-checking-journalists-create-more-initiatives-verify-public-discourse-and-reve
Moschella, M., Watts, R. (2017) “UK Election: What we learned working with Full Fact” https://firstdraftnews.com/joint-venture-learnings/
Nagle, B. and Burstein, S. (2009) The Ancient World: Readings in Social and Cultural History 4th Ed. New York: Pearson
Nyhan, B., Reifler, J. (2014), “The Effect of Fact Checking on Elites: A Field Experiment on US State Legislators”, American Journal of Political Science
Swift, A. (2016) “Americans Trust in Mass Media Sinks to New Low”, Gallup News http://news.gallup.com/poll/195542/americans-trust-mass-media-sinks-new-low.aspx
Watts, R., Ma, A., Dias, N. (2017) “How we fact-checked the UK election in real time” https://firstdraftnews.com/uk-real-time-factchecking/
Wardle, C. (2017) “Fake news. It’s complicated”. https://firstdraftnews.com/fake-news-complicated/
Wood, T. and Porter, E. (2016) “The Elusive backfire effect: mass attitudes’ steadfast factual adherence” Social Science Research Network
Young, K. (2017) Bunk: The Rise of Hoaxes, Humbug, Plagiarists, Phonies, Post-Facts and Fake News Minneapolis: Gray Wolf Press
*For those who are interested, Hunt Alcott and Matthew Gentzkow have probably some of the most comprehensive research regarding fake news and the US election.
**As an aside, one of Birks’ interesting conclusions was that BBC Reality Check examined claims by both parties equally, whereas Channel 4 News’ FactCheck and First Draft News/Full Fact tended to fact-check more Labour claims than Conservative ones (Birks, 2017)).