It took just a day for the Ukrainian site stopfake.org to published details of the real origin of the photo – one of hundreds of fake reports that the website has debunked since it started in March 2014.

The StopFake site is run by a small group of current and former students from the Mohyla School of Journalism in the Ukrainian capital, Kiev. When pro-Russian forces started to invade Crimea back in February, the school’s director, Yevhen Fedchenko, called students and alumni together for a brainstorming session.

“There was a lot of Russian propaganda and very little hard news,” Fedchenko said to onMedia via Skype from Kiev. “One of our students suggested starting a website debunking propaganda, and so we immediately decided to set up StopFake on WordPress.”

The amount of disinformation about Ukraine is staggering. Russian media, much of which is state-controlled, has been broadcasting for months what the New York Times has called “bluster and hyperbole, … misinformation, exaggerations, conspiracy theories, overheated rhetoric and, occasionally, outright lies about the political crisis in Ukraine.” Other media organizations, from the BBC to the Guardian and Germany’s Spiegel magazine have written recently about Russia’s “relentless propaganda war”.

Because of the sheer flood of Russian-language misinformation, StopFake curates and refutes false claims spread only by Russian media and doesn’t examine false claims potentially made by the Ukrainian media or government.

The fake-busting site has proved hugely popular. In its first few weeks of operation, it had 1.5 million visitors and some of the weekly video round-ups detailing the fakes have more than 100,000 views on YouTube.

Verifying images

Photos and videos aren’t that difficult to check, says Fedchenko. The journalists working for the site do this by looking at metadata which can reveal the date and location the footage was shot, in combination with search tools like Google or the Firefox add-on, Who stole my pictures?.

This AP Photo by Guillermo Arias shot is Mexico was published on Politikus.ru as showing dead Ukrainian militants in Sloviansk

“There are pictures or footage supposedly shot in Ukraine which actually come from Iraq, Syria, Mexico and even from Tiananmen Square in China and they were quite easy to debunk,” says Fedchenko.

Stopfake.org have even prepared a page with information on how readers can identify a fake for themselves in the hope that people will become more critical about the information they find on the Internet.

For other stories, the journalists rely heavily on that basic journalism skill – fact checking. For example, the Russian President Vladimir Putin recently claimed one in three Ukrainians – or thirty percent of the population – was working in Russia. A couple of clicks through official statistics and Stopfake.org proved – with screenshots – that only seven percent work in Russia. That’s a massive difference.

But other rumours are more difficult to verify, says Fedchenko, especially when false information is mixed together with real facts. Earlier in June, a bomb was dropped on story the city of Luhansk. One theory was that it was the Ukrainian airforce, the other was it was terrorists who stole the plane.

“We took all the hypotheses and collected all the videos and analysed them from different angles and tried to give explore the different theories,” he explained. “It was very popular.”

Getting people involved

Another success is the bright red “Report a Fake” button on the site’s homepage. By clicking on the button, visitors can give details of potential fakes. When they first added the button, the group were worried that it would be misused by so-called trolls – people who are paid to leave comments on the Internet. But, says Fedchenko, they have received “tons of feedback” via the button and it has the added advantage of giving users a way of participating in the project.

The website was originally targeted at Russian speakers in the Crimea, where independent Russian language media was shut down at the beginning of the invasion. However, it seems Russians are also hungry for independent information, with one third of Stopfake.org’s visitors coming from Russia. They are also very active in reporting fakes, Fedchenko says.

The site has an English version too – it was important for the team to give Ukraine a voice outside of their country. After several journalists were detained in Eastern Ukraine, many Western media organizations, including Germany’s public ARD and ZDF broadcasters, have recalled their correspondents.

Despite StopFake’s popularity, most of the team work as journalists for other media organizations and can only work on the site in their spare time. They have received some 8,000 euro in donations, but are otherwise paying expenses out of their own pockets. The hope is that in the future, crowdfunding will enable them to payroll full-time staff.

Template for others

Financial issues aside, Fedchenko says the project is easily replicable in other countries with similar problems. The team have the skills expected of any online journalists – writing, editing, photo-cropping and resizing, video editing and creating infographics for the weekly video wrap-up.

“We do the same things as anyone else would do – find the truth, check the facts, talk to people, and these are universal ways to improve journalism in any country.”

StopFake won the People’s Choice for Russian section of Deutsche Welle’s Best of the Blogs Awards – better known as the BOBS.

Related Posts

Verifying facts in the digital age

How ARD, BBC and CNN verify social media content

Checking the facts in Africa: AfricaCheck

Written by Kate Hairsine

Photo: Munir Uz Zaman/AFP/Getty Images

Critics of using social media to crowdsource news argue that people on the street aren’t objective and that their impressions of an event can’t be trusted. But in this guest blog post for onMedia, journalist Ranty Islam argues crowdsourcing is a key journalism tool that stems from a tradition older than the media itself – the bustling marketplaces and squares that have been used for information exchange since time immemorial.

There’s no truth. Only best guesses. The trend to use social media to crowd verify news is a reminder of how people have always shared and used information. To be useful, facts have to be reliably true enough. And this is the basis of oral traditions which have been around for centuries, if not millenia, in every culture. It’s also the basic operating principle of crowdsourced news verification. Both tap into a collective memory that surrounds any time and place where people are present to witness an occurrence.

So when news agency Storyful polls its network via social media about the veracity of a YouTube video claiming to document government crimes in the Syrian war, this isn’t too different from medieval town squares where storytellers and bards passed on news collected from battle fields or royal courts far away. Your Storyful dashboard and the square only really differ in the size and speed of the network. This means today there are more checks and balances in place to hopefully lessen the number of false judgments.

When it comes to crowdsourced news verification, a knee-jerk reaction from newsroom veterans often sounds something like this …

“How can you trust the subjective testimony of individuals who may be biased or even actively pursuing an agenda, who do not know how to research or cross-check information and who are not journalists?”

The answer is: you can’t.

Image – (CC) Gavin Llewellyn/Flickr

But instead of asking just a few John Does grinding their own ax, you are able to ask hundreds or even thousands of individuals with the click of a button, you will get many different views. These competing perspectives and variety of biased perceptions will ideally drown each other out. Let’s call this ‘noise’. Because they have all witnessed the same event, however, some sort of common theme will surface. Let’s call this the ‘signal’.

The parallel with telecommunications can be pushed further. To get a better ‘signal’, you need a greater input, larger sampling and more exposure. Similarly, the bigger the crowd, the better are your chances of sniffing out a common thread, i.e. a verified version of an event. (Also take a look at the Verification Handbook that’s just been released by the European Journalism  Centre.)

However, there are at least two caveats here. To start with, ‘verified’ does not mean it’s ‘true’. If a piece of news has been crowd verified, it just means there is a higher probability it conveys something that has really happened, and therefore is more likely to be true. It’s a best guess, but not the truth. The second issue is that there may be some systemic distortions. In some places, for instance, particular segments of society are less likely to be Facebook or Twitter power users simply because they don’t have regular internet access, can’t afford a smartphone or are less tech-savvy. If these strata of society additionally hold views that are very different from, let’s say, those of articulate, social media savvy, smart-phone wielding urbanites, we have a problem. Verifying news by just polling Twitter users under these circumstances would introduce a selection bias – when ideally more and different channels need to be used.

If these problems can be kept in check though, there’s no reason to believe crowdsourced news verification is less reliable than traditional journalism. In fact, it may score better. The Arab Spring playing out on the streets of Cairo or the fighting in Syria are examples of how crowd based reporting and verification can break news and provide vital background information before traditional journalism.

On the other hand, traditional journalism has been undoing itself. Its often stated commitment to truth, impartiality, objectivity and professional scrutiny are increasingly difficult to maintain in a world which is becoming faster and more complex by the day and where those who are traditionally reported on, now talk back.

A social media-ted public makes itself heard in a way that can naturally be harnessed by crowd-based reporting. But is also forces traditional journalists to act more transparently. It is much harder to buy into the objectivity of foreign correspondents today once you know they draw on their own mere handful of informants. Even a decade ago, this fact may have gone unnoticed.

And impartiality is at long last revealed for what it is: a fiction. Twenty years ago, the likes of the New York Times and the BBC were unquestioningly trusted as authoritative sources of unbiased information. In the massively interconnected world of today they are mere instances (albeit important ones) of a particular culturally and historically conditioned vantage of the world from North of the Equator and West of Greenwich.

This then, is where crowd based reporting really comes into its own. In contrast to traditional journalism, it puts together its story from those who are living and experiencing it rather than those who are observing it. And in a couple of decades from now, the notion that a few professional, trustworthy institutions can reliably inform people about the world will be viewed as an odd historical aberration – wedged, as it is, between centuries of oral traditions and the 21st century take on the same thing.

Ranty Islam is a digital anthropologist and an editor at DW for Global Ideas, a leading, award winning multimedia program covering climate protection and biodiversity around the world. Follow Ranty on Twitter: @2ranty

In this blog post, I’m going to talk about WHEN we can be happy that a story has been verified enough to publish it. It’s an issue which comes up often in DW Akademie trainings I’ve been part of. I think it’s a key skill for journalists to master.

If you are interested in HOW to verify, have a look at this page on verifying news tips from Twitter or this post with a list of useful sources. Also check out the amazing detective work of this journalist from Storyful – a social media fact-checking agency.  And keep your eye out for this publication coming out early in 2014 looking  at verifying user-generated content during emergencies.

How sure is “sure”?

When I was first learning about journalism, it seemed pretty simple to know when you could publish or broadcast a story – you could run it if you had two independent, trusted sources or one of your own correspondents as an eyewitness.

It soon became clear to me, however, that things in the real world are more complicated than that.

Below is a scenario we often give to participants in DW Akademie trainings to tease out some of the complexities of when to run a story.

You are working at a radio station in a small town when you start hearing reports about a toxic waste spill in the local river that people use as a source of drinking water. You get calls and texts and people are posting about it and even uploading photos on Facebook and Twitter. People say the spill is coming from a factory owned by a prominent local politician. It’s just a week until local elections. You have one minute until your main midday news broadcast and so you don’t have time to contact anyone to check out the story.

Not easy to work out what to do, huh?

Concerns about public health and safety might lead you to broadcast something even though you don’t have time to verify the facts for yourself. Perhaps you’d issue a warning not to drink the water but include phrases like, “This is still unconfirmed…” or “We are hearing rumours…”

You obviously wouldn’t mention the theory that the spill is coming from a certain factory or the fact that the factory is owned by the politician. But in a small town, people are likely to know this sort of information anyway and are likely to come to their own conclusions once they hear your news bulletin. Your broadcast would give credence to the rumours people are already hearing – whether you like it or not.

I don’t think there’s any one right answer here. Most journalists I’ve done this exercise with decide to broadcast something like, “We’re hearing about a problem with the river water, making it unsafe to drink. We are looking into the reports and we’ll update you as soon as we have more information.”

The thought of someone getting sick because they missed the news is the overwhelming consideration for most journalists. On the other hand, the consequences for the politician could be horrendous if the whole thing turns out to be a hoax. Even if you correct the story later, considerable damage to the politician’s reputation may have already been done.

Screenshot source – CNN

Even journalists working for large media organizations get this stuff wrong – sometimes in highly sensitive situations. A day after the Boston marathon bombings in April, CNN announced someone had been arrested (watch the announcement here and read about it here). An hour later, other news organizations like Fox and AP also ran with the story. But in fact, the information turned out to be false: no one had been arrested at all and CNN especially came in for heavy criticism.

Part of the problem is that news organizations feel pressured to put out the news as fast as they can. We often fear our competitors will get the story out before us and win over the audience.

The question you need to ask yourself is: do you want to be “first and pretty certain” or “later but sure”?

Personally, I prefer to wait.

“In this age of instant reporting and tweets and blogs, there’s a temptation to latch on to any bit of information, sometimes to jump to conclusions. But when a tragedy like this happens, with public safety at risk and the stakes so high, it’s important that we do this right. That’s why we have investigations. That’s why we relentlessly gather the facts.” US President Barack Obama speaking after the CNN debacle

Be it the death of Osama bin Laden, the emergency landing of a plane on the Hudson River or armed conflict in Syria, photos and videos made by eye witnesses usually reach the public as initial evidence through breaking news. Today, media organizations are virtually flooded with digital content from all over the world which makes it even more important to pay attention to the sources of information. That is why large media organizations have set up special research teams to verify the content from social networks. Although most of them follow the same rules, it is worthwhile to compare the separate approaches. Konrad Weber shows how renowned international media outlets such as ARD, BBC, CNN and others check the content coming from social media.

BBC

A UGC Hub has existed at the BBC since 2005. Its primary task is to verify pictures, news and videos that come from the users and viewers. Initially, the team mostly focused on the information mailed to the BBC on a daily basis. But today people are more likely to distribute the information themselves through different social networks. As a result, the number of contributions has dropped to about 3,000 a day, and the Hub, which has in the meantime grown to 20 staffers, is now focused on self-researched materials in the social web.

The rule of thumb for the Hub staffers is: always get on the phone or contact the author via Skype. By doing this, the journalists cover two issues. First, they can query the source on the circumstances under which the photo or video was taken. Second, they request the source’s permission, as the copyright holder, to use the material in their coverage. The journalists clarify the following issues:

• Where was the footage uploaded originally?
• When was the footage uploaded?
• Are there any special reasons for which the footage was uploaded?

However, the footage might be posted anonymously for safety reasons, which makes it difficult to contact the author. In that case, the Hub staffers have to go through other verification steps:

• Contents: Does the location actually seen in the footage match the indicated location? Are languages or dialects in the footage understandable? To check these issues, the Hub staffers turn to local specialists such as field reporters, immigrants or guides.
• Technical issues: Possible changes in the footage and its local nature are traced.

Thus, in May 2011 the Hub disproved a fake image of Osama bin Laden’s head that circulated online. It turned out to be a photo of another corpse’s face, onto which bin Laden’s features were grafted using Adobe Photoshop. The journalists used TinEye to search for images similar to the photo. The same search can also be performed with Google Picture Search.

Storyful

The news agency Storyful employs 23 staffers and is specialized on separating “actionable news from the noise of the real-time web”. The agency has developed a checklist to verify footage from the web:

• Can the footage be geo-located? Are there any landmarks that help to verify the location via Google Maps or Wikimapia?
• Are streetscapes similar to geo-located photos on Panoramio or Google Street View?
• Do weather conditions correspond with reports on that day?
• Are shadows consistent with the reported time of day?
• Do vehicle registration plates or traffic signs indicate the country or state?
• Do accents or dialects heard in a video tell us the location?
• Does the information correspond to the other imagery and reports people are uploading from this location (e.g. via Twitter Advanced Search?

The Storyful staffers have also created an additional checklist to verify the sources:

• Where is this account registered and where is the uploader based, judging by his or her history?
• Are there other accounts – Twitter, Facebook, a blog, or website – affiliated with this uploader? How can they help us identify location, activity, reliability, bias and agenda?
• How long have these accounts been in existence? How active are they?
• Does the uploader write in slang or dialect that is identifiable in the video’s narration?
• Can we find WHOIS (domain registration) information for an affiliated website?
• Is the person listed in local directories? Does the person’s online social circles indicate a proximity to the story/location?
• Does the uploader “scrape” videos from news organizations and YouTube accounts?
• Are video descriptions dated? Does the title of the video have file extensions such as .AVI or .MP4?
• Are we familiar with this account? Has the content and reporting been reliable?

CNN

When CNN launched its citizen journalism platform iReport in 2006, the company probably didn’t expect over a million of iReporters who would register on the platform over the next six years and 2.4 million unique users each month.

All iReporters contributing to the platform with their tips, photos and videos are unpaid. Still, the quality of the submitted content is often high and CNN has made use of it in its coverage. iReport’s team of eight full-time producers receives more than 500 pieces of content on average each day. If a submission could be used in whole or in part by a CNN property other than iReport, the iReport team verifies it thoroughly.

That’s where a two-step verification process comes into play:

• Verification begins with an iReport producer reaching out to the uploader in order to ask questions by phone, email, Skype, etc. The author of the footage is asked about their actual location, the camera used to take the footage, what they think about the event and if they have been paid for reporting. At this stage, journalists also hope to find out additional information on the event, which can be included in the coverage.
• As a second step, the iReport staffers verify the footage itself.  They resort to CNN-ers in the field, subject-matter experts, affiliate networks, local media and local police reports to verify all the aspects of the story. The journalists also cross-check what they have learned from citizen journalists with other social media reports and use technology.

Reuters

Let us have a look at how English news agencies verify their sources in the social web. Whereas Associated Press (AP) uses a clear verification model to check the authenticity of pictures, Reuters provides its reporters with just a few general tips on dealing with information from Twitter.

The Reuters Handbook of Journalism explains that reporters should be aware of the dangers of the social web. There are, however, no concrete guidelines on the verification process.

ARD

In contrast to Anglo-Saxon media, German media organizations are just beginning to implement verification processes in their work. For example, the German organization of regional public-service broadcasters ARD developed a verification checklist in April 2011 which is used by a journalist in the company’s Content Center. Basically there are four steps in the verification process:

• Examining metadata (exif data, time of shooting, Google Maps, weather)
• Checking the source’s credibility by contacting the source. At this step, journalists also ask for permission to publish the footage.
• Check through experts (ARD reporters at the location, language specialists, Skype contacts)
• Technical check (image editing, sound)

ZDF

Reporters at ZDF approach the verification process in a more flexible way. For example, some staffers use Facebook groups to crowd source information without bringing the sources into danger. Emergence teams may be set up when reporting on crises.

Author: Konrad Weber

Original publication: Wie ARD, BBC und CNN Inhalte aus dem Social Web verifizieren

Translation: Natalia Karbasova