Breaking news is when social media comes into its own – displaying both its best and worst qualities, instantly. On the one hand, it allows for the near real-time dissemination of stories as they happen, distributing essential information that goes beyond pure news: it is a data-driven, 21^st century equivalent of an air-raid siren going off. At the same time, this very immediacy co-opts us into being participants rather than just distant observers or bystanders: we are the media, whether we are in Paris, or in front of a computer thousands of kilometres away. We participate through supplying content, comment, analysis, and antagonism; we participate through the expression of grief or shock or outrage, and through sharing those same responses with others.

But sometimes our desire – our need, even; and this is a very human response – to participate, or be seen as contributor-participant outweighs our other more critical functions: to stop and think about [the impact of] what we are seeing, what we are saying, and what we are sharing. Social media platforms are, by their nature, designed to ensure you share first and think later. This is problematic on many levels.

Even as events in Paris were unfolding, the terrible gap between social media intention and effect was on full display: a hashtag (#PorteOuverte, meaning “open door”), designed to help Parisian Twitter users find nearby places of safety, quickly became overrun with endorsements posted by well-meaning Twitter users nowhere near France, or Paris, or the affected arrondissements. These solidarity Samaritans generated so much white noise that it would have been almost impossible for anyone actually affected by the attacks to use the hashtag for its created purpose at that time. [As a pointer: this kind of issue can easily be circumvented by posting info about a hashtag, without actually using the hashtag, through embedding the text/tag in images like this one]. This is a direct illustration of how even the benign misuse or appropriation of seemingly passive content can have unintended consequences.

For active content, the consequences become incrementally significant – although the behaviour patterns on social media remain much the same. The urge to share or, perhaps more pressing, to be shared, overrides almost all other cautions. The result is a cluttered and frenetic digital landscape where, for many, it can be almost impossible to distinguish the relative value (or cost) of different content. When bombs and gunshots go off, everything becomes equally portentous, and equally meaningless.

The blind broadcasting, sharing or republishing of active content during a mass-event emergency situation is not a new phenomenon. Nor is our understanding of how damaging this can be. We saw this during the Boston marathon bombings, for example, where a missing student, Sunil Tripathi, was mistakenly identified on social media as one of the suspects. His name, and photograph, were widely shared by both citizen and professional media accounts, many posts adding the further broken-telephone error of stating that the Boston Police Department had confirmed Tripathi as a suspect (they had not). Tripathi’s family, who were frantically searching for their missing son and relative, had to endure days of resulting hostility, even once the actual perpetrators were identified, and eventually discovered that Tripathi had, earlier, taken his own life.

In recent years media advocacy organisations have compiled disaster response verification checklists, to try and encourage a more critical and objective approach to how we engage with and share breaking news online, particularly when large numbers of lives and deaths are involved. Although these typically describe and prescribe digital platform processes, they are all based on what should be straightforward common sense: that breaking news (particularly bad news) can get muddled up, even by professional news organisations; that unidentified or anonymous sources should be treated with extreme caution, as should unverified “reports of…” incidents; and that conspiracy theories, typically, pan out to be just that: theories.

In the case of the recent Paris attacks, the formal French media did an excellent job of publishing frequent, responsible updates under incredibly difficult conditions – sources were clearly cited, unconfirmed information was listed as such. It was these media agencies, including France24, Le Figaro, and Agence France-Presse, which provided much of the content subsequently used by other international news organisations.

At the same time, there was an equivalent flood of unofficial content being posted online – citizens themselves caught up in the attacks or the aftermath, or speculators (including journalists not actually on the scene) simply watching the news and providing reciprocal broadcast, commentary and analysis.

The nature of the Paris attacks meant it was not always possible to contact the original sources or authors of many such posts – because they were still quite literally caught up in the attacks themselves, and it would not have been safe to attempt to speak to them (I don’t imagine for a second that stopped some media from trying). In one particular instance, a man named Benjamin Cazenoves appeared to be posting on Facebook, from inside the besieged Bataclan theatre. This first-person account was one of the first to alert us (my timeline, in any event) to the fact that the theatre might be a massacre rather than a hostage situation. At the time, however, French security forces were still outside the Bataclan, and it was impossible to assess what was known rather than suspected-speculated. It was, however, possible – within less than one minute – to establish whether or not Benjamin Cazenoves existed (by checking his Facebook account), and whether or not he was who he claimed to be (a quick read through his older timeline indicated, for example, that he had posted about the Eagles of Death Metal – the band playing that evening at the Bataclan – a few days earlier).

This one-minute shallow verification indicated a good possibility that this was a real person and a genuine account, caught up in the tragic events that were unfolding. Not to put too fine a point on it, but in a real emergency humans behave badly, and it is far from uncommon to encounter fake accounts (and, indeed, it appears fake accounts for Benjamin Cazenoves have been created subsequent to the attacks).

Performing an initial verification on a single Facebook user was relatively simple. Even then, based on the way Cazenoves’s content was broadcast, many on social media, including journalists and news organisations, seemed to have shared the information without necessarily having repeated this checking process for themselves (or, if they had checked, they gave no indication of this). In other words, they relied on the assumption that other users would or might have already checked, and simply re-broadcast based on that assumption.

In my experience there is a very short list of things that do not need to be (personally) checked/authenticated before reposting:

1. Opinions (even here, you should read the entire piece – in case, say, it endorses apartheid – to make sure you know what you are posting; also, always verify the author when reposting from a screenshot, or a RT).

2. Cat memes (again, note that cat and other cute animal pics may not be kosher, and may involve cruelty to animals).

Fact-checking is about simplicity and consistency. Probably for this reason, fact-checking is often interpreted as the petty application of discipline. In the past few days, I have been called both a “Cassandra” (fondly) and a “prefect” (or, “hall monitor” if you like). Part of being a teacher of fact-checking is studying how such processes work; it’s no surprise to me that people do not like being corrected, or told what to say or what not to say. But fact-checking is not about monitoring, or censoring. It is about transparency. Post what you like, but say where it comes from, share your sources, be open about your process. In this way, content sharing becomes responsible rather than just reflexive. Why does it matter? This is at the heart of much of my training work: not just explaining how to fact-check and verify, but to provide constant reminders of why we need to do so.

None of us – even those of us with decades of experience as professional journalists –can accurately predict which information will wind up where, and what the impact of such information might be. The premise for Sunil Tripathi’s online lynching originated from just one person, who knew Tripathi, but had not seen him since high school, and who made an identification based on poor quality CCTV footage, before sharing it on a single social media platform (Reddit).

If we look at the events that took place in Paris, what might have been the consequences of too-rapid sharing of unverified social media content? Would a hoax account posting about hostages at the Bataclan have swayed or influenced the operational decisions of security forces (or the attackers themselves)? Could unverified social media posts have contributed to public outrage that would, in turn, have put pressure on politicians and security forces to act in a particular way? What if an account had posted fake information about another attack in another area of Paris? Would security forces have been deployed (and diverted from real events) to deal with such a hoax?

These are realities, if not in Paris then in many other mass events. [Author’s note: Indeed, shortly after the Paris attacks, a doctored image, claiming to be that of one of the terrorists, was released and widely re-published. The individual targeted in the hoax, Canadian Veerender Jubbal, is a vocal opponent to GamerGate].

In addition we know, from repeated studies of fakes, hoaxes, and plain old bad information, that bad facts not only contribute to bad decision-making; they can also incite fear, paranoia, and suspicion. The latter is very much in evidence with speculation around those responsible for the Paris attacks: whether or not they shouted the name of Allah as they opened fire at the Bataclan; if, indeed, a Syrian passport conveniently managed to survive the detonation of its suicide-bomber owner, to be found right next to the body…. Another odd lesson in human behaviour is that conspiracy theories are almost always more plausible than boring old facts. But it is on the simple facts where, immediately, we fall down.

In the hours after the attacks had ended, the online rhetoric machine juggled breaking news with newer constructs of solidarity, and increasing scrutiny. Images of crowds gathered after the Charlie Hebdo shootings were quickly scraped and reposted, labelled as if they were from November, instead of January (a dateline scrape). One of Donald Trump’s many offensive tweets, about France and gun control (also made in January), was similarly repurposed, and reposted, apparently leading to a much-retweeted (and subsequently deleted), angry response from Gerard Araud, the French ambassador to the United States.

Thousands of composite posts were soon being shared showing international landmarks lit up in a tricolour of red, white and blue – the spire of One World Trade Centre in New York, the Sydney Opera House, the statue of Christ the Redeemer in Rio de Janeiro. Except that some of the images (Cristo Redentor for one) may have dated back months, even years. This timeline/image faking and sharing extended beyond the immediate epicentre of Paris, to other stories that seemed connected – even if only by concentric tragedy. When artist and musician Meredith Yayanos noted that a photo depicting explosions in Beirut (where 43 people had been killed in twin suicide bombings the day before the Paris attacks) was nearly a decade old, and not from 2015 at all, one Twitter user railed against her correction: “… is pointing that [out] either helpful or kind?” they asked. There it is again: the suggestion that fact-checkers are the killjoys of social media, the hall monitors, the uninvited police. Possibly a sub-type of Grammar Nazi.

I often hear (when correcting an image, or more specifically the caption on an image) that intention matters more than facts. What is the harm, people seem to ask, in posting something that speaks to feeling rather than data. Fact-checking, I have been told, denies the importance of the lived experience

“When I speak to the depravity I don’t have a time frame. Whenever that img was captured is not a qualifier (4me),” a spokesperson-correspondent working on the Palestinian crisis told me, after posting an anonymously submitted image of what was claimed to be Israeli security forces taking a cellphone photo of a young Palestinian man they had just murdered. [As a rule, I personally verify images of Syria, Israel, and Palestine, before reposting – preferably including information about the photographer, the date of the event, and the location – because political and religious rhetoric and a dismissive “all middle eastern people/places look the same” approach have made this an exceptionally exploited area, with frequent fakes/hoaxes/scrapes].

But outrage is not a get-out-of-jail-free card for fact-checking; and neither is empathy, or sympathy, or even solidarity. There is nothing kind or benevolent about failing to check your information. It is not a charity to post bad data, however good your intentions are.

It is easy to see (or, at least, I hope it is) the potential hazards of reposting unverified images of dead bodies or explosions. It is often harder to explain why the same cautions should, absolutely and without exception, apply to passive content, to buildings coloured in different lights, to statues and lightning storms and hurricanes and animals… Why? Because the same processes should be applied to everything we share, irrespective of the relative graphic nature of the content. The reasons for this are not to induce some kind of fact-check paralysis, where verification makes it impossible to post anything, but because fact-checking only works when it is consistent. We cannot only apply fact-check standards when we think the news or content is important, but discard the same processes when we (subjectively) think it is not.

Facts work in partnership, not in isolation. The best story can unravel over one tiny mistake (just look at American presidential campaigns to see how easily this can be exploited). A small error can also become a metaphor for deeper flaws, and this process happens quicker than you think. Bad facts also travel faster than corrections, which means that – particularly for professional journalists and media organisations – we need to keep emphasising accuracy over agility. DM

Nechama Brodie is the head of Africa Check Tri Facts: Training, Research and Information

Photo: A Pakistani lights candles for the victims of the 13 November Paris attacks, in Karachi, Pakistan, 16 November 2015. EPA/SHAHZAIB AKBER.