How is the Live Feed of Christchurch How is It Affecting Facebook

It was around 1.35pm last Friday in Christchurch when the horrific video began. And a full 29 minutes before Facebook - the very platform on which it was hosted - even knew about it.

The massacre by an Australian white supremacist, captured on a GoPro headset and streamed directly to the world's largest social network, has now been well documented. Fifty innocent people lost their lives, dozens more were injured.  "That quiet Friday afternoon has become our darkest of days," New Zealand Prime Minister Jacinda Ardern said this week.

Funerals for some of the 50 victims of the mosque shootings started this week. Credit:Jason South

It was a deranged act, engineered to generate maximum publicity. Unfortunately, in that sense, it worked. And for the past week, everyone has been trying to understand how, and why.

Facebook said this week the actual live stream of the incident, which lasted for 17 minutes, was only viewed by 200 users. Yet due to a series of operational failures and the nature of the modern internet, millions more ended up seeing it.

Loading

The $US473 billion ($A663 billion) company's vaunted artificial intelligence technologies, designed to stop extremist and illicit content such as pornography or videos depicting violent acts from appearing on the platform, failed to detect anything objectionable in the Christchurch shooting stream.

It took until 29 minutes after the live stream began, and 12 minutes after it finished - when a user reported it- for Facebook to begin to realise what was happening.

At around 1.40pm the alleged shooter entered the Al Noor Mosque and began firing.  At 1.47pm, according to local reports, and just six minutes after the first emergency call was made, police had arrived on the scene. A little after 2pm, or 21 minutes after the first emergency call, they had rammed into the attacker's vehicle outside a second mosque, Linwood,  and detained him.

But it wasn't until 2.29pm, when New Zealand Police used a special escalation channel for law enforcement and intelligence agencies to alert Facebook to problematic content, that the social network acted.

The police alert triggered notifications to Facebook's most senior executives in Australia and New Zealand, including local policy chief Mia Garlick (who was at head office in California at the time) as well as content moderators in places such as the Philippines. Minutes later, the video was finally removed.

The roughly hour-long period that elapsed between the start of the live stream and the video being taken down would prove to be a crucial delay that enabled the video to spread uncontrollably.

During that period, the original video had been viewed by 4000 Facebook users. At least one of them copied and posted the video to 8chan, the online message board frequented by alt-right trolls, incels (so-called "involuntary celibates") and the alleged attacker himself. 8chan was also where the accused terrorist had posted a manifesto before the rampage began.

From there, the video spread rapidly across the internet, quickly surfacing on YouTube, Twitter and Reddit.  It was also picked up and aired by some news websites and TV stations - usually edited to remove the most violent moments. A groundswell of interest in the video developed online. Google searches for the video spiked.

Even though Facebook removed the original video and "hashed" it using technology designed  to prevent it from being uploaded again, it couldn't stem the tide.

Loading

In the 24 hours after the incident, Facebook was hit with a barrage of 1.5 million videos containing footage of the attacks. The social network's systems thwarted 1.2 million of them from being uploaded.  That still means 300,000 videos made it through its defences. Many of them were screen recordings, or videos with slight alterations, designed to evade Facebook's controls.

YouTube, meanwhile, was facing similar challenges. It described the volume of videos related to the incident as "unprecedented both in scale and speed". At certain points, there was an upload of the video every second.

The Google-owned site says it has removed tens of thousands of videos containing footage of the attacks since then. It has also terminated hundreds of accounts that tried to upload them.

In an update on Thursday, Facebook said it believed there was co-ordination by "bad actors" to distribute the videos through as many channels online as possible.

New Zealand Police would not say when asked by the Herald and The Age whether there was  a concerted effort by online communities to which the alleged perpetrator belonged to spread the videos.

The incident was also amplified by the traditional media's coverage of the event. The unavoidable fact remains though, that it all started on Facebook. Which means the biggest questions encircling the giant social media company, which has 2 billion users, are being asked again, at even higher levels than before.

An 'uncontrollable digital Frankenstein'

For example, is Facebook doing more harm to the world than good? And has its billionaire founder Mark Zuckerberg created a beast he cannot control?

"It's almost impossible for governments to put the tech genie back in the bottle without sending us back to the Dark Ages or making us similar to the totalitarian countries that do limit the internet," says Foad Fadaghi, an analyst at tech researcher Telsyte.

Tristan Harris, a former Google employee turned critic of the tech industry, this week described Facebook as an "uncontrollable digital Frankenstein".

The Christchurch attack  - and Facebook and Google's role in it - has occurred in the middle of an increasingly high-stakes, global conversation over how to regulate the two digital giants, which have grown to dominate the internet.

Australia's competition regulator is currently investigating how their stranglehold over online audiences and digital advertising is affecting traditional media companies, such as Nine,the owner of this website, and Rupert Murdoch's News Corp. The European Union, which has been hostile towards American tech companies for years, is also sharpening its knives.

Meanwhile, Elizabeth Warren, a left-wing United States senator and candidate for the Democratic presidential nomination, has called for Google and Facebook to be broken up.

The Christchurch attack also occurred on the eve of the first anniversary of Facebook's biggest crisis to date. It was in March last year when it emerged that the data of millions of Facebook users had been harvested without consent by Cambridge Analytica, a consulting firm that had worked for the Trump campaign during the 2016 US presidential elections.

Loading

Zuckerberg was hauled before the US Congress to explain the scandal, as well as the platorm's failures to prevent fake news from spreading, and from being infiltrated by troll groups linked to the Russian government.

During his testimony, Zuckerberg expressed confidence that artificial intelligence would eventually solve Facebook's biggest challenges, including fake news and terrorist propaganda. "Building AI tools is going to be the scalable way to identify and root out most of this harmful content," he said.

Christchurch was a striking illustration of just how far it has to go on that front. Yet experts in artificial intelligence were far from surprised.

"Mark Zuckerberg fronted up to Congress last year and said 'we are developing all these AI technologies', well, that is not going to solve the problem," says Toby Walsh, a professor  in artificial intelligence at the University of New South Wales. "We know that artificial intelligence is not capable of the subtlety of reasoning or the depth of understanding required to filter content in a fail-safe way."

Facebook founder and CEO Mark Zuckerberg spent about 10 hours answering to US Senate and House politicians in April last year. Credit:Bloomberg

But Walsh, one of the world's foremost experts on AI, says Facebook should still have done better in containing the Christchurch videos. Facebook has a strong reputation in the emerging field (some of Walsh's peers are employed by the company) and the social network's systems should have been able to detect certain videos with minor alterations designed to evade its controls.

"If someone has deleted the first five minutes of the video, or changed the resolution to make it look like it is different, or downgraded the quality, or changed the file type, something like that should be easy [for the AI] to spot," he says. "I can't believe that it is rocket science, especially if I had the resources of Facebook behind me."

That said, simple screen recordings are more difficult for AI systems to detect. And it is going to be a long time before that will even be possible.  "There are still very trivial ways that you can spoof computer vision systems," says Walsh. "To put your complete confidence in AI algorithms to filter out vision or audio is placing too much trust in what we can do today, and probably for the next decade."

That raises all sorts of questions about live streaming to mass audiences.

In a Thursday update, Facebook said that some people who shared the videos were supporters of the killer's actions, while others shared it out of (morbid) curiosity. Some intended to actually denounce the act. And it was amplified by reporting in traditional media, Facebook said. Broad reporting on the mere existence of a video "may have prompted people to seek it out and to then share it further with their friends," it said.

Candles are lit by school children for the victims of Friday's attacks in Christchurch. Credit:Jason South

There has been a debate within traditional media this week about coverage leading up to the incident, and of the incident itself. There have been arguments that inflammatory right-wing commentators sowed the seeds for anti-Muslim sentiment, but also introspection over whether reporting on the incident fed into the terrorists goals.

Peter Fray, director of the Centre for Media Transition at the University of Technology, Sydney, and a former editor of the Herald, says the Christchurch incident put traditional media outlets in a difficult position. "There have been images of equal barbarity and intensity that have changed the course of history,"  he says, pointing to images of a general assassinating a member of the Viet Cong, or of naked children running a way from a cloud of napalm, also during the Vietnam War.

"What the events in Christchurch bring home is it's no longer journalists framing these events. Here we have a self-confessed white supremacist, who straps on a GoPro, and he is the first recorder of history, not the journalist. And all of a sudden [millions of] people have got this video on their networks".

Would this have unfolded differently in the pre-Facebook, pre Google era? It is impossible to say for certain. But at the very least, the information would have been vetted by human editors.

Then again, TV stations in Australia were rebuked by the media regulator for airing sections of the terrorist video. And Murdoch-controlled Sky News Australia stopped its live simulcast into New Zealand's Sky News (which is not aligned with News Corp) citing a need to comply with strict legal requirements in the country.

"The real question is, to what extent is the journalism glorifying the madman, amplifying the madman, encouraging potential copycats?" Fray says. "I don't think the media has done that in this case, but I think it has got very close to it."

'If you go on live TV, there are controls'

Sheryl Sandberg, Facebook's second in command, was the first person to reach out to the New Zealand government. Within minutes of the video being removed, she contacted Prime Minister Jacinda Ardern, who she knows personally, to express her concern and condolences.

On Saturday morning at 9.15am, Garlick emailed senior advisers in Australian Prime Minister Scott Morrison's office to detail how the company had responded to the incident. "In general, since the attack happened, teams from across Facebook have been working around the clock to respond to reports and block content," Garlick wrote.

Facebook's attempt to soothe tensions at the political level didn't work, though.  Both Ardern and Morrison condemned Facebook last weekend for failing to take down the live stream and, by extension, for allowing it to spread.

"In the past, they have suspended this sort of Facebook live-streaming and assurances were given that when it was put back up, it could avoid this," Morrison said last Sunday morning. "Clearly it hasn't."

On Wednesday, Morrison confirmed he would push for co-ordinated action at the next G-20 summit to force tech giants to implement controls that prevent episodes like the Christchurch live stream from happening again.

"I want the social media companies to use their technology to ensure that instantaneously, their platforms cannot be used as weapons by terrorists. If they can geotarget an ad to you based on something you've looked at on Facebook within half a second - it is almost like they are reading your mind - then I'm sure they have the technological capability to write algorithms that screen out this type of violent and hideous material at a moment's notice," he said at a press conference.

The problem though, as Walsh points out, is such technology does not exist, and likely will not for years. That means there is no simple solution to the problem Google and Facebook have helped create.

YouTube's AI has encountered similar challenges with its videos, and in the comment fields of its videos, where paedophiles and extremists used coded language to communicate and avoid detection. The eventual solution to that problem was to eliminate comment fields on some videos altogether.

Loading

So, should Facebook suspend live streaming?

"I think there is a responsible way you can do it [live streaming] says Walsh."But if you want something to be broadcast to the world, you have to put traditional media safeguards in place."

Walsh, like many others, says tech platforms should face the similar types regulation and controls as traditional media companies.  "If you go on live TV, there are controls. When they fail, we get really upset at traditional media companies. So we should get really upset at tech companies as well."

"This isn't the first time this has happened and sadly wont be the last,"  he says. "At the moment, these safeguards don't exist at all online. These are the challenging questions we need to be pressing the tech companies on."

But Telsyte's Fadaghi says regulation may backfire if it makes it harder for new entrants to emerge and compete with Facebook, or forces users into even less regulated corners of the internet.

The regulatory disparity between old and new media was illuminated when TV networks where threatened with action for airing the videos, but Facebook and Google escaped any censure because no agency has jurisdiction over them.

One week on from one of the worst terrorist attacks in this region's history, videos of the incident are still being shared in the darkest corners of the web. And people are still trying to upload them to Facebook.

roewitioned.blogspot.com

Source: https://www.smh.com.au/business/companies/how-facebook-s-hour-of-inaction-enabled-the-christchurch-video-to-spread-20190319-p515k6.html

Share this post