Facebook overlord Mark Zuckerberg with a padlock, which for the purposes of this article is both metaphorical and literal (Getty Images)

The data does lie: how Facebook’s fake video stats smashed NZ journalism

A lawsuit has revealed Facebook inflated its video statistics for years, inspiring the ‘pivot to video’ which made thousands of journalists redundant. Duncan Greive looks at its impact on New Zealand.

So Facebook’s been lying again, at least according to a suit filed last week. Days after admitting that its new portal device would eavesdrop on its owners for advertising purposes, and weeks after revealing a giant hack. The platform’s blunders are hardly scandalous any longer, more a known attribute that must be weighed when assessing anything it claims.

The company which preaches connection and the free flow of information between people all over the world is much less passionate about those same people learning anything about its inner workings – they tend to leak out years later under duress.

The latest situation is this: for years it overstated its video statistics, discounting views of less than three seconds from its average duration figures, which had the effect of inflating the amount of time they plausibly claimed people watched its videos. With video autoplaying within Facebook’s feed format, billions of people would let a video start, but a much smaller subset would watch it right through. By excluding the former group, the company was able to imply a far more substantial engagement with video than actually occurred.

This might seem a small thing – some data misfiled, a minor categorising error.

Only, it’s a lot more than that. The story has come to light because of a lawsuit in the US, brought by advertisers who say their clients were induced to spend money on Facebook by these misleading statistics (Facebook deny the suit’s charges). The plaintiffs have 80,000 pages of internal Facebook records which show that not only was the data wrong, but that Facebook knew it was wrong and failed to act on that knowledge for over a year. Not only that, but that when they did admit the problem they claimed the issue inflated figures by orders of magnitude less than it really did.

The suit does not allege that the higher numbers themselves cost businesses money, but that they acted like neon signs in Vegas to clients, enticing them to pump cash into Facebook’s boosting machine, paying to push these videos into more feeds. Businesses were operating under the assumption that their videos were going out into the world and being watched for much of their duration, when most thumbs simply hovered for a moment. (I use thumb here for a reason – Facebook is essentially a mobile advertising company – 89% of its income comes from that source.) Once a clip flicks over three seconds, it counts as a view – and one way or another, someone’s probably paying.

This error was extraordinarily lucrative: in the four completed years since it introduced video, its profits have grown by over 1000%. The flood of money into Facebook was torrential enough already without the huge head start this mangled data gave them. But, as with most prickly situations they encounter, Facebook made a few tweaks and kept collecting the money.

The suit is a good one, and it’s nice to see a front in the war to tame big tech open up in the US, rather than having it all sit on the immensely strong shoulders of the EU’s superhero prosecutor Margrethe Vestager, who has levelled multibillion-dollar fines at the giants.

People working in marketing all over the world will be given pause by this moment, though it’s regulators who should be most troubled. The platform is brilliant in many ways, woven into our lives, and network effects mean that people are likely to stay on Facebook regardless of how it behaves. So it’s only governments which can effectively supervise it.

There are many groups who’ve been impacted by Facebook’s poor behaviour, but I want to focus on the way the latest one affected New Zealand – particularly on its impact on journalism here. The decline in revenues for New Zealand’s print and television journalists was already well in motion by 2013, when Facebook video launched, with rounds of redundancies regularly announced at the Herald and Stuff (then Fairfax NZ).

Against this backdrop the extraordinary performance of Facebook news clips gave some in journalism that rarest of sensations: hope. Mark Zuckerberg confidently said at the time that in a few years “the vast majority of the content that people consume online will be video”. In the next couple of years international news organisations began posting video direct to Facebook and seeing millions of views, and re-angling their newsrooms accordingly. All this activity became known as the ‘pivot to video’, one of the biggest media trends of the past decade. Could it happen in New Zealand?

Our news organisations all wanted to find out. The Herald launched Focus, hiring Campbell Live veteran Tristram Clayton and moving Laura McGoldrick from Hauraki to front a daily current affairs wrap, with the now-exiled Tony Veitch on sports. Mike started his Minutes, and WatchMe appeared as a home for entertainment programming. (As with the whole pivot, not all of it happened on Facebook – but video’s performance there and Facebook’s evangelism for it was the largest driver of the change).

Not to be outdone, Stuff hired Paula Penfold and her  investigative broadcast team to launch Stuff Circuit, while asking many of their journalists to shoot video or solicit it from the public. RNZ started streaming Checkpoint live, and creating video for key interviews elsewhere. The NBR made arguably the biggest investment of all proportionally, recruiting top shelf talent like Simon Dallow and Susan Wood to front video programming.

At the same time, Facebook was touring the major newsrooms, teaching staff how to use its then prioritised Facebook Live video feature (this was before it became most famous for a live execution) and building small studios in some newsrooms to encourage publishers to migrate to their platform.

Around the same time they were launching Instant Articles, which promised faster load times and higher engagement rates, and into which Facebook would sell advertising on behalf of publishers (a service the Guardian wholesale adopted and completely spiked within two years). It feels so long ago – before Brexit, before Trump, before Cambridge Analytica. Before Facebook’s  yawning response to news of its complicity in the Rohingya genocide. Before it announced that it actually wasn’t about news at all, but friends and family, and the whole media (including New Zealand’s) saw their traffic nosedive. When we still believed that boyish figure in the grey t-shirt wanted the best for us.

So when we saw the numbers, we acted. We made redundant boring old print journalists, built multimedia studios and asked young reporters learning an already complex trade, made more so by the maelstrom of its social media, to learn yet more skills.

It seems like a collective madness, looking back. We already knew more than enough about Facebook – already knew that we were the product, that when we liked pages or posts we were providing data so that we could more efficiently be packaged up into different interest groups and sold. That when we did this diligently we were making the machine even more efficient at diverting money which previously sustained the mass media into Facebook.

We knew that Facebook would always act in the best interests of Facebook. That it had encouraged millions of brands and thousands of media organisations to spend heavily to grow their Facebook audiences based on the idea that they could speak to them through the feed, then changed the rules to show only a tiny fragment of the audience that content unless a toll was paid.

We knew enough already to know not to trust them. Yet we were desperate, starving, and easily seduced. We knew a lot of people were on Facebook, we knew a lot of people watched video on their phones, so I guess it was possible? That’s what we told ourselves when videos spiked from the tens of thousands to the hundreds of thousands. We wanted to believe that we weren’t buying a monorail.

We as publishers felt we had to pay attention. Had to roll Facebook into our distribution plans. And NZ on Air, itself trying to make sense of the new environment just as publishers were, would have been negligent not to follow the data too.

Some time in the past year, despite allegedly having known since 2015 it was inflating numbers by as much as 900% (if you’re going to lie, lie HUGE, right?) Facebook finally started to bring its numbers back to earth. They didn’t make any major announcement, and they didn’t do it overnight. In a classic example of one-party state behaviour, they quietly changed the names of metrics, replacing ‘average duration viewed’ with ‘average watch time’.

A fake stat replaced with a real one. No harm done. Except that NZ on Air and our media companies had committed millions of dollars and dozens of jobs on the basis of a platform which had failed to provide accurate data to its customers. And part of the reason they were acting with such urgency was because so many advertisers believed the numbers too – agencies started hawking social video, and even ropey brand pieces were clocking up far greater durations than they were likely really earning. This was money diverted from other sources – outdoor, radio, display, print, TV.

Facebook user Mark Zuckerberg delivers the opening keynote address at the f8 Developer Conference April 21, 2010 in San Francisco, California. Photo by Justin Sullivan/Getty Images

Now we know why. The New Zealand media were collateral damage in Facebook’s obsessive desire to grow at all costs. As Facebook’s Sandy Parakilas put it in Evan Osnos’ extraordinary recent New Yorker profile of Mark Zuckerberg, the sense he got on arrival was “we believe in the religion of growth”. At any and all costs. The media was just one of the things broken in the quest to move fast.

Getting accurate data which can cross sites, let alone media, is already difficult enough without episodes like this. I can already hear the chorus coming from legacy media: stop spending money online. Here they have established a lobby group under the name ‘Think TV’, and are telling anyone who’ll listen that the world isn’t changing – or not nearly so fast as we think.

Only, it is changing. NZ on Air’s own research is easily the most comprehensive survey of media consumption in New Zealand, and its release earlier this year showed that, if anything, it’s accelerating. The most shocking statistic: Netflix is already the second-largest channel in New Zealand by daily reach, trailing only TVNZ1.

There are plenty more though, most showing the divide between New Zealanders aged 15-39 and 45+. Just one example: 50% more of the young group watch YouTube or Facebook video each day than watch linear television. What we really have now are two New Zealands, living alongside one another.

One is older, more rural, whiter, effectively set in its ways. It is well served by existing media and infrastructure – it listens to the radio in the morning, watches the TV news at night, reads the paper on the weekend. The second is younger, more ethnically diverse, less likely to vote or own a home. It consumes the vast majority of its content online, and is currently very poorly served by our publicly funded media. Not because there’s no attempt, but because it’s difficult.

Facebook being both enormous and untrustworthy does not change the fact that this second group deserves to see itself reflected in ways more sophisticated than an unfiltered look at social media. It just makes an already complex job harder. The response of government and of funders cannot be to abandon any attempt to reach younger New Zealanders – it has to be to think about all the tools it has at its disposal to try and tame the digital giants which are increasingly embedded in their lives.

This saga is just the latest proof that Facebook is not a benign actor. For a long time its scale has been discussed by comparison to other nation states. So let’s assess its behaviour under those terms.

Facebook is entirely undemocratic, with no ability for its citizens to influence its governance. It has proven itself willing to undertake undisclosed experiments on its people, with express intent of altering their mood. It follows the techno-libertarian aversion to tax. When its citizens provide it crucial information to protect themselves, it sells that information without their permission. It allows itself to be a portal for nations to meddle in one another’s elections, and profits from that meddling. It stifles dissent, refusing to allow any post referencing Facebook to be boosted. When it has been co-opted to spread false stories regarding religious and ethnic minorities which have aided a genocide, its response is far too slow due to its obsession with finding engineering-based solutions over human moderation. And when its systems prove faulty internally it does not inform those impacted by its failures.

Of course, this is not all it is: like most tech companies, it’s a mix of the good and bad. There are many incredible things about Facebook, for both users and clients – this is my favourite place on the internet – but you already know what they are and they have a whole machine dedicated to telling that story. It’s about whether we can have the good with less of the bad. Specifically, here in New Zealand, where we live.

This is a corporation which has more customers than any other in our country: far bigger than Spark or Progressive or The Warehouse. Yet its actions – repeated breaches of trust – appear to have no consequences.

The one entity which can influence its behaviour is government. It doesn’t lack for tools – consumer and commerce protection law could easily be applied to what is functionally a monopoly. Yet there has never been any serious attempt to confront the monster that Facebook has become. Not here, not anywhere.

Instead successive New Zealand governments have actively supported the expansion and hegemony of Facebook. Almost every department and agency has a Facebook page – which means it has someone to run it, and likely an agency to advise it. It creates content to fill the feed, and inevitably spends money directly on boosting content to try and find its audience. In all probability, government entities will have strategised based on the algorithm as it was, not as it would be – and accelerated spending in response to the gaudy and deeply misleading numbers it was seeing from 2013-2015. 

I know journalists here who have been trying to find out just how much state entities spend directly and indirectly Facebook. It’s a difficult task, corralling hundreds of agencies. An educated guess suggests it’s well over eight figures, and probably over nine – an extraordinary sum to spend on a malign foreign entity which barely pays tax in New Zealand. Put it another way: there is almost no chance that the government spends anything like as much on public journalism as it does on Facebook.

This latest scandal barely rated last week. When we were on Facebook, the only thing we saw was Jami-Lee Ross’s rictus grin and the chaos it birthed. Still, it is up to the politicians and regulators now. Someone has to try and wrestle the big blue monolith. It has already hacked democracy and shown indifference to the consequences. Does it have to happen here before anyone in power starts to pay attention?


The Spinoff’s business section is enabled by our friends at Kiwibank. Kiwibank backs small to medium businesses, social enterprises and Kiwis who innovate to make good things happen.

Check out how Kiwibank can help your business take the next step.

Related:


The Spinoff is made possible by the generous support of the following organisations.
Please help us by supporting them.