This is the second chapter of the book All Your Friends Like This, which I wrote with colleagues Andy Hunter and Dom Filipovic and was published in 2015 by Harper Collins. The book summarises the Share Wars project, which collected Facebook sharing data for stories from the world’s biggest news publishers from 2012-2015. Download this PDF version for an easier offline read.
Chapter 2 The Birth of Share Wars, by Hal Crawford
Picture yourself on the verge of a job interview.
It’s well known that assessment in a spoken interview has no correlation to subsequent success in the job. A general discussion designed to determine the character and ‘cultural fit’ of a candidate is a festival of delusion and cognitive bias. Daniel Kahneman, the Israeli-American Nobel Prize- winning psychologist whose work comes up several times in this book, established the principle while working for the Israeli Defence Forces, observing that expert interviews do little better than chance in predicting the future success of candidates. Interviewers are swayed by what people look like; how recently they spoke to them; their mood; and above all by the misplaced belief in their own prowess. It is unusual to come across anyone who rates themselves a bad judge of character. Interviews that rely on the intuition and expertise of the interviewer are almost useless.
The problem is that the charade suits both sides: neither the interviewers nor the interviewee care to admit they may as well be rolling dice. So job interviews persist, and everyone reading this book will have experienced the feeling of waiting outside the closed door while the panel sits within.
This is not about how well you can do the job. This is about how well you can be judged. nIn July 2006 I was preparing for that judgement.
I’d left The West Australian newspaper eight years before and gone wandering the world, returning to Australia to work at La Trobe University in Melbourne. I was teaching a journalism course – an education I had never believed in. My problem, as I saw these hundreds of sleepy faces assembled for the morning lecture, was that journalism didn’t seem to me to be a fit object of study. The practice itself is a combination of mundane technicalities (details better learned on the job), general knowledge and curiosity. How do you teach it, really? Maybe my scepticism seeped out. The university decided not to renew my contract for the next semester, which left me with two weeks to find a new source of income. My girlfriend and I had enough money left for Plan B – two tickets back to Perth. I was becoming desperate. I had applied for jobs I thought I could do pretty well. Little reporting gigs, like writing shipping news for Lloyd’s List. Stuff in public relations (how hard could it be?), academia, and even branching out into the public service. I applied to work at a local council. Their rejection confirmed my suspicions: I was not qualified to write parking tickets.
With a few days to go before our self-imposed deadline to leave Melbourne I landed two interviews: one for a job on legendary Australian magazine The Bulletin’s website, and one for internet portal ninemsn. Both jobs were based in Sydney, so both interviews would be by phone.
The day for the Bulletin interview came. I sat in the front room of the old worker’s cottage we had in the suburb of Prahran and tried to be the person they wanted to hire. On the line were editor-in-chief Garry Linnell, editor Kathy Bail, columnist Tim Blair and internet editor Lynda Dugdale. They asked me to rank a bunch of stories in order of importance. My diary notes, ‘What a crew … I may have come across as too touchy feely’. I also wrote, ‘I have never wanted a job so much.’
It didn’t work. Dugdale let me down gently – so gently, in fact, that initially I couldn’t understand what she was saying. The Bulletin was a publication with a pedigree, a proud history and the big names. As it sank in that I hadn’t got the job, I felt like I had been kicked in the stomach.
The interview for ninemsn was different. A bloke called Andrew Hunter and a guy from HR were on the line, and the conference phone kept cutting out. I have always liked the way technical failure screws with expectation. There’s freedom in a debacle. I was in the front room again, but I’d read about a trick to make yourself sound more confident: I was standing up. I paced back and forth across the floorboards, expounding my digital news vision. I’d done my homework and I was on the front foot.
In 2006, ninemsn was going strong. The portal had been launched in 1997 as a 50/50 joint venture between Microsoft and PBL Media, the holding company for the Nine television network and ACP Magazines. At the launch party a young, slim-faced James Packer had stood in front of a wall of heavy monitors and declared, ‘This is the train leaving the station.’ nFrom launch, ninemsn played a supporting role in most of the big success stories in Australian digital history. SEEK, eBay, realestate.com.au, carsales.com.au: few big sites rose without help from a portal that funnelled all the exit traffic from Microsoft’s Hotmail, MSN Messenger and Internet Explorer to one place. Ninemsn was home to all the digital properties of the Nine Network and had the exclusive right to commercialise all the content that came out of ACP Magazines, including The Bulletin. For millions of Australians coming online for the first time, ninemsn was the start of the internet. For those in the industry, it was a traffic machine that distributed its bounty with a generous hand.
I got the ninemsn job. It was Andy Hunter who lowered the rope down from the digital chopper and broke me out of my Melbourne predicament. Andy was a big part of why ninemsn had been a success. A powerful mix of consideration and enthusiasm, Andy came first from music – he’d been the bassist in a band called the Daisygrinders – then newspapers, the ‘street press’ and magazines. He’d done time at ninemsn’s main competitor, news.com.au, and had a better grip on digital news than anyone I’d met.
For all the research I had done into the role, and my previous work as a digital editor in the Netherlands, I was pretty clueless. I sat in the ninemsn newsroom in central Sydney, staring at the internet portal’s custom-made publishing software and wondering what to do. As Kahneman had discovered 50 years before, vetting officers for the Israeli Defence Forces, my performance in the interview had been no predictor of fitness for the job. Every screen that flashed in front of me looked the same. Here journalists were not reporters or subeditors but ‘producers’. What they produced was not stories but ‘content’. There was this weird process where you sat down with your manager and talked about yourself. These conversations would begin with ‘How are you going?’, accompanied by a meaningful look that indicated your psychic well-being was of the utmost importance. A refugee from newspaper culture, I was longing for the reassuring growl of the misanthropic editor and distrustful of an organisation that seemed not so much to observe the world as to collate and package it. There was no perpetually boiling urn, no International Roast and very little foul language. nMy breakthrough came with breaking news. One of the commonalities between legacy and digital news media is that other people’s disasters are the lifeblood of both. This is not some sick fixation but a reflection of a general truth about people: we need all the information, immediately, particularly if it’s about something bad.
Fame, death and a new kind of news nOn 4 September 2006 an ambulance press release dropped about a man having been killed by a stingray near Port Douglas in Queensland. This was extraordinary. A few seconds later we learned the unfortunate bloke had been 44 years old. I started writing the first paragraph of the story before the real shock hit. The dead man was Steve Irwin. The Crocodile Hunter was gone.
All my doubts about what to do were dispelled. The death and identity of the victim were confirmed with a phone call while I was still working on that first paragraph. The story fragment was published within seconds, along with a breaking news strap on the ninemsn home page. Traffic to the servers spiked as Australia came online. I proceeded to fill out the story. Unlike in newspapers, my corrections and extensions could be published as fast as I wrote them. Information came in: the ray’s barb had gone through Irwin’s heart. He’d died almost immediately. His crew had been filming at the time.
Years later, in 2014, the cameraman who was with Irwin finally revealed what happened that day on the Barrier Reef. This was how Justin Lyons described the stingray attack in an interview with Channel 10:
I remember it very clearly … we’d been filming with crocodiles and sea snakes — milking sea snakes — and we were looking for tiger sharks on this particular day. We’d had a bit of bad weather, and Steve was like a caged tiger, particularly on a boat, so he said, ‘Let’s go do something.’
We’d been motoring for a few minutes when we found a massive stingray. We’d swum with stingrays many times before. This one was extraordinarily large — it was eight foot wide — it was very impressive. We were only in chest-deep water. We stood up and chatted about what we were going to do, we made a plan, and slipped into the water.
Stingrays are normally very calm. If they don’t want you to be around them they will swim away, they are very fast swimmers … I had the camera running and I thought, this is going to be great shot, it’s going to be in the doco for sure, and all of a sudden it propped up on its front. It started stabbing wildly with its tail — hundreds of strikes in a few seconds. nI panned with the camera as the stingray swam away. I didn’t even know it had caused any damage. It wasn’t until I panned the camera back [and saw] that Steve was standing in a huge pool of blood that I realised that something had gone wrong.
The stingray barb is about a foot long extending out of the middle of the tail. It’s a bit like a fingernail, the other half is embedded in the tail of the stingray … It’s a jagged sharp barb and it went through his chest like a hot knife through butter. nHe thought it had punctured his lung, and he stood up out of the water and screamed, ‘It’s punctured me lung.’
Within a few seconds the inflatable that had been motoring about 30 metres away was there … He had about a two-inch wide injury over his heart with blood and fluid coming out of it and we thought, we’ve got to get him back to the boat as fast as we can. As we’re motoring back to the boat I’m screaming at one of the other crew to put their hand over the wound. And we’re saying, ‘Think of your kids, Steve, hang on, hang on.’
He just sort of calmly looked up at me and said, ‘I’m dying,’ and that was the last thing he said.
Less than an hour after Irwin spoke those words the story broke. The load on the servers – the computers that presented the pages to the public – became unlike anything ninemsn had experienced. They began malfunctioning. Occasionally sites suffer malicious concerted traffic requests, where hackers orchestrate what are called denial-of-service attacks. The news of Steve Irwin’s death created an unintentional but massive denial-of-service attack as millions of Australians crammed online. As people saw error pages instead of the story and the technicians began to panic, I experienced a weird sense of peace. The freedom of the debacle. I had landed in the right spot.
From that day, Andy and I and others at ninemsn worked to build a newsroom culture that fused the best of the old with the possibilities of the new. The infrastructure team increased the server capacity, and we made the pages look better, with bigger pictures, headlines and video. We tried hard to be fast and accurate. By far the biggest change for me was the discovery of the audience and the revelation that we could actually see what they wanted to read using software. This discovery fitted hand-in-glove with the commercial needs of our organisation: if we maximised both the number of people looking at our site and how many pages they viewed every time they came, we could make more money. In that era, advertisers were still willing to buy as many banner ads as we could create.
Digital display advertising works by selling standard ad spaces on the page by the thousand. The most common ad is the ‘medium rectangle’, a piece of desktop real estate 300 pixels wide by 250 pixels deep. Within those 75,000 pixels the advertiser may present any message to the audience there to consume the site’s content. A typical CPM (cost per mille, or thousand) of these medium rectangles on an established site would be $10, meaning that for every ad displayed the advertiser pays the publisher just one cent. A single ad displayed once is called an ‘impression’.
When you understand the dynamics of the relationship between the publisher and the advertiser, you begin to see why certain things online are the way they are. For example, a publisher may only get one cent for every ad impression, but there is nothing stopping him from loading pages up with multiple ads, thereby increasing the number of impressions every page view delivers. This leads to ad proliferation and cluttered pages. The relationship also encourages the creation of lots of page views. Both of these factors – ad and page- view proliferation – are only relevant so long as the advertiser continues to want the impressions. nThat was the game we had been in, the driving commercial force behind the ‘pure traffic’ era of digital. The problem with this approach was that it was very short-term, flooding the market while compromising audience experience. The push for impressions also came at an editorial cost: the art of the page view slipped easily into an art of darkness.
Running any kind of publication, you are met with the requirement to keep the feeling right, to stay true to your voice while engaging an audience. What Andy and I had found in our modestly sized newsroom was that our devotion to traffic inevitably led young editors to go down dark paths in their daily pursuit of the page view. Digital news outfits publish constantly, with no space constraints, and because of this, maintaining control of voice is more difficult than in a newspaper with a daily deadline. The short-term feedback of the analytical tools shows the site editors what is working, so they obediently ‘starve the losers’ and then end up with a ghastly news mix. A typical page under this kind of unimaginative regime might feature a foreign murder picture story, with a news lead of a local rape, followed by a sex abuse scandal and a bestiality yarn thrown in for laughs. Sometimes when you challenge a producer on this kind of mix, you see them look at the page as if waking from a dream. They were only doing it for the traffic.
We instigated regulations that limited the number of simultaneous rapes, murders and other gory stories that were permissible on the home page. This eliminated the ‘pall of gloom’ effect from a story mix that generated traffic but risked leaving the audience depressed and anxious and undermined the long-term future of our site. Bestiality was banned unless there was an overwhelming ‘public interest’ reason to publish. The in-house style guide specified exactly how much butt- cheek was too much butt-cheek in a thumbnail image.
But we couldn’t regulate the problem out of existence. Rules tell you what you must not do – they won’t tell you what to create, the vast number of subjects the audience might like but does not yet know about. We needed something – some guidelines, a framework of understanding – that brought science to the process of selecting which of the world’s myriad stories we should pay attention to. nFive years after Steve Irwin’s death, news media had transformed again. The power of print had evaporated as audiences went online and changed their daily habits. Many of the old print mastheads still had big audiences, but they were now digital and pulled in less money. Back in 2006, newspapers and magazines had been packed with talent, highly resourced, focused on publishing original content. It was a high-cost set-up. We knew that the kinds of revenues we were making at ninemsn, the biggest digital publisher, would have struggled to fund a newspaper newsroom. What had seemed like bare operating costs for print now looked extravagant.
The Bulletin shut down 18 months after my failed job interview. Everybody was laid off. I went to the magazine’s farewell party and realised how much my view of the industry had changed even in that short time. All the people who had interviewed me had been experienced journalists, heavy hitters. Gary Linnell would go on to head up Fairfax. Kathy Bail ended up a publishing CEO. Estimating their salaries based on what I know now, I reckon there was a million bucks of total cost on the other end of that phone line, interviewing a candidate for a minor position. Based on the concentration of cost, maybe I should have been able to predict the fall of the magazine I was then so desperate to join. But in that era, most journalists like me didn’t think about the business side of their business. I knew of expensive journalists working for prestigious newspapers who were comfortable filing one story a month. They were doomed.
In digital things were changing too. The pure traffic era was over. Commercially, the market was flooded with ad impressions – prices for excess ‘remnant’ impressions were dipping below a tenth of a cent – and the audience had tired of Britney Spears and seemed to be moving beyond pure celebrity culture.
Andy and I were sitting around the stone tables in the shade of the Australia Square tower building contemplating what was next. Working with Andy, I’d gone from distrusting anything that didn’t look like old newspaper culture – the ‘coal face’ mentality – to understanding something about technology companies.
The challenging thing about Andy is that he doesn’t mind contradiction. Sitting at Australia Square that day, we both agreed it would be good to get some ideas together about the new digital world and publish something. My suggestion was to document what we knew about the art of the page view.
‘No, that’s the old world,’ he said. ‘The question is, what’s next?’
We sat and watched an ibis circling the tables. When I’d arrived in Sydney, I’d been interested to see these large birds standing on rubbish bins and fossicking in parks. I was amazed when Andy told me they had not always been part of the Sydney city wildlife: ‘They weren’t around when I was growing up. One day they just appeared.’ nI’d imagined the city eternally as I found it. I’d been wrong. Nothing stays the same. nBy the time the pungent animal had finished his round, we had our hypothesis about the future of the industry.
Social networks – the platforms that were already sending us thousands of people and growing every day – were going to make news media better.
That day we founded Share Wars – our project to investigate what made news stories share on social networks. The name came from our belief that the social news feeds of individuals would become the next big battlefield for news media, and that the battle would change the nature of the combatants.
The significant part of [sharing news on social networks] is that it promotes a different kind of feedback loop. It’s a feedback loop that stretches out from the individual independent of the publisher, and it’s moderated by value. Public value.
It’s a channel unlike any the news world has seen before. It will shape not only the way we consume news but also the very stuff of news – the stories themselves.nFrom Post 1, Share Wars blog, 4 July 2011
The new feedback loop between publisher and audience would see news remade in the image of the stories that individuals chose to share. On the whole, that makeover would be an improvement, because people only share what they value.
There were reasons why investigating news on social networks made a lot of commercial sense as well. With the market flooded with dirt-cheap ad impressions, publishers had to try to find a way to prove to advertisers that their content actually meant something to an audience. Getting someone to share a news article, to put their own name and identity to it, represents a deeper commitment than merely getting them to look at it. We reasoned that advertisers would want to be associated with content that was so highly valued by an audience. Sharing indicated true engagement with the content, a trait that could also come in handy in situations where advertisers turned to the publisher for advice on what content would make potential customers pay attention to their message. In Andy’s experience, this occurred frequently with big spenders like cosmetics, car and drink manufacturers, who wanted to create bespoke ‘hubs’ of content they felt suited their products.
Andy and I took our proposal to ninemsn: we wanted to start a private project investigating the drivers of news- sharing on social networks. We decided to pursue a policy of openness with our findings, in the belief that this would create opportunities, but we also wanted to retain ownership. To give over the intellectual property in Share Wars to ninemsn would be signing its death warrant – too many committees, not enough true ownership – and also foregoing any potential commercial value down the track. After some negotiation, our bosses agreed. Newly appointed ninemsn CEO Mark Britt understood that having employees engaged in world- first research could not be a bad thing for his organisation and was unequivocally supportive. Share Wars would remain independent of Australia’s biggest internet portal, but we would use the knowledge we gained from the project to help our day jobs. If we discovered the key to news ‘virality’, we would use it to ninemsn’s advantage. nWe had our mission. Now we had to pull off the hard bit – gathering the data, studying it, and forming a model for the perfectly shareable story. We never thought that would be straightforward, but neither were we quite prepared for the trip our quest would take us on. We set up a blog and began publishing our ideas on social networks and news. nWe knew our message wouldn’t be welcomed by many in the media. Not only were we revealing ourselves to be ‘demise deniers’ and sceptics about the supposed high quality of old media, we were proposing that journalism could be improved by the digital scourge. That message approached sacrilege to journalists in organisations undergoing regular rounds of cost-cutting, who could only see editorial values being undermined.
If you are going to take on the sacred, you’d better have some factual backup. Our first task was to build an engine to harvest the world’s news. We would need another Share Warrior: a software engineer.