The first data collection period of the Share Wars project ran from March-June 2012. In that period we tracked every article published on the homepages of 118 global news sites. We tracked the articles’ share counts at regular intervals over a 24-hour period – on Facebook, Twitter and a few other social networks.
In total we captured 1.4 million articles and their individual “sharing curves”. By design and necessity, the project favoured fast-response news-based material over slow-burn content that might take days or weeks to build up its share count.
Only now have we arrived at a categorisation scheme for articles that we think is solid, consistent, and useful. The idea of the Share Wars project is to find sharing drivers within the content of articles – based on the assumption that content is the biggest driver. Other drivers are product (placement of buttons etc) and the broader uptake and evolution of social networks within the audience. We are ignoring those – we’re content people. From years of observations in digital newsrooms we know that, whatever dreams you may have for your new app or your new site, it’s the content that accounts for 90 percent of consumption.
We started the categorisation process from the top – the top shared stories across the British, Australian and US publishers. The process is manual. We examine an article and place it in a series of categories. The first category looks at the overall purpose of the sharing act, the second the state inside the sharer, the third is used for drawing out technical features of the story and everything from category 4 on relates to story subject.
There’s a number of reasons it has taken this long to arrive at a good scheme. One is that these stories are infernally, magnetically powerful and as you examine them you find yourself sucked into their narratives. It’s natural that this would be so – we distilled the most shared stories from three months of the world’s news output. These stories are designed to suck you in. I spent days just reading them, and I enjoyed it.
Another reason is that as we developed models and threw them out, we came to realise that a lot of what was happening with these “shared” stories was not really about sharing. Often it seemed to us to be more about reinforcing a view of the world and strengthening group identity. We called this “norming”. Sometimes it is simply the digital equivalent of yelling out a warning or exclamation. We called this one “newsbreaking”. Yes, they are all types of sharing behaviour but they are not what we thought of as classic sharing: reading a story, valuing it and forwarding it based on your evaluation of the quality of the content. Here’s a breakdown of the first two categories:
We have now categorised the top few hundred stories in the UK, US and Australia – not even a blip on the radar of the whole data set – and we have seen an interesting pattern. Hitting the “Like” button on an article is more often about norming than it is about pure sharing. Furthermore, there are significant differences between the Anglophone cultures in which types of stories are shared the most. Perhaps not surprisingly, the US top shared stories are far more issues-based (gay rights, race hate) than those of Australia. More on these differences and what they mean in coming days.