This is the second of a three-part guest post by Paul Lewis that originally appeared in the book Investigative Journalism: Dead or Alive? You can read the first part here.

The investigation into Tomlinson’s death began in the hours after his death on 1 April 2009, and culminated, six days later, in the release of video footage showing how he had been struck with a baton and pushed to the ground by a Metropolitan police officer, Simon Harwood. The footage, shot by an American businessman, was accompanied by around twenty detailed witness accounts and photographs of the newspaper seller’s last moments alive and successfully disproved the police’s explanation of the death.

The result was a criminal investigation, a national review of policing, multiple parliamentary inquiries and, by May 2011, an inquest at which a jury concluded Tomlinson had been “unlawfully killed”. At the time of writing, Harwood, who was on the Met’s elite Territorial Support Group, was awaiting trial for manslaughter.

In media studies, the case was viewed as a landmark moment for so-called “citizen journalism”. Sociologists Greer and Laughlin argue the Tomlinson story revealed a changing narrative, in which the powerful – in this case, the police – lost their status of “primary definers” of a controversial event.

Significantly, it was the citizen journalist and news media perspective, rather than the police perspective, that was assimilated into and validated by the official investigations and reports. Ultimately, it was this perspective that determined “what the story was”, structured the reporting of “what had happened and why” and drove further journalistic investigation and criticism of the Metropolitan Police Services.

The initial account of Tomlinson’s death put out by police was that he died of a heart attack while walking home from work in the vicinity of the protests, and that protesters were partly to blame for impeding medics from delivering life-saving treatment. Neither of these claims were true, but they fed into coverage that was favourable to police.

A public relations drive by the Met and City of London police was bolstered by “off the record” briefings to reporters that suggested – also wrongly – that Tomlinson’s family were not surprised by his death and upset by internet speculation it could be suspicious. These briefings contributed to a broader media narrative that endorsed police and criticised protesters.

How the police account left so many questions unanswered

The morning after father of nine died, the newspaper he had been selling outside Monument tube station, the Evening Standard, carried the headline: “Police pelted with bricks as they help dying man.” But it was plain to us, even at an early stage, that there could be more to the story. The overlydefensive police public relations campaign gave the impression there was something to hide. Embedded in the small-print of press releases, there were clues – such as the Independent Police Complaints Commission’s notification of the death – that left unanswered questions.

Most obviously, anyone who had ventured near to the protests near the Bank of England on the evening Tomlinson died would have known he collapsed in the midst of violent clashes with police. It seemed implausible, even unlikely, that the death of a bystander would not have been connected in some way to the violence. But pursuing this hunch was not easy, given the paucity of reliable information being released by police, who at times actively discouraged us from investigating the case.

All that was known about Tomlinson in the 48 hours after his death was that he had been wearing a Millwall football t-shirt. That, though, was enough to begin pursuing two separate lines of inquiry. One involved old school “shoe leather”; trawling through notepads to identify anyone who may have been in the area, or know someone who was, who could identify Tomlinson from press photographs of him lying unconscious on the ground.

That yielded one useful eye-witness, with photographic evidence of Tomlinson alive, with images of him walking in apparent distress, and lying at the feet of riot police 100 yards from where he would eventually collapse. Why was Tomlinson on the ground twice, in the space of just a few minutes? And if those photographs of the father of nine stumbling near police officers, moments before his death, were put online, would anyone make the connection?

Becoming part of a virtual G20 crowd

The answer was yes, as a direct result of the second line of inquiry: by open sharing information online, both through internet stories and Twitter, we became part of a virtual G20 crowd that had coalesced online to question the circumstances of his death. In this environment, valuable contributions to the debate, which were more sceptical in tone than those adopted by other media organisations, worked like online magnets for those who doubted the official version of events. Twitter proved crucial to sharing information with the network of individuals who had begun investigating the death of their own accord.

I had signed-up to the social media website two days before the protest, and became fascinated with the pattern of movement of “newsworthy” tweets. For example, a YouTube video uploaded by two protesters who did not see the assault on Tomlinson, but did witness his collapse minutes later and strongly disputed police claims that officers treating him were attacked with bottles, was recommended to me within seconds of being uploaded. Minutes later, Twitter investigators had identified the protesters in the film and, shortly after that, found their contact details.

Similarly, those concerned to document Tomlinson’s last moments alive, including associates of the anarchist police-monitoring group Fitwatch, were using the internet to organise.

Through Twitter I discovered there were Flickr albums with hundreds of photographs of the vicinity of this death, and dissemination of blog-posts that speculated on how he may have died. None of these images of course could be taken at face value, but they often contained clues, and where necessary the crowd helped locate, and contact, the photographer.

Journalists often mistakenly assume they can harness the wisdom of an online crowd by commanding its direction of travel. On the contrary, in digital journalism, memes (namely, concepts that spread via the internet) take their own shape organically, and often react with hostility to anyone who overtly seeks to control their direction. This is particularly the case with the protest community, which often mistrusts the so-called mainstream media. Hence it was incumbent on me, the journalist, to join the wider crowd on an equal playing-field, and share as much information as I was using as the investigation progressed.

Establishing authenticity and context

There were times, of course, when we had to hold back important material; we resisted publishing images of Tomlinson at the feet of riot police for four days, in order to establish properly their authenticity and context.

Internet contact usually does not suffice for verification, and so I regularly met with sources. I asked the most important witnesses to meet me at the scene of Tomlinson’s death, near the Bank of England, to walk and talk me through what they had seen. We only published images and video that we had retrieved directly from the source and later verified.

A different standard applies to sharing images already released on Twitter, where journalists such as National Public Radio’s Andy Carvin in the US have proven the benefits from sharing information already in the public domain to establish its significance and provenance. The break, though, as with most scoops, was partly the result of good luck, but not unrelated to the fact that our journalism had acquired credibility in the online crowd.

Chris La Jaunie, an investment fund manager, who had recorded the crucial footage of Harwood pushing Tomlinson on a digital camera, had become part of that crowd too, having spent days monitoring coverage on the internet from his office in New York. He knew the footage he had was potentially explosive. The options available to Mr La Jaunie were limited. Fearing a police cover-up, he did not trust handing over the footage. An alternative would have been to release the video onto YouTube, where would it lack context, might go unnoticed for days and even then could not have been reliably verified.

He said he chose to contact me after coming to the conclusion that ours was the news organisation which had most effectively interrogated the police version of events. It was more than a year later that my colleague Matthew Taylor and I began inquiring into the death of Mubenga. By then we had recognised the potential reach of Twitter for investigative journalism and our decision to openly investigate the death of the Angolan failed asylum seeker was a deliberate one.

Not all investigations are suited to transparent digging, and, indeed, many stories still demand top secrecy. This has been true for the three outstanding UK investigations of our times: the Telegraph’s MPs’ expenses scandal and, at the Guardian, the investigations into files obtained by WikiLeaks and phone-hacking by the News of the World. However, Tomlinson had shown that open investigations can succeed, and there were parallels with the death of Mubenga.

In the third and final part, published tomorrow, Lewis explains how he used Twitter to pursue that investigation into the death of Jimmy Mubenga, and the crucial role of verification.