periodismo deportivo de calidad: Burn-Murdoch: "There’s a golden opportunity for sports journalists to add more quantitative analysis to their repertoire to better serve their audience"

John Burn-Murdoch is one of the most outstanding sports data journalists in the world. He has been working for the data and visual journalism unit at The Financial Times for the last two and a half years and also is a visiting lecturer at City University London.

His particular road into data journalism started in 2011 when he was brought in by the Guardian's data team as one of a group of students to work on their analysis of the London Riots. After that he completed a Masters in Interactive Journalism at City University London and then he took on a longer term position with the Guardian’s Datablog, where spent 18 months working with Simon Rogers, James Ball and Mona Chalabi. During that time he also launched and edited the Guardian's section on 'big data', before departing for the FT.

- How many people work in FT Data (journalists, data scientists, ...)?

- As is often the case, it depends where you draw the line! There are three of us whose role could reasonably be described as ‘specialist generalist’: we have backgrounds in traditional written journalism but have since developed varying levels of skills in areas such as data wrangling, statistics, web design and development, and digital cartography. This means there are plenty of projects where we could be self-sufficient, but for most pieces we collaborate with dedicated developers, designers and statisticians from elsewhere in the FT's broader data/visual journalism family.

- More and more outlets are launching their own Labs or Data Departments in order to tell stories in a different way. Why is data-driven journalism becoming a priority for the media?

- The question for me is actually the inverse: why are some news outlets still reluctant to adopt this approach? Right across the spectrum from the likes of ourselves and the New York Times, to digital-only startups like Quartz, to tabloids like the UK's Daily Mirror and its regional subsidiaries, news organisations have realised and demonstrated the value of data- and visual-journalism.

If you broaden out the meaning of the term data journalism to include any use of structured information and related technologies to find and tell stories, you really see how it adds value. There are the pieces that simply wouldn’t be possible without it, the pieces that allow you to come at a story from a new and more creative angle, the pieces where it adds immense value to an ongoing story, and the pieces where it saves you — the news organisation — vast amounts of time and resources through the automation or replacement of a regular series of stories.

Those are just a tiny subset of the stories and use-cases for data and visual journalism, but it seems clear that these all help fulfil core requirements of a newsroom in 2016. For the subscription-funded media organisation, data/visual journalism can provide the exclusive stories, unique and rigorous analyses and stunning visuals that leave readers feeling their money is being well spent. For the smaller or more traffic-focused organisation, the time and resources that data journalism can save make it an obvious choice.

- Data-driven journalism is still developing in many countries though. What is needed to have this kind of journalism more implanted in the outlets? Should be there more training of (future) professionals from the University, is it a question of journalistic culture (changing concept according to the country) or just a bet that must be done by companies?

- A great question, but one to which I hope the answer is “it’s getting better!”. To set up a dedicated or semi-dedicated data/visual journalism unit requires investment, just like with any other specialist desk. As a result, news organisations are understandably keen to make sure their investment will pay off. This means there’s a Catch 22 situation to a certain extent: editors and managers are reluctant to hire data journalists without evidence that the result will be beneficial to the newsroom, and as a data journalist in a country where there are no data teams,it can be difficult to get the opportunity to prove that you can add that value.

Training can only be a good thing, but academic institutions have the same problem: why should they invest in setting up a specialist course or module(s) in data journalism if there are few-to-zero employment opportunities in the sector?

As an optimist, I look at how things have developed organically first in the US, then the UK, and now dozens more countries, and I see signs that this is a process that will naturally follow elsewhere.

Even as recently as five years ago, there would probably be one or two data journalism jobs created in the UK each year, but now there can be as many as ten openings a year that are either listed specifically as “data journalist”, or describe that role in as many words.

"One of the big growth areas at the moment is on the data creation side. I don’t mean making up numbers, but rather the use of specialist equipment and or expertise to build datasets that have never existed"

- How is evolving data-driven journalism? In what fields of information is being more and more relevant?

- Again this comes down to where you draw the line for what is and isn’t data journalism, but if I go back to my broad definition — the use of structured information to find and or tell stories — I think one of the big growth areas at the moment is on the data creation side. Before anyone recoils in horror, I don’t mean making up numbers, but rather the use of specialist equipment and or expertise to build datasets that have never existed.

A completely different growth area is in the visual side of data journalism, where it’s clear that more and more news organisations are realising the immense value that powerful visualisations can have in drawing and engaging audiences.

- What is the potential of Big Data in Sports Journalism? Here we need to handle such a huge number of stats?

- I’m reluctant to use terms like Big Data, which holds very little practical meaning, but on the broader point of the crossover between data journalism and sports journalism, I think the two lend themselves to one another amazingly well.

Sport is by no means alone in being an area of news where a lot of traditional punditry and conjecture can be improved upon by incorporating quantitative information, but it does have the relatively rare benefit of this information being quite widely available. Dedicated sports fans have compiled some incredibly detailed datasets over the years on dozens of sports, and when you add these to the sporting results published by governing bodies and media organisations, plus dedicated sports data collection businesses, you have a gold mine of resources with which to find and tell stories, or fact check widely believed myths.

One could of course argue that the same type and volume of data is available for financial reporters, but the difference here is that both the readers and subjects of financial journalism have long been accustomed to using statistics to prove or disprove a point, whereas many sports-people, sports journalists and sport fans have yet to warm to the idea, particularly outside of the US.

As a result, I feel there’s a golden opportunity at the moment for sports journalists to add a little more quantitative analysis to their repertoire — or for data journalists to collaborate with sports desks — in order to better serve their audience.

I think it’s fairly clear that there’s a growing appetite from readers for this kind of work, and I’m optimistic that the growing number of fantastic pieces of stats-assisted sports reporting will help persuade editors that there’s value here, too.

- In which sort of sports and topics data journalism has more scope to be developed according to your experience in FT?

- I think we’re likely to see more and more sports undergoing something of a statistical revolution. Fairly advanced statistics are already widespread in mainstream US sports journalism, sometimes woven into long-form features, other times as standalone tools. As I see it, there are two fairly good indicators of a sport that’s about to go mainstream in terms of stats writing: availability of data (in this case thanks to Jeff Sackmann’s tireless efforts), and an active blogging community. Staying with tennis, BuzzFeed’s attempts to find suspicious patterns in betting activity (and responses to their work) demonstrate another way sport- and data-journalism can work well together.

Moving away from sport, politics is one area where there is both a wealth of data-driven work already going on, and scope for real growth. The stronger tradition of Computer Assisted Reporting (CAR) in the US means it once again has a head start here, with several major news organisations running in-house polling models, but the recent UK general election was a first in terms of seeing the BBC collaborating with statisticians and the New Statesman running a data-heavy site of their own.

Elsewhere, we’re increasingly seeing major news organisations either collaborating with external, non-journalistic bodies, or using techniques such as web scraping, to bring quantitative techniques to other areas of their reporting. Each of these trends, in its own way, is the result of having more data-savvy people in editorial roles in newsrooms, and as a result I expect both to continue to grow.

- FT Data always bet on Sports competitions and results (not only the industry and all their economic dimension) as one of their favourite fields of development? Why?

- Good question, but the answer perhaps owes more to circumstance than design.

Before September 2014, the FT had been steering clear of covering sport as a topic in its own right for several years, though coverage of the business of sport had of course continued. But as part of the redesign of the print product, which was re-launched that month, the decision was made to reintroduce a small sports section on the back page.

Alongside that move came a feeling that as well as a traditional sports column, it would be good to give readers something that might differentiate our offering from those of other mainstream newspapers. Fortunately for me and my FT Baseline co-founder Gavin Jackson — both of us comfortable finding stories in datasets — it was decided that this unique angle should be stats.

In the eighteen-months since the launch, alongside our primary jobs one or more of us (fellow sporting stats geek Rob Minto is also a regular contributor) has the weekly task of generating an idea, [usually] building a dataset from scratch, analysing it, writing up a story and producing graphics for the web that can also be repurposed for the paper by our wonderful print graphics team. It can become a challenge to fit everything in, but it’s great fun!

So in short, I think it’s fair to say the general opinion of the FT is that sport is something that can be a nice add-on, but will always be secondary to our core responsibilities of reporting on the events, people and places that are shaping the business world in which our readers operate. Luckily for me, feedback on Baseline has so far been great, so the challenge is to continue to keep both its readers and my editors interested!

"The thing to remember is that most of your prospective readers aren’t anywhere near as interested in the numbers side, they just want to know what you found out and what it means for them"

- And how about the story? How important is the story when so much data is handled? Is it possible a storytelling only with data?

- I really can’t overstate the importance of a good story and strong narrative, however nerdy the work that underpins them. After putting a lot of time and effort into a particularly tricky data-gathering or analysis exercise, there’s a danger that you stay in that mindset when writing the story itself, ending up with something closer to a scientific paper than a piece of accessible journalism.

The thing to remember is that most of your prospective readers aren’t anywhere near as interested in the numbers side, they just want to know what you found out and what it means for them, or the person/place/thing concerned. I would even go so far as to say that in many cases, you want a reader to be able to get through the entire piece without knowing you even did any quant work behind the scenes.

For me, one of the most skilled writers in this area is FiveThirtyEight’s Benjamin Morris, whose features maintain a flow that allows stats-averse readers to appreciate the story without needing to head down to the footnotes.

Another good example would be this piece by my colleagues Erika Solomon, Robin Kwong and Steve Bernard, who took data on the extraction, processing and transport of oil in territories held by ISIS, and turned it into a fascinating and extremely numbers-light feature, with particular attention given to crafting a narrative that would compel readers to keep scrolling for more.

- Working with data means also applying a scientific method. What is the methodology you and your team follow in FT to bring the appropriate data into a feature or a report? What are those steps?

- Another great question! One of my favourite quotes about data journalism is from Steve Doig, who describes it as “Social science done on deadline”. I’ve always tried to stay particularly faithful to this line, since I believe it is where data journalism can add the most value: not simply in reporting that number A is greater than number B, but in attempting to prove or disprove theories, and to explain why patterns in datasets are or are not meaningful.

In terms of how we ensure our work meets the required standards, there are a number of practices and influences to consider. Just like any of our colleagues in the newsroom, we on the data/interactive team are driven by a desire to produce journalism of the highest possible standards of accuracy and presentation, because anything less would mean failing our readers. For me there is something of a “This is Anfield” effect from the dual considerations that 1) our readers are paying good money to read out work, and 2) my colleagues are very intelligent, work very hard and hold themselves to very high standards.

A separate and equally important influence specific to my sports data work is the blogging community. The quality of prose, analysis and visualisation on football blogs, for example, is extremely high. Conversations on Twitter flow back and forth for days about data visualisation techniques, methodological idiosyncrasies and the interpretation of results. I try to play an active part in that community, and I’m driven as much as anything else by the pressure to ensure that my work is as rigorous, as valid and as interesting as theirs, without simply re-treading ground they’ve covered.

I’ve also discussed ideas for the FT column with leading sports analytics bloggers and have no qualms about seeking their advice on some of the more complex analysis techniques I’ve used. In one recent case an analyst was kind enough to let me use a proprietary dataset of his to validate and illustrate a point I was making.

- In what way such a global event like Olympics is also an opportunity for data journalists to do different things than usual in media?

- For me there are three primary ways that intelligent use of data can add value here.

First, a chart or a statistical story can quickly allow the reader to become an ‘expert for the day’, as it were. This is what I set out to achieve with my piece for last year’s World Athletics Championships on the fastest men in the world. The aim there was to make sure anyone following the men’s 100m would know exactly why the Usain Bolt v Justin Gatlin narrative was such a big issue, and would also be able to impress their friends with their knowledge of the up-and-coming youngsters.

Of course, any form of journalism can achieve the same result, but the old aphorism that “a picture is worth a thousand words” rings true here. The digital news distribution landscape in 2016 dictates that news organisations often need to tell their story without a reader ever actually visiting their site. During an event like the Olympics this need is even more acute, since huge numbers of people will look to the pace of social media to keep them up-to-date with the dozens of events taking place each day. A clear graphic has a good chance of standing out in that stream of information, and can be consumed in full in situ.

The next way in which data can open up new avenues is through personalisation. The BBC made a fantastic tool for London 2012 allowing users — whether experts or just passing by — to see which Olympic sport their body shape would be most suited to. Others have since superbly executed similar pieces for different sports. The key here is that huge sporting events are one of the few topics for news organisations that hit the sweet spot for personalised storytelling. As mentioned earlier, sports are very well served with data, and when you add to this a vast audience with enough reference points in the subject matter to engage enthusiastically with an interactive tool, the rest is a case of ‘build it and they will come’.

Finally, we have predictions. FiveThirtyEight are streets ahead of the mainstream media pack when it comes to data-driven predictions and ratings, sports or otherwise, but The New York Times, Bloomberg and other types of media organisations such as Infostrada Sports have all done great work in this area too.

As we’re seeing with the 2016 US election, debate is fierce and endless over who’s winning, who’s losing and whose opinions on the above you can and can’t trust. Whilst the end result of a major sporting event like the Olympics is infinitely less significant than that, the back-and-forth during the run-in is similarly heated, and as a result there is huge demand for predictions. Evidence has shown that — on average — statistical models outperform even expert opinion, and as such, there are huge opportunities around major sporting events for data journalists and their employers to establish reputations for accurately predicting the future(!).

And here some of the best data-driven pieces recently published by John Burn-Murdoch on FT: