UK newspaper, The Guardian, has done some pioneering work to use data, and to engage readers in exploring data to share their own insights. The paper’s Simon Rogers and Google’s Kathryn Hurley shared some of the lessons at Strata this morning.
Rough notes follow.
Not going to talk about big projects like riots and Wikileaks and MP’s expenses… Going to talk about the day-to-day process of hacking around with data.
Open data journalism – more than just Google spreadsheets. Much more of a two-way process than simply writing and disseminating stories.
Numbers need context. Journalists need the skills to interpret, probe, and tell a data-backed story.
First rule of what we do – find the key data behind a story and make it public. Guardian Datablog and Data Store used to push out data relevant to the main news stories of the week.
Lots of data is available, but it’s locked up in a wide range of data sets. A lot of the Data team’s work is involved with pulling freely available data together in one place – making it comparable and useful.
Get past raw numbers, and show how they have changed over time. Measures and units and groupings change, so how do you actually compare like with like?
Don’t always just rely upon the algorithm… Need the knowledge and the question-asking capabilities to wonder whether or not the result is too good to be true. Often, it will be wrong.
Olympics… lots of data, but very little was open. IOC sold the data, and refused to allow it to be shared.
Kathryn Hurley at Google… spent the last week working directly with the Guardian team… Learned…
- News drives the stories
- Data journalism moves fast
- Quick and easy tools reign supreme
What does this mean for other businesses?
- Know what matters
- Find the data to back it up (internally, from government, from public data sites, from data markets, etc)
- Clean the data (a lot! Sometimes just normalisation, sometimes more serious)
- plugging tools like Google Refine
- Sometimes the data you have isn’t enough – find more
- Tell the story – visualisation matters, interactivity helps
- Sharing the data to support your story – make it available for download, or offer an api
Tools need to get easier to use and richer, to let data journalists (and others) get the results they need more quickly, and with less coding.
Published data needs to be more logically formatted… PDFs derived from printed documents are designed for human reading, not for machine processing.
- Data journalism at the Guardian: What is it and how do we do it? (nextlevelofnews.com)
- Open data journalism (guardian.co.uk)
- Four key trends changing digital journalism and society (radar.oreilly.com)