Photography: Anirudh Koul
In 2009, UK newspaper The Guardian was in trouble. Its rival broadsheet The Daily Telegraph had been publishing a series of spectacular accounts from leaked documents that exposed expenses fraud by some of the country’s politicians. It was the political story of the decade and the Telegraph had an exclusive that enabled it to grip the nation alone — at least until the government spoiled its party by making all of the documents available to the public.
When that happened though, The Guardian, like the rest of the press, was faced with the task of reading 457,153 pages of dumped expenses reports in order to separate MPs’ bills for paperclips and printer ink from claims for private moats and duck houses.
Unlike other media outlets though, The Guardian turned to its readers. As Michael Anderson of the Nieman Journalism Lab at Harvard University explains, the newspaper placed all of the documents on its servers then invited volunteers to sort through them, marking each page as “not interesting,” “interesting but known,” “interesting” or, most urgently, “investigate this!”
Within 80 hours, Guardian readers had sorted through 170,000 documents with a visitor participation rate of 56 percent.
It was a perfect act of crowdsourcing, completing an important task that the newspaper could not have done alone. And it worked because the project combined all of the elements necessary for a successful appeal to the masses.
Know Your Crowd
The Guardian’s readers are a feisty lot, rebellious, anti-establishment, and keen to point their fingers at greedy politicians. Giving them an opportunity to dig up evidence that an MP had been cheating allowed them to play at investigative journalism. It was the kind of request that would have appealed precisely to the newspaper’s readership.
Finding the right crowd is usually relatively straightforward. Website visitors are self-selecting, but seed social media with hashtags (as The Huffington Post did to write some headlines) and you’re likely to be sorting through a lot of irrelevant contributions to find the ones worth noting.
It’s a problem that might be solved by crowdsourcing the selection process too, allowing visitors to vote for the submissions they like the most. But if the risk of working with crowds is that the free work is worth what you’re paying for it, then targeting your audience carefully will help to reduce the poorer offerings. If Wikipedia had a less open contributor policy it might be less democratic but its entries might be a lot more valuable.
Make It Fun
The Guardian’s project got off to a flying start but it really took off when developer Simon Willison added each politician’s picture to their pages in the database. The crowds were no longer looking at random numbers; they were plowing through invoices filed by real people, often their own representatives.
It’s a strategy of personal engagement that creators of other crowdsourced projects believe to be essential. Threadless, a t-shirt company that uses a community of 100,000 graphic designers to source designs and highlight the best submissions, recently launched a new platform to help non-profits and campaigns make use of crowdsourced designs for their campaigns. The DNA Foundation, for example, is looking for submissions to raise awareness of child sex slavery.
“One thing is that it is very important is to present challenges that are inspirational to the creative community,” Threadless CEO Tom Ryan told Techcrunch. “It lets the artists get excited about what they are designing.”
A successful crowd-sourced project then has to match the organization to the right crowd, and that crowd to a project that they find personally engaging.
But that personal engagement doesn’t have to be driven by the desire to donate time to a good cause. The Huffington Post, which effectively crowdsources its entire content, used participation to build engagement with its brand and win traffic. The Economist has recently taken a similar route. Each week the newspaper now runs a weekly online competition that invites readers to submit a witty caption for one of its images. The contest is regularly one of the most commented posts on the site. It’s unlikely that the newspaper really needs editorial help — it seems to do fine with its other images — but it may well need assistance building its page views.
Make the Request Simple — and Urgent
Crowds are working for free which means that organizations who operate them can’t make the demands too onerous. [email protected], one of the most successful crowdsourced projects, merely requires computer users to leave their machines running at night so that its software can make use of their computing power. The Guardian didn’t ask its volunteers to write summaries of each document but only to give them one grade among four, and Flickr users helping the Library of Congress to categorize its image collection just have to add keywords to the tags.
Make the demands more challenging — as Wikipedia does — and you run the risk of creating dedicated elites within the crowd who are more willing to invest their time and effort than others. It’s a system that works for Wikipedia but might not be suitable for other crowdsourced projects like those run by Threadless.
PR expert J.W. Alphenaar also recommends that the crowds be given a deadline. Without that urgency, he argues, there’s no action and no results.
That may not always be essential though. The Guardian, which needed to dig up scoops while interested lasted and before its rivals found them, motivated its volunteers by creating leaderboards to show who had managed to sort through the largest number of documents. Threadless’s campaigns show clearly how many days designers have left to submit their entries.
But the Library of Congress’s images are going to remain on Flickr, picking up more tags all the time, in the same way that Wikipedia will remain an ongoing, continuous project. Urgency is useful when the work needs to be completed but not every crowdsourced project has to come with an end-date.
The real secret of successfully using crowdsourcing then is to pick the right project — ideally one that only works with large numbers of people — make the task simple, the goal interesting and the crowd packed with the right sorts of people.