I’m just back from the meeting of the Open Society Institute Information Program sub-board – the board that takes more of my time than any other, as it meets three or four times a year in odd corners of the globe. It’s probably my favorite board as well – my fellow board members and the program staff are incredibly sharp folks and we get to spend a lot of time together, which means we’ve had the chance to become friends as well as colleagues. My friend Jean-Claude Guédon, who was attending his last board meeting with us (we’re limited to two two-year terms to keep the board fresh), observed that one thing he’ll miss about not meeting with us is the way this board keeps us abreast of some of the smartest projects in the technology for social change space.
I was thinking precisely the same thing while listening to Ronaldo Lemos da Silva, Jr., one of the leaders of the open culture movement in Brazil. as he told us about Overmundo, a truly remarkable website dedicated to the arts and culture in Brazil. The site was commissioned to solve an interesting problem – to ensure coverage of cultural events in Brazil outside the major cities of Rio and São Paulo. With funding from Petrobras, Brazil’s state-owned oil company (the largest arts funder in the country), Ronaldo and his colleagues put together a remarkable resource which incorporates some of the smartest thinking I’ve ever seen about citizen media.
Each story that appears on Overmundo is the output of a complex community process which involves several opportunities for community participation. Stories are posted in draft and wait a minimum of 48 hours before being brought live. During that two-day wait, readers can offer suggested editors to the story, as well as voting on whether or not it should be published on the site. If a site reaches the voting threshhold – roughly 30 votes at present, Ronaldo tells us – it gets published.
Once published, the stories are in competition for position on the site’s front page. Readers use a digg-like mechanism to determine which story is the lead on the page. Votes “decay” over time, which allows newer stories to replace the old, but means that a story that remains popular – sometimes because they generate very active comment threads – might remain on the front page for a long time.
The voting system incorporates a karma system, a concept borrowed from community sites like Slashdot and Kuro5hin. Authoring useful stories and comments improves karma – Ronaldo proudly points out that, despite being one of the four architects of the site, he’s not listed until the second page of the profiles page since his karma is so much lower than many other community members. And karma ages – if you were active on Overmundo and dropped off, your karma will be lower than that of active users.
The site is one of the more dynamic and interactive I’ve ever seen – stories generate good comment threads, and Ronaldo tells us that the site gets about 12,000 unique views a day, a great number for a young project. There’s a wide mix of news stories, personal reflections, and works of ficton, music and film posted each week. Overmundo hasn’t experienced the sort of vandalism you might imagine from a site that’s open to public posting – the voting process tends to eliminate noise and spam.
Most remarkable to me – the project began as a paid content site, hiring contributors in each of Brazil’s 27 states to ensure geographic coverage from the whole country. But the site was sufficiently popular a few months in that the team decided to stop paying contributors. Petrobras came back to the project offering them more funds, and they refused, explaining that they’d found a way to keep the site running without inputting more money.
Explaining the community to our board, Ronaldo insists that the challenge in building a site like this isn’t the technology, but the community dynamics. This is pretty similar to something I often say about Global Voices, which is built on pretty simple software, and pretty complicated group dynamics. But it seemed like a bit of an oversimplification in this case, as there’s clearly some very complex thinking about tools which makes the community behavior possible.
Ronaldo conceded that the site took eight months of work to build, most of which was spent planning, not programming. The credits page of the site shows the wide range of tools and projects they were inspired by. Hoping to make other sites like this possible, Overmundo now plans on releasing their tool under an open source software model, allowing other communities to try it out as well.
I wonder to what extent different tools are effective in different communities for different goals. Overmundo is remarkable, but requires a pretty high level of engagement from readers to make the system work. This may work well in Brazil because there’s such widespread usage of community systems, including social networking sites like Orkut – Overmundo clearly has learned a lot from these different social software projects. I can imagine that some dynamics of this community – the idea that votes don’t count equally, for instance – doing very badly in other communities, or the whole system failing to take root because readers are more used to reading passively and not to participating. Still, it’s an amazing project and one I’m looking forward to following closely and learning from.