Every newsroom’s got them: A few dozen gadflies who’ve been in town forever and are proud to have their favorite reporters on speed-dial.

The little team at New York City’s Web-only Gotham Gazette — two reporters, two geeks, and a boss — wants to recruit more people like that. In fact, they want to train them. And they think the way to do it is with a closely edited wiki.

The Gazette’s plan for Councilpedia, a planned guide to the filthy lucre that links real estate and politics on New York’s city council, just made the Gazette the first two-time winner of a Knight News Challenge grant, this one worth $250,000 over two years. (Editor-in-chief Gail Robinson’s team won the same amount in 2007 for a series of educational Web games, such as one that asked readers to balance the city budget.)

The idea is to combine the anyone-can-contribute model of Wikipedia with the editing and fact-checking that marks good journalism. The hope is that by directly enlisting the eyes and ears of the public, Councilpedia will uncover watchdog stories that would have otherwise gone unnoticed.

I talked with Robinson, a veteran journalist, and her top geek, Amanda Hickman, about their strategy for launching a topical local wiki. Here are the six most interesting choices they’ve made:

— Keep it simple, stupid. While Wikipedia’s increasingly strict citation system has made the world’s biggest wiki less accessible for inexperienced editors, the Gazette says they’ll be fighting that tendency with their interface, which will be built on the same MediaWiki software Wikipedia uses.

“We’re counting on a body of users that doesn’t necessarily have any technical expertise,” said Hickman. “I think the project will succeed or fail based in large part on the usability of our edit screens. If it’s too hard to put information in, or it feels like they’re always doing it wrong, then they’ll stop.”

— Choose one problem to solve together. One reason for Wikipedia’s success: every volunteer editor knows intuitively what an encyclopedia is supposed to do. By setting narrow goals for Councilpedia’s content, the Gazette hopes to signal to their own volunteers how to direct their energy.

“We don’t even particularly want to know if the council member kicks his dog,” Robinson said. “People are less likely to respond [to], ‘Tell us what you think about things that are going on in New York City!’ If you say to people, ‘Who’s giving money to your council member?’ people will show up.”

— Think of a wiki as community organizing. Robinson said the first six months of Councilpedia will be an “intensive outreach” effort, with paid interns fanning out to civic groups across the boroughs to promote and explain the site.

Hickman, who worked for three years as a community organizer before becoming the Gazette’s technical director, expects to spend about half her time in 2010 administering the site.

— Focus on landing one blockbuster success. The Gazette folks are eager to start their wiki grinding at a few particular rumors that, if confirmed, would spread like fire through local media.

“If we can really tease out one story that really grabs people’s attention, then I think we’ll get a flurry of activity,” Hickman said.

— Make it useful immediately. To win loyalty, Councilpedia would need to become something people feel they “weren’t able to do their work without,” Hickman said.

She and Robinson aren’t sure yet how much information they’ll need to seed the site with to start attracting contributions, but they’re certain no one would contribute to an empty site.

— Keep it accurate. Unconfirmed data and constant refinement may be good enough for Wikipedia, but it’s not good enough for the Gotham Gazette, Robinson said.

She has no idea how much staff time it’ll take to confirm every claim that is made on the wiki. But their goal is to hold everything to the standard of anything else on their web site. Anything less, Robinson said, would be unethical.

But if the Gazette’s staff swoop in to police every statement made on Councilpedia, will volunteers lose the sense of ownership that motivates their participation?

And just how accurate will the information people post turn out to be?

That’s exactly what the Knight Foundation is paying them to track, Robinson said.

“It’s always very surprising what people buy into and what they don’t,” she said. “It’s anybody’s guess at this point.”

The Knight News Challenge’s biggest winner, with a two-year grant of $719,500, is DocumentCloud, the primary-source index conceived by journalists and developers at ProPublica and The New York Times. Here’s why you should care: There’s good reason to believe the project will transform how some investigative journalism is conducted — and who conducts it.

Like a lot of software in the cloud, this one is complicated to explain. I wrote a long overview of DocumentCloud in November, and you can read their initial grant application in my first post about the project. Aron Pilhofer, editor of interactive news technologies at the Times and one of the project’s creators, told me on Monday, “DocumentCloud isn’t really conducive to a two-minute elevator pitch.” But later in our conversation, he ventured one: “It will turn documents into data.”

In the analog version of investigative journalism, a reporter obtains documents from sources and freedom-of-information requests, writes a story, and… that’s it. If we’re lucky, the materials are posted as unwieldy and barely searchable PDFs.

DocumentCloud’s vision is to collect, archive, and index the text and metadata of all documents used by participating news organizations, advocacy groups, bloggers, and others — “so they’re not just sitting in the corner of a newsroom collecting dust,” Pilhofer explained. That way, anyone — from other news outlets to curious readers — will be able to search across all documents in the project to find information that might not have been relevant to the original piece. If it were an animated TV series, the catchphrase might be, With our newsrooms combined — we are DocumentCloud!

Early partners in the project include the Times, ProPublica (the non-profit investigative journalism outfit) Gotham Gazette (a New York City news site published by Citizens Union Foundation, themselves winners of two Knight News Challenge grants), TPM Muckraker (the investigative arm of Talking Points Memo), and the National Security Archive (home to the largest public repository of declassified government documents). Are you salivating yet?

Anyone who has waded through the National Security Archive’s wealth of FBI files and CIA reports will immediately recognize the benefit of DocumentCloud. What if you could search across the entire archive for a particular topic of interest (Pilhofer suggested Marilyn Monroe) and get pinged whenever that topic shows up in a new document? Or what if you were a business journalist up to your neck in SEC filings? Or a local blogger keeping tabs on your congressman’s earmarks?

Details are still being worked out, but running materials through DocumentCloud will involve some sort of optical character recognition (to render those pesky image-based PDFs favored by some government agencies into searchable text) and Open Calais (to extract metadata like names, locations, and dates for more effective indexing). The news organizations that contribute documents will likely host them as well, said Scott Klein, editor of online development at ProPublica and a creator of DocumentCloud. He described the project as more of a “card catalog” than, say, a repository.

There are other aspects of the project that I’d be happy to discuss in the comments, and maybe we can get Pilhofer and Klein to weigh in here as well (as though I’m not hyping them enough). The other creators are Eric Umansky of ProPublica and Ben Koski of the Times. So if you have any questions about DocumentCloud, feel free to ask, and I’ll work on getting answers.

UPDATE, 3:30 p.m.: The Document Cloud folks introduced their project at the Future of News and Civic Media Conference at MIT today, and I shot some raw video with Qik: