Around ten in the morning this past Labor Day, the sky above Boulder, Colorado turned a dusty orange. A fire had sprung up in nearby Fourmile Canyon, and was spreading. Fast. Black smoke billowed up, visible for miles. Since it was a holiday, many Boulder residents were out of town. Likewise, government offices and news outlets were either closed or incredibly short-staffed, causing an understandable but dangerous lapse in communication and emergency response.

At first, the websites of local news outlets like the Boulder Daily Camera, The Denver Post, and Denver’s didn’t have any information. As a result, the town saw the smoke and blaze out their windows before they knew exactly what it was, where it was, or how quickly it was spreading. So many people there did what they’d been conditioned to do when they need answers fast; call friends, blast something on Facebook, or click around on Twitter until they make a connection.

As the fire spread throughout the following week, residents turned increasingly to these social media outlets for information, photos and maps. Local newsrooms struggled to keep up with the flow of unconfirmed reports and rumors, and delayed reporting those that they could not verify; some readers weren’t satisfied with their performance. After the smoke cleared, those who covered the fire, both the professional journalists and those trading information on social media sites from home, have asked who better served the community’s needs during this fast-changing emergency, and what they can both learn for next time. While social media cheerleaders see Twitter and other tools as essential for keeping up with the news cycle, more traditional news outlets maintain that they did the right thing. Editors at the Daily Camera and the Post both said that even if they had had more staff on this story, they would not necessarily have dedicated the extra manpower to watching or reacting to Twitter reports.

Probably the most prolific social media user in the first few days of the fire was Sandra Fish, an instructor at the School of Journalism and Mass Communications at the University of Colorado. When she saw smoke, she tuned in to the Boulder County police scanner, which is broadcast online to the public. Fish was used to listening to scanners from her time as a full-time journalist—she wrote for the Daily Camera and the now-defunct Rocky Mountain News, among other papers—and she also teaches her students to use Twitter to take notes during a breaking news story. She instinctively started tweeting what she overheard on the scanner about the fire: information about its location and size, and evacuation instructions from the authorities.

Some tweets had specific information, some were educated guesses based on what she could make out. Some were just quotes without context. Here are three samples of the hundreds she put out that day, one every few minutes:

scanner: hearing explosions. “unless we get some eyes in the sky, we’re really throwing dirt at this thing.” #Boulderfire
Mon Sep 6 17:40:36 2010

scanner: “we’ve definitely got an hour or so before it gets really bad.” sounds like someone near Sunshine? #Boulderfire
6:00 PM Sep 6th

scanner: fire running parallel to Gold Hill and Sunshine Road. “Everybody west of Gold Hill should be evacuated.” #Boulderfire
6:38 PM Sep 6th

Fish also re-tweeted a lot of information from the authorities, who were themselves using Twitter and Facebook to get information out to residents. (The more traditional “Reverse 911” system that autodials home phones with emergency alerts failed, in many instances.)

Meanwhile, more and more people joined in and traded tips: everything from offering a place to stay or a free meal to those who had been evacuated, to passing along information about how people could help—things like ‘please don’t come bring cookies to the firefighters because they are busy’. A lot of people also took advantage of Twitpic to post photos of the blaze from where they stood.

After a few false starts, “#boulderfire” was the tag soon adopted by everyone trading information about the fire online. A whole ecosystem of information developed, quickly and organically. In one especially impressive feat of crowdsourcing, by 1 p.m. that day, graduate student Amanda Pingel had put together a collaborative Google map of the areas affected by the fire, embedded with geo-specific alerts and residents’ photos.

There is no way to know for sure exactly how many people evacuated from their homes as a result of seeing an alert on Twitter or Facebook, as opposed to another mode of communication. But social media certainly had a wide reach. By social media analyst Tery Spataro’s account, the Twitter tag “#boulderfire” was seen by 985,000 people and was used by 688 Twitter users within the first five days of the fire.

Lauren Kirchner is a freelance writer covering digital security for CJR. Find her on Twitter at @lkirchner