Post needs to get over fear of crowd estimates

By Andrew Alexander
Friday, November 5, 2010; 7:44 PM

"As I look out here today, I can see we have over 10 million people," Jon Stewart joked to the crowd on the Mall at last weekend's "Rally to Restore Sanity and/or Fear."

Some readers seemed to take him seriously. They insisted that the rally was one of the largest in Washington's history and complained to me that The Post didn't say so.

Crowd counts, inexact and exploitable, are a no-win problem for the media. Event organizers tout their own estimates to promote their cause. If a news organization's estimate is lower, it gets accused of bias. If it's seen as too high, the charge is favoritism.

Crowd size has become such a critical metric of a rally's success that it sometimes seems more important than why the event was staged. Style Editor Ned Martel, who directed much of the Oct. 30 rally coverage, said Post reporters and photographers "were accosted throughout the Mall by people who wanted to know the crowd size." Estimates have become so controversial that the National Park Service no longer provides them.

Tim Curran, the editor in charge of the Sunday A-section, insisted that The Post avoid crowd estimates in its coverage. Absent a credible count, it was the right call.

Instead, The Post offered broad descriptions of packed Metro trains and clogged entry points to the Mall. Beyond that, Curran said, "we're not going to try to pretend that we're in a position to make an accurate crowd estimate."

"To me" he said, "it's an attempt to be as sensible as we can be and [avoid] claims of leaning one way or the other."

While the printed Post avoided estimates, the online Post encouraged them. After the rally, a "User Poll" asked, "How many people do you estimate attended" the event?

Unscientific user polls, more entertaining than enlightening, are intended to engage online readers. But some found this one silly because it encouraged participation by those who had no clue of how to estimate crowd size and may not have even attended the rally.

Ann Chih Lin, an associate professor at the University of Michigan's Ford School of Public Policy, objected: "It's akin to sending out a message on the Internet saying, 'I have a jar of jellybeans on my desk. You don't know the size of the jar or the size of the jellybeans. Guess how many are in the jar.'"

Mike McPhate, The Post Web editor who wrote the question, said most readers know that "user polls" offer more fun than fact. "I think that people recognize that this is sort of like a game," he said. "I didn't see the harm in letting people weigh in on what they thought."

Conventional crowd counting techniques are fraught with imprecision, especially on the Mall. Aircraft restrictions limit direct overhead photography, and merely extrapolating from a crowd's geographic dimensions can yield little more than guesstimates.

But new computer-analyzed aerial imaging is producing what experts say are highly accurate crowd counts. The Post should embrace the technology and get in the game, for two reasons. First, it should lead in coverage of rallies in its back yard. Second, providing credible crowd counts is informative and counters unchallenged, publicity-driven estimates from organizers.

CBS News has used Digital Design & Imaging Service Inc., a Falls Church firm that employs devices such as satellite images and balloon-tethered cameras. The super-clear shots provide head counts in grids, and computers analyze areas of similar density to produce precise estimates.

Conservative commentator Glenn Beck claimed that his late August "Restoring Honor" rally in Washington attracted "between 300,000 and 500,000" participants. But Digital Design's estimate for CBS put the crowd at 87,000. Its estimate for last weekend's "Sanity and/or Fear" rally was 215,000 and was briefly noted on The Post's Web site.

"We have a very sound methodology," said Curt Westergard, president of Digital Design, which makes the data for its crowd counts available online for the public to evaluate.

"It behooves the hometown paper to make a stronger effort on crowd estimation," said Steve Doig, an expert in computer-assisted reporting at Arizona State University's Cronkite School of Journalism. Doig, who has done work for Digital Design, suggests that The Post and other major media outlets form a consortium to share the costs of using cutting-edge technology.

Estimates will always be challenged. Were people obscured by tree cover accurately counted? What about rally-bound participants stuck on Metro trains? Anything that moves estimates further from the jellybean jar is still worth exploring.

Andrew Alexander can be reached at 202-334-7582 or at For daily updates, read the omblog.

© 2010 The Washington Post Company