The crowd of protesters in Hong Kong on Sunday stretched more than a mile. As the city’s legislature considered a bill that would allow extraditions from Hong Kong, which is semi-autonomous, to China, Civil Human Rights Front, the group that organized the demonstration, estimated that more than a million people had come out to march. Hong Kong police, however, put the number at 240,000 people.
Crowd size is a marker for energy around a cause. “It’s hard to measure enthusiasm based on speeches or noise level,” says Steve Doig, who specializes in data-based crowd-counting as a professor at the Walter Cronkite School of Journalism and Mass Communication at Arizona State University. “So the size of the crowd is the token that various sides will want to inflate or deflate.”
In 1996, Congress banned the National Park Service from releasing crowd size estimates after the activist Louis Farrakhan threatened to sue over its calculation that the Million Man March attracted only 400,000 people. (Researchers at Boston University did an independent analysis and found that the crowd was closer to 800,000.) Donald Trump’s claim—that the audience for his inauguration in 2017 was the largest ever—was clearly not true; but his urge to surpass previous numbers was based on comparisons to other fictitious crowd counts. Lyndon Johnson’s swearing-in supposedly had 1.2 million attendees, in 1965. “Those are impossible standards to beat,” Doig says. “The claim that ‘I’m the most popular’ becomes ridiculous.”
Large crowds typically aren’t contained neatly enough for an easy body count unless people gather in an arena or convention hall. “The gold standard is to measure the area containing the crowd and then make a reasonable estimate of the density,” Doig says. “It’s simple math, it’s reality-based.” That’s why parades, or migrating marches, are the hardest of all to quantify, for journalists and police alike.
Doig has a few go-to ways of making his calculations. His first tip: learn The Jacobs Method. “Herbert Jacobs should be enshrined,” Doig says. In the 1960s, Jacobs was a professor of journalism at the University of California, Berkeley, whose office overlooked the plaza where protests against the Vietnam War were held. Bricks divided the ground into a pattern of squares. As Jacobs looked down, he created a technique: in a loose crowd, measure one person per ten square feet. “It sounds like a lot of space,” Doig says, “but you’re arms length from everyone’s shoulder.” In a tight crowd, each person has five square feet. “You’re really in each other’s space then,” he says. “There are a lot of ridiculous estimates for one square foot per person, but that’s beyond mosh pit density.”
Doig also likes to imagine the view from a balloon in the sky. Standing at head level, as a new president would during an inauguration, makes crowds look immense—“a sea of humanity,” Doig says. An overhead shot gives better perspective—though that’s hard to come by in Washington, DC, where there are no helicopter flights and no tall buildings; the capital region is also a no drone zone.
He’ll also enlist a team. In 2010, for instance, Doig covered labor protests in Lisbon with the help of several students. They came up with a method based on time: “Ten seconds is pretty much the limit of attention span,” he says. His students positioned themselves at different points along the parade route and counted how many protestors marched by every ten seconds. The team checked for consistent crowd density, and later they compared numbers, before multiplying their tally by the length of time the march took. In the end, the labor unions claimed to have 100,000 protestors. Doig’s team estimated that the actual number was closer to 10,000.
In Hong Kong, because the protests moved through so many streets, the real challenge, Doig says, is in knowing the “available population.” In 2017 Hong Kong had 7.4 million residents. “If they’re claiming 1 million marched, ask yourself if truly one in every seven people is here,” Doig says. “Ask: Are enough people not here to explain how everything in this city is still operating? How many of the population are kids in school right now, not on the street? What percentage of the total population are able-bodied adults?”
It’s also worth checking Google Earth, and trying to measure the potential area in which a crowd can gather. Using this method when reviewing a photograph of the Hong Kong demonstrations (above) published this week in The New York Times, Doig figures that the space in view, Hennessy Road, is 210 meters long and 25 meters across, or some 5,500 square meters. At the front, people were a little less than an arm’s length apart, he observes—tight for a parade, but not as dense as a standing crowd—which would make for one person per square meter (roughly ten square feet, per the Herb Jacobs rule). “An Asian crowd tends to be filled with people who aren’t as, er, large as us beefy Americans,” he says, “and they are more accustomed to being in tighter quarters. So I am arbitrarily going to call this a density of 0.8 square meters per person.” That makes for about 7,000 people in the photo.
“That doesn’t help estimate the total crowd that day because we have no idea of all the streets that may have been filled,” he adds. Buildings in Hong Kong are so tall that satellite images can be tough to collect, and overhead footage becomes nearly impossible to decipher. From what he’s seen from sources on the ground, he would accept conservative estimates—of some quarter of a million protesters in Hong Kong. It would be impossible to make a reliable calculation himself. If 7,000 people fit in 210 meters, there would need to be 142.9 street segments, or about 18.6 miles of streets packed at once to count a million protestors, which seems like too many. None of this is exact science, Doig says. “Is that what happened? Dunno.”