Counting the number of July 1 marchers is difficult for those who want to do it right because, as statistics expert Paul Yip Siu-fai puts it, whatever he announces, everyone says it's wrong.
The University of Hong Kong social work professor said he had thought seriously about not counting this year's marchers.
It has been a thankless task since his first one in 2003, when half a million took to the streets against the controversial national security bill, which was eventually shelved.
'Whatever I do, people politicise it,' Yip said. People were obsessed with the biggest number, he said, so the party releasing the largest figure should be viewed with scepticism.
'Everyone likes the big numbers, including your newspaper. A big number is a good number because it's sensational. I have considered not doing it ... it's hard to swallow people's accusations, especially when they aren't reasonable,' he said, referring to some rally organisers' remarks in previous years that Yip had 'deflated the number'.
But after much consideration, Yip decided he could not abandon a job he has done for nine years.
His method involves counting the number of people who pass a certain point in Causeway Bay and Admiralty for a minute at 15-minute intervals. However, this does not cover protesters who start at Victoria Park and leave the march before Yip's first checkpoint in Causeway Bay.
His team found that in last year's march, about 200 people passed a certain point in a minute. And since the rally took five hours, his final estimate was 60,000 to 70,000 people.
The police said 54,000 marched.
'[The organisers] said there were 218,000 people, which meant 1,000 people passed through a place in a minute. That's impossible. It doesn't make sense,' Yip said. 'The media and those who don't like my findings do whatever they can to bad-mouth me. That's why I insist on releasing my findings in academic journals.'
His methods have been detailed in the Australian & New Zealand Journal of Statistics and Significance. His doctoral thesis involved estimating the number of birds in a forest, and he said he was confident he could estimate the number of jelly beans in a jar, given the right sampling method.
'Our count has its limitations, but it is the most verifiable of them all. But when people don't like the number, they connect it with my being part of the [government think tank] Central Policy Unit,' he said.
The method used by HKU's public opinion programme involves six researchers counting protesters as they pass a footbridge in Wan Chai.
The Civil Human Rights Front, the rally's organiser, deploys 10 volunteers at four points along the route of the march, and the average of their results is taken as the final count.
A police spokesman said officers used the size of the venue and how crowded the rally was to reach an estimate.