Photo: RNZ/Samuel Rillstone
Over nine days, people from across the country made their way to Wellington to protest ACT's Treaty Principle's Bill, which aims to legally define the principles of Te Tiriti o Waitangi. Hīkoi mō Te Tiriti was said to be a march against the government's policies affecting Māori and for tino rangatiratanga.
On 19 November, participants gathered at Parliament. A sea of red, white, and black, the crowd filled the forecourt and surrounding streets. It was obvious this was one of the country's biggest protests. But exactly how big became a topic of debate that day, and for days afterwards.
Police estimated more than 42,000 people took part. Meanwhile, Te Pāti Māori co-leader Rawiri Waititi estimated 100,000 were there, and New Zealand First's Winston Peters said 22,000.
Why the discrepancy? Large crowds are difficult to count, and the numbers are often politicised, according to experts.
The police estimate
Crowd estimates aren't an "official responsibility" of police, a spokesperson told RNZ, but were provided to media to assist with reporting. The numbers "are only ever estimates based on various information available... at the time and may include subjective opinions".
On 19 November, as protesters walked from the capital's Waitangi Park to the Beehive, police used a map showing the surface area of the Parliament precinct and surrounding streets.
"A formula of 1.5 people per square metre resulted in an estimated crowd size of nearly 42,000 people.
"Earlier that week, on the Auckland Harbour Bridge, hīkoi participants were escorted across the bridge in groups of about 250 people which allowed transport agencies and police to estimate approximately 5000 people."
Counting from afar
Caleb Moses (Ngāpuhi), a data scientist and, currently, a PhD student in computer science at Canada's McGill University, used drone footage to estimate the crowd size remotely.
"I'm personally invested in the outcome of the [Treaty Principles] Bill," he told RNZ. "But, also, it was an interesting mathematical problem."
He likened it to "working out how many jellybeans are in a jar": "First, you work out the size of the jar."
Moses said the footage was shot by a friend in Wellington (and he noted the flight complied with Civil Aviation rules). Using Google Maps, Moses drew boundaries around the crowd.
"Then, you look at how tightly packed the beans are in the jar."
After trying machine-learning approaches, he found counting people by hand was more accurate. "It's hard to measure how tightly the crowd is clumped together, and if you get that wrong you can be off by several tens of thousands in either direction."
In addition, because a protest of this kind only gathers every few years, "you have very little reliable prior data to base your figure on". "Even past crowd estimates are also only guesses, and you don't sell tickets to a protest, so you don't have an easy way to keep count."
The footage provided multiple perspectives of the crowd, from both Waitangi Park and Parliament. Moses' final, conservative, estimate was "around 40,000-60,000 people".
Given the likelihood of future protests, he said, "it seems like this is a problem worth spending some time on".
What can we learn from all of this?
Dr Chris Cocking, a leading United Kingdom crowd behaviour researcher and lecturer at Brighton University's School of Humanities and Social Science, agreed crowd counting "isn't an exact science".
The most effective way to count a protest, in his opinion, is to measure its density and how long it takes to pass a certain point.
Remember, police estimated a density of 1.5 people per sqm. The usual safe crowd density is 2, Cocking told RNZ. The maximum is 4.7. After that, it becomes "deadly dangerous".
"One of the dangers of concerts is people might tolerate dangerously packed crowds without realising it.
"When you get close to those numbers in protests, it's often because police are charging a crowd and there's a surge."
Protest organisers will always overestimate attendance, and people on the political right or left - whatever ideology is in opposition to the protesters - will underestimate it, he said. Simply because it's in their political interest.
Police also tended to underestimate crowd sizes, he added.
How does this influence crowd behaviour and management?
In contrast to the 2022 anti-mandate protest at Parliament, where hundreds of people were arrested, just one person was arrested on 19 November.
Wellington District Commander Superintendent Corrie Parnell in a statement thanked those involved in the hīkoi "for how they conducted their movements" and "the positive way in which they engaged with us throughout".
Cocking told RNZ crowds often behave much better than they're given credit for, and he has spent his career "undoing irrationalist myths about crowds".
"In the United Kingdom, at least, if there's a crowd incident the language used in reporting will be peppered with phrases like, 'hooliganism', 'mindless violence', 'mob', and so on. And these words almost never match up with the behaviour on the ground."
Historically, these myths have guided policing tactics that sought to tightly control large groups of people for fear of violence.
"The risk is that's a self-fulfilling prophecy of escalating behaviour. And it just unites people against police. Disorder escalates on both sides. And it's very difficult to de-escalate a riot."
There was a shift in tactics after Metropolitan Police officers were found to have acted unlawfully in surrounding G20 Climate Camp protesters in London in 2009.
A controversial tactic first used in the 1990s, called kettling, attempted to contain people inside police cordons, often for hours on end. The result was bloody clashes as protesters attempted to break free. A bystander on his way home was pushed to the ground by an officer, and died.
Police have since turned to "crowd psychologist models", Cocking said.
"There's a lot of evidence to show that when police follow crowd psychology models, there's much less violence."