You can make a difference.
An oilpatch lobby group paid Facebook to boost an online poll that skewed the level of support for Kinder Morgan's Trans Mountain expansion. But the group won't say why it did it and Facebook didn't flag the activity as breaking any rules.
The Texas giant's pipeline has sparked a fierce battle between the neighbouring provinces of Alberta and British Columbia, with new threats surfacing from Alberta Premier Rachel Notley to send B.C.'s gas prices soaring.
The paid-for boost was made through an ad posted to a Facebook page for Canada's Energy Citizens, which is managed by the lobby group that represents Canada's major oil and gas companies, the Canadian Association of Petroleum Producers (CAPP).
While the post remained on the page, the lobby group appeared to stop paying for the advertising a few hours after National Observer asked what it was doing. CAPP declined to answer that question.
Ultimately the poll, run by the Oak Bay News on Vancouver Island, was liked on Facebook thousands of times and by April 18, 2018 nearly 80 per cent of those who participated voted in support of the pipeline. That kind of support is a departure from scientific polls that show more divided opinions among British Columbians.
Sponsoring posts on social media is a common and accepted practice to boost the visibility of a message through advertising (in fact, this publication pays to boost posts on Facebook regularly).
But this ad is only one example of the style of online warfare that has pitted fossil fuel supporters against clean energy advocates in a battle to win the social media newsfeeds of the nation. The online war also includes automated messages and accounts — or "bots" — designed to amplify the reach of targeted statements to manipulate public opinion. Programming automated messaging can promote fake social media profiles to share and repost desired messages, ensuring that people see them while creating a sense that an opinion is popular and held by many people already. In the last two years, people have become more familiar with "Russian bots," or automated message swarming aimed at influencing public opinion, most notably in the last U.S. election. But researchers find bot activity on other issues, too. And it can be difficult to narrow down where bots originate, in terms of where the people who put them to work live in the real world.
The oilpatch lobby group says it doesn't engage in bot activity.
It describes the role of Canada's Energy Citizens as a voice for members of Canada's oil and gas community. The Facebook page for Canada's Energy Citizens works like a newsfeed for its followers, now numbering more than 220,000, increasing online traffic for articles, videos and opinion pieces that promote the oil industry. Each post shared can get dozens, or even hundreds, of shares from this following.
Those numbers matter because the more likes and shares a post gets, the more likely it will show up in someone else's newsfeed.
In the end, oilpatch marketing tactics, plus the marketing tactics of environmental detractors, plus the activities of bots add up, drowning out some people's voices and making it difficult for the average person to evaluate facts.
It also isn't always easy to identify bots (or who is programming them, or why), though there are some clues.
"When you see profiles... tweeting 100, 200, 300, 400-plus times a day, we're fairly confident that we can argue that that's non-human behaviour. (Or) that's a lot of Red Bull and a lot of typing if it's a human being trying to create that much contact," said John Gray, CEO of the Vancouver-based firm Mentionmapp Analytics.
Researchers are still grappling with what effect automated, bot-like behaviour online can have on how people talk and think about big political problems.
Recently, someone who has been a contractor for Greenpeace admitted to deploying their own "bots" in an online poll on a news site, to prove that the other side was playing dirty. Interviewed by National Observer, he asked not to be identified, acknowledging that his tactics were also trying to game the system.
"It was a spontaneous morning's effort to show what a toy system the poll was using. I shared the knowledge with my Greenpeace friends after I'd looked at it because they also weren't aware that these polls didn't adequately stop multiple votes," he added in an April email, noting he was not working on a contract with the environmental group at the time. "I haven't looked at the polls since."
Greenpeace flagged the possibility of bots swinging the results when two Vancouver Sun polls asking readers about Kinder Morgan started getting thousands of votes each. The number of participants seemed outsized compared to how many people normally vote in such polls, and pro-pipeline votes dwarfed anti-pipeline votes, contrary to a series of scientific polls that showed people were split on the matter.
"The biggest concern to us is that if someone is deliberately trying to manufacture the appearance of consent for Kinder Morgan by manipulating polls... people aren’t getting an accurate sense of how people in the country feel," said Jesse Firempong, a Greenpeace Canada communications officer based in Vancouver.
"These false results, if they are using bots, could be waved around to decision makers and to investors as evidence of greater support and confidence for the project than there actually is.”
In a follow-up email in April, Firempong emphasized hers is not the only group concerned about bots, and, "Greenpeace has never run any bots on any of these (or other) polls, nor would we."
She also said Greenpeace hadn't spent money on promoting participation in newspapers' online polls, but, "if pro-pipeline groups and Greenpeace use (at times) promoted posts in general, I don't think this signifies much except that organizations around the world share some pretty standard online engagement tactics — and it highlights the difficulties of Facebook's algorithm."
In the case of the online poll promoted by the oilpatch lobby group, CAPP was directing its fanbase to a Vancouver Island newspaper's poll on the Kinder Morgan Trans Mountain pipeline expansion.
These activities are just examples of how the online conversation around the troubled oil pipeline is, increasingly, being taken to extremes by actors on both sides of the debate. While federal, British Columbia and Alberta politicians fight over the future of the $7.4-billion project, casual opinion polls online and social media platforms are open battlegrounds. In these fights, messages can be spread faster and the debate can become more polarized with sponsored ads and automated bots.
How it works
First, Facebook ads: when someone sponsors a post, they can reach a bigger audience than those who already follow them. In Canada, users can see lists of sponsored posts as part of a Facebook experiment to enhance transparency.
CAPP spokeswoman Chelsie Klassen described the energy citizens group as giving "supporters a voice to express their enthusiasm and enhance energy awareness in Canada." She declined to share the cost of urging people to vote in an Oak Bay News poll or explain why the post was sponsored.
"Canada’s Energy Citizens asked the following question: 'Do you support the Trans Mountain pipeline?' This question does not solicit a bias to vote a certain way," Klassen wrote in an email to National Observer. The ad was posted before noon eastern time on Fri. April 6, and was no longer on the citizens' group's list of purchased ads at 2:45 p.m. the same day.
The group paying to promote the poll didn't ring alarm bells for Facebook, which is instead focused on giving people tools to see where ads are coming from and ensuring that advertisers are authorized to run "political" or "issue" ads.
The Facebook ad was just one part of Canada's Energy Citizens' online efforts before and through last weekend to harness public opinion in polls published by some of Canada's largest and smallest newspapers. Only the Oak Bay News link appears to have been sponsored. But by Monday afternoon, polls run in the Toronto Star, Saanich News, Victoria News and Oak Bay News — all of which got a signal boost from the oil and gas group's Facebook page — showed overwhelming support for the pipeline.
'Something very peculiar'
Green Party leader Elizabeth May said she would "expect no less" than a sponsored post from CAPP.
"I know my community and I know there are people who support the pipeline, but they are a minority. I respect their views, but the overwhelming sentiment of southern Vancouver Island and the Lower Mainland is clear, and there's something very peculiar about some of the online polling," May, who represents the federal riding of Saanich-Gulf Islands, told National Observer.
"But of course it's not controlled... I don't know how these things get manipulated but they certainly aren't legitimate polls."
Whereas the recent online newspaper polls show a great deal of support for the Kinder Morgan Trans Mountain expansion, scientific polls indicate a range of more nuanced positions.
As examples, one Angus Reid online poll of 2,501 Canadians showed people are mostly split on the pipeline debate, while another, of 2,125 Canadians conducted in mid-April, showed 34 per cent of British Columbians feel environmental risks outweigh economic risks and 35 per cent feel the opposite. Angus Reid pollsters found about 30 per cent said the pipeline's risks and benefits are about equal. They also found that 54 per cent of British Columbians supported the pipeline expansion now, compared to 48 per cent in February. (Angus Reid estimated its margin of error to be plus-or-minus 3 percentage points.)
Meantime a Forum Research telephone survey of 1,061 eligible voters in B.C. showed 43 per cent of respondents believe the pipeline will be good for the economy, and 50 per cent believe it will be bad for the environment. And an online Abacus Data poll of 900 adult British Columbians found a wide expanse of wiggle room between those who absolutely oppose the pipeline (22 per cent) and those who absolutely support it (23 per cent).
These polls don't just show diverse opinions among British Columbians, they are transparent about how people are surveyed, how a sample of respondents compares to the wider population, and what the likelihood is of findings reflecting the positions of the broader public.
If a sponsored social media post aims to reach a lot of people, online "bots" can act like a lot of people.
On social media, automated bot accounts can be programmed to post or "like" messages en masse. Depending on their authors' intent, a simple bot strategy on Twitter might be to re-post messages of interest, making it look like there are many people who hold the same position, or actively working to bully and drown out contrary voices.
"These are hyper-partisan, highly partisan, profiles. There's just no question that their raison d'être is to push particular perspectives and points of view," said Gray, from the analytical firm Mentionmapp.
Such tactics have been used for multiple political campaigns around the world, including the recent scandals involving tech firms that used personal data to sway voters in the recent U.S. elections as well as the Brexit referendum in Europe.
Twitter is working on weeding out bots and recently made changes to "stay ahead of malicious activity targeting the crucial conversations taking place on Twitter — including elections in the United States and around the world."
But bots aren't a Twitter-only issue. Gray noted they can also be found at work on Pinterest, Instagram, dating apps and other online platforms.
They also aren't inherently bad: a company might build a chat bot into its website for customer service, for example. In that case, when you make a complaint online, a window might open and you'll be asked a series of standard follow-up questions before you are connected to a real person. The order of questions and pre-set responses are written by a person. If you say you want to cancel an order, the bot can be programmed to ask you why. Depending on what you say, the bot will provide you with additional information, or ask you more questions according to a script.
To Gray, who recently studied the presence of "non-human" participants in Twitter conversations about the Kinder Morgan pipeline expansion, bots can make it more difficult for real people to be heard in an increasingly polarized conversation (unless they are paying for advertising). On Twitter, he found bot activity — like accounts posting hundreds of times — on both sides of the pipeline debate.
That, he said, puts public opinion at risk of manipulation.
In the news
In December, someone who specializes in computer security research tried their own bot out on a Vancouver Sun poll and shared their results with Greenpeace.
Speaking to National Observer on condition of anonymity, the Canada-based researcher — who has worked on contracts for Greenpeace and other organizations — wrote a script that allowed him to vote "no" to a question about the pipeline numerous times, very quickly.
He said he was prompted to game the poll when he saw that "11,000 people had voted or something ridiculous before lunchtime."
As his automated anti-pipeline votes ticked up, pro-pipeline votes appeared to increase in response. In the end, more than 20,000 people weighed in (most on the side of the pipeline expansion going through).
Sun editor-in-chief Harold Munro noted that the company they use for their daily polls, Poll Daddy, tracks users' IP addresses and cookies to prevent people from voting more than once.
"Some polls do attract significantly more responses, particularly those on hot-button issues such as pipelines, whales in captivity, pit bulls and gun control. Special interest and advocacy groups on both sides of an issue often share a link to a poll with their members and like-minded groups, urging them to vote. This is when we see the biggest spikes," Munro wrote in an email response to National Observer questions.
"The poll is in no way scientific, nor do we claim that it is. Rather we use the daily poll each morning to encourage conversation about a news event or trending topic."
The Sun's daily polls often ask readers to respond to news of the day, and generally seem to net just a few hundred votes. For example, in a Nov. 28, 2017 survey, 596 people weighed in on news of Prince Harry’s engagement to Meghan Markle. In May of last year, when the Sun asked about a Vancouver Aquarium decision to ban new dolphins, whales and porpoises from captivity, nearly 14,000 votes were cast (about 46 per cent of respondents in favour of the ban, 54 per cent against)
The outliers that Greenpeace picked up on were published on Dec. 9, 2017 and Jan. 31, 2018. The first — “Do you agree with the NEB’s approval of Kinder Morgan’s request to bypass Burnaby laws to build a pipeline?” — received more than 25,000 votes. The second, “Do you agree with B.C.’s proposal to restrict shipments of diluted bitumen from Alberta?” netted almost 24,000 responses.
Munro did not respond to follow-up questions about whether his newspaper is worried about or looking into bot interference in its online polls.
Elisabeth Besson, another spokeswoman for the Canadian Association of Petroleum Producers, said the oil industry group has "never seen any indication of bot activity," nor has it made any use of automation to mobilize supporters.
Canada's Energy Citizens started to expand rapidly in 2014, when Canadian pipeline operators such as TransCanada and Enbridge were aggressively trying to promote pipeline projects such as Energy East, Keystone XL, and Northern Gateway. A leaked TransCanada strategy document at that time revealed details of a plan to attack the reputation of critics and build an army of online activists to promote the company's messaging on Energy East. Following the leak, TransCanada confirmed that it had solicited the strategy but said that it had never fully been deployed.
Yesterday, @NEBCanada ruled in favour of Kinder Morgan's efforts to bypass #Burnaby bylaws to build its Trans Mountain #pipeline. It's time to speak up - take this poll and make yourself heard >>https://t.co/WBdrFYRe2z #StopKM pic.twitter.com/0BqFPaCmNh— Greenpeace Canada (@GreenpeaceCA) December 8, 2017
Bot activity new normal, but impacts untested
At the University of Ottawa, the Vancouver Sun poll numbers are no surprise for Elizabeth Dubois, an assistant professor in the department of communication.
"Frankly I’m surprised that the little polls that show up online, particularly the ones on media outlets... don’t get manipulated more frequently, or that we don’t notice them being manipulated more frequently," Dubois said. “Polls have a really powerful role to play in how people decide who to vote for or what they think on an issue. And so it’s obviously politically powerful."
Dubois and Concordia University assistant professor Fenwick McKelvey recently analyzed the effects of Twitter bots in Canada's 2015 election. So far, they have found people writing scripts that put bots to work amplifying and silencing messages, but how influential this work is remains to be seen.
Looking toward the 2019 federal election, Dubois cautioned real people do write the programs that put bots to work.
“If you’re a political operative, are you going to prioritize engaging people and then go to bots, or are you going to start with bots and then engage people based on the now-inflated numbers in terms of support that you’ve got? It’s hard to say and we don’t have, at least to my knowledge, in Canada, any good understanding of that yet,” Dubois said.
Kennedy Stewart, the NDP MP for the British Columbia riding of Burnaby South and a professor of public policy on leave from Simon Fraser University, easily dismissed the newspaper polls, comparing them to a radio station that asks people if a hockey player should be removed from their team roster.
"The polls aren't scientific so I don't pay any attention to them at all. These are gimmicks," he said. "There's no way of knowing who these people are who answer, and all it takes is a pro-pipeline group that has a large list of emails to say, 'hey, ... click yes to this poll.'"
Last month, Stewart and May were both arrested alongside others blockading Kinder Morgan's pipeline terminal in Burnaby, B.C.
To date, nearly 200 people have been arrested at the site; last week a B.C. Supreme Court judge suggested that those arrested under civil charges for violating a court injunction should face criminal charges.
For Kennedy — who was interviewed for this story before the March protests began — the real battleground is offline. "You could fight it on Twitter, or social media, or these online polls... but really nothing gets political attention like on-the-ground activity."
— With files from Mike De Souza
Editor's note: The name of the firm Mentionmapp Analytics was corrected at 11:50 a.m. EST on April 20, 2018.