I'd like to begin by acknowledging that the land on which we gather is the traditional unceded territory of the Algonquin Anishinaabe people. Further, I'm on parental leave now, and I would like to thank my family for giving me the time to speak here today.
I hope my comments will be relevant to the committee and provide evidence to support its preliminary recommendations, which I largely support as well. I appreciate its willingness and dedication to keep pulling a lone thread that unravels this tangled web of data, surveillance, campaigning and advertising. These issues have been a great preoccupation for me, bringing together previously separate research into Internet policy, digital political communication, and algorithmic governance.
I would like to focus my comments on three areas of investigation before the committee today. In many ways, they complement some of the findings and conclusions of Taylor Owen, such as the focus on online advertising, third party data brokers and analytics, and finally political parties. My comments highlight my concerns and potential policy remedies to these issues based on my own research. I hope the committee will also look to new ways to support more research in these areas, giving researchers better access to data under clear ethical guidelines.
First, online advertising is more than a political problem. The Cambridge Analytica Facebook scandal has exposed more than anything the public's unawareness, resignation or willed ignorance about the sophistication of online advertising. It might not tip the next election, but reforms to the sector will go a long way toward restoring public trust in the Internet writ large—speaking to the structural issues of the presenter before me.
Online advertising means a few things today. It concerns programmatic banner advertisements of the kind we see around every website. These ads account for a $12-billion industry in Canada, according to the Media Concentration Research Project, and Google and Facebook account for three-quarters of the revenue. However, there are new types of advertising. Native advertising, or sponsored content, confuses the line between advertising and advertorial. With influence marketing, informal brand ambassadors fill our social media feeds with their often unacknowledged endorsements. There's also spam and bot activity.
In general, I question the public benefit of all these forms of targeted advertising. In my mind, we have too little accountability and too much in the belief of data and targeting. New kinds of advertisers will present problems for political campaigns. Political campaigns may turn to these grey markets, using influencers or “for rent” social media accounts to fake grassroots support. We must recognize the extent of this promotional content in our culture, make steps to be able to qualify it, and also work to ensure proper disclosure and fair play for these third party advertisers. One tangible step might be to work with Elections Canada to clarify the placement cost criteria to ensure that the new types of advertising count in electoral spending.
In regard to programmatic advertising, we need to consider what are appropriate limits to data collection and targeting. There's evidence in the political literature that the multitude of microtargeting does not necessarily help campaigns better engage with voters. In my opinion, the current situation overstates the value of targeting data, omitting the potential harms in over-collection. We can name a few risks of over-collection. Conceivably, we can think about advertising profiles being used as a proxy for protected categories like race and political belief. Targeted advertising is increasingly used to justify growing online and offline surveillance. Finally, all this data can be leaked or improperly handled, as we've seen time and time again.
Data protection is an important remedy. By limiting what can be collected and used for targeting, we can diminish the race to monetize more personal information for advertising. Without change, I fear a time when large social media companies compete against Internet service providers on how much data they can collect and turn into targeted advertising portfolios, collecting as much data as they can.
AggregateIQ is part of a global technology industry. Canada, like many other western democracies, has witnessed political parties go digital to better run their campaigns. Many companies now sell services to help parties manage, analyze and use their data to, among other things, buy ads and gauge support.
For its proponents, technology-intensive campaigning gets out the vote. It also helps parties find the right supporters, be more responsible with their limited funds, and ultimately win. I do not dispute these claims, but it has become clear to me that the global scope of the industry today creates new regulatory challenges, particularly in ensuring that offshoring data analytics or digital services does not evade national spending or national privacy law.
These industries warrant greater scrutiny, particularly in how they move data across borders. Offshoring data analytics should not evade privacy laws. International companies should be mindful of how they transport models, particularly models using machine learning algorithms that might have been collected and developed using loose privacy laws, and make sure they do not find their way abroad.
I believe these issues can be addressed by adding enforcement powers to the office of the Privacy Commissioner and continuing to support its multi-jurisdictional enforcement.
Third and finally, with regard to political parties, I was not surprised that AggregateIQ has little uptake in Canada. This is not because there is an aversion to technology in politics but because parties already have their own solutions in place. The Conservative Party uses NationBuilder, together with its proprietary database. The Liberal Party uses the U.S. Democratic Party-affiliated NGP VAN. The NDP works with other Democratic-affiliated firms, Blue State Digital, and its own tool, Populus. I have to admit I'm surprised that no representatives from these political parties or from these companies have appeared before these committees investigating these matters of political data.
In general, political parties have much to do to be more accountable about their data habits. Again, in my research I've been impressed by the professionalism of campaigners on all sides, and I believe these professionals will ultimately embrace these new rules. I understand reluctance, too, to impose more regulation on already taxed organizations, but greater accountability for digital campaigning should benefit all parties.
I support the committee's recommendation for privacy laws to apply to political parties. I'd like to add one other reason.
In my own research I've found that lax rules have created real challenges for political campaigns. Data is a strategic resource for parties. Lax rules, however, translate into real inequities. Incumbent parties have better access to data than new entrants. To compete, all parties have to be constantly maintaining their lists and collecting more data, since they cannot rely on the data collected by Elections Canada. This leads to an overall concentration in the central party, which often becomes the database, and an overall logic of permanent campaigning.
Parties might be reluctant to adopt privacy law, given the importance of digital fundraising. If we believe that parties should collect less data, then we may want to consider reinstating the per-vote subsidy that diminished the need for funding and its associated data collection.
Also in terms of data, most parties use some form of predictive analytics to examine the political data they have collected and make predictions about voter behaviour. Either the party or, more often, the consultant analyzes the data to calculate the probability that each voter will support the party and the probability that a voter will be persuaded to vote for the party. The parties use these to make important decisions, like who to target and who to encourage to vote. Predictive analytics exacerbates low voter turnout in Canada, allowing parties to continue to distance many voters from the electoral process. Parties should agree to audit their scoring of voters, and other analytics, for potential race or gender biases. As well, they should also make sure that these decisions about which voters to contact and which voters to ignore are auditable and explainable.
Finally, my suggestions about reform to digital campaigning are my own experiences alone.
Political parties ultimately need to work together on the rules of the game. Codes of conduct have long been recommended to improve Canadian politics. I believe that now is the time to move toward the drafting of a code. In many ways, when we're trying to deal with these consequences of foreign interference, we can only begin to look to ourselves as a first step in rectifying those potential threats. However, parties need to be able to take the first step.
I commend the committee for continuing this project and I hope my comments help support the recommendations for online advertising, data protections for political technology firms, and reforms to privacy and the activities of political parties.
Thank you very much.