Policing Info World: conference review
On May 23, I attended a free, one-day conference on data, law enforcement, and surveillance: “Policing Info World”. It was hosted by the BC Freedom of Information and Privacy Association (BC FIPA), the BC Civil Liberties Association (BCCLA), and Kwantlen Polytechnic University (KPU).
There were four panels: one on surveillance, one on predictive and algorithmic policing, one on the ideal of bias-free policing, and one on searches for data when crossing the border.
On surveillance
The panel:
- Michelle Davey, Vancouver Police Department’s Superintendent in charge of Investigative Support Services,
- Wade Deisman, Associate Dean of Arts at KPU whose research interests include policing, law, technology, and security, and
- Josh Patterson, Executive Director of the BCCLA.
On this panel, Michelle Davey presented the VPD’s plans for a drone/uncrewed aerial vehicle. They are still developing the policy that will control its use, but she told us that it will primarily be used for emergency response and for gathering data to analyze an accident or crime that has already happened, especially vehicle collisions. She said that it would be also used during large public events to monitor unlawful gatherings of people (mentioning the 2011 Stanley Cup Riots as an example where this would have been useful). Each use of the drone will require high-level approval and they will document and publish each use of the drone.
Josh Patterson expressed support for most uses of the drone that Davey mentioned, with the exception of the potentially indiscriminate wide-area surveillance at public events. When people expect to be surveilled, they may be less likely to show up at an event they would have otherwise been present for, even when their presence would have been wholly peaceful and lawful.
I hope that that VPD’s drone policy will disclose the factors that they will use to determine whether a public event justifies the use of a drone for pre-emptive monitoring and what is done with the data in the case that no criminal investigations follow from the event.
The VPD also already operates a helicopter. While the use of a drone may bring certain privacy issues to the forefront and may subject more people more frequently to surveillance, it is not different in kind from that which is already possible.
On predictive policing
The panel:
- Ryan Prox, from VPD’s Crime Analytics Advisory & Development unit, and
- Mike Larsen, President of BC FIPA, and a criminology professor at KPU whose research interests include the security state and access to information.
Ryan Prox presented how the VPD uses predictive policing to inform where they should place officers. They use predictive policing only for the most patterned and predictable crime: property offenses. You can see the data that they use as input to the system at geodash.vpd.ca.
Of course, police presence itself would increase the observations of crime in a particular area, thus potentially creating a feedback cycle where the system would suggest deployments to areas where crime had been observed by officers, because that’s where they had been deployed in the past. To avoid this, the system does not include property crime that was reported by an officer; it only bases its predictions on property crime that is reported by the property owner.
The system also isn’t used for predictive deployments to the downtown eastside or in areas around temporary modular housing. These areas already have their own community-policing programs and officers that are specially trained to work in these areas and with these communities.
There are risks inherent in off-loading our decision-making to a mathematical/algorithmic process. It seemed like the VPD is aware of this and approaching the issue thoughtfully. Prox pointed to the Federal Treasury Board’s Algorithmic Impact Assessment which, while not legally applicable outside of the federal government, gives a useful framework for managing the human risks of algorithmic decision-making in general. For example, if an automated decision will “likely have very high impacts on […] the rights of individuals or communities”, then the government must (among other things) obtain peer review by non-government, publish descriptions of the system and training data, require the final decision to be made by a human, and conduct ongoing monitoring to detect unintentional outcomes.
On the ideal of bias-free policing
The panel:
- Dylan Mazur, community lawyer with the BCCLA, and
- Michelle Cameron, former analyist and then Senior Investigator with the Civilian Review and Complaints Commission for the RCMP, now with UBC as an advisor/investigator for investigations of sexual misconduct on campus.
Dylan Mazur discussed street checks and their place on the spectrum of political violence as a means to disorganize social movements. (In June 2018, the Union of British Columbia Indian Chiefs and the BCCLA filed a complaint with the Office of the Police Complaint Commissioner regarding the over-representation of Indigenous and Black people in street checks conducted by the VPD.)
Michelle Cameron talked about her experience as an analyst and investigator and discussed the need for good data in order to detect expressions of bias and monitor improvement (or lack thereof).
The panel talked about the difficulty in obtaining accurate race-based data in the context of a street check. The categories that people slot themselves into on a census don’t match how they might identify themselves to an officer after an adversarial, power-imbalanced encounter.
This need for data is consistent with many of the Calls to Action from the Truth and Reconciliation Commission of Canada which call for data on criminal victimization and to “monitor and evaluate progress” in eliminating over-representation of Indigenous people in custody, for example.
On searches for data when crossing the border
The panel:
- Peter Edelmann, an immigration lawyer, and
- Meghan McDermott, BCCLA’s policy staff counsel.
Peter Edelmann presented the state of the law on data searches when coming into Canada. The power for these searches comes from customs and immigration statutes that give the Canadian Border Services Agency (CBSA) heightened ability to search for material that might reveal customs- or immigration-related offences.
It is clear that CBSA requires no reasonable grounds (as would normally be required for a search) to search your goods at the border. And “goods, for greater certainty, includes conveyances, animals and any document in any form.”
Less clear is how the government is to treat the data that is on your phone. Is it a ‘good’? All we have today are some lower court decisions that say the data is a ‘good’ for the purposes of the Customs Act.
Edelmann noted that Canada doesn’t have a consistent approach to data that crosses the border. Data can cross the border over the internet and data can cross the border on a cell phone. The data carried over the border on a cell phone is minuscule compared to the data crossing over the internet. But the government treats these two streams very differently. He argued that the government needs to clarify with legislation how data should be regulated and then do that uniformly, predictably, and transparently, rather than shoehorning our new electronic devices into old categories of searches as a way of getting at only a tiny fraction of data that is crossing the border.
Meghan McDermott presented portions of the Electronic Devices Privacy Handbook: A Guide to Your Rights at the Border, published by the BCCLA and the Canadian Internet Policy and Public Interest Clinic (CIPPIC) in 2018.
The main takeaway for me was the lack of transparency from the CBSA about their border-search practices, especially regarding the National Targetting Program. What might increase the likelihood that your electronic device is searched at the border? It is impossible to know, but court cases and access-to-information requests provide some hints. The BCCLA/CIPPIC handbook identifies a list of potential factors, including having “anime and manga” or being “a single man travelling alone.”
Connection with the history of government searches
I’m reading a book right now on the history of protections against search and seizure in England and the US. A theme that comes up time and time again is how popular opposition to government searches did not crystallize until the practice became oppressive to the right group of people. The law didn’t suddenly give the government more search powers between 1580 and 1642, but a new opposition to government searches became apparent during that time. What changed? Instead of the government searching only the lower classes and “vagrants in particular”, searches started to be “the frequent experience of the many.” And the Kings were using these new tactics as tools for disorganizing social structures and political opposition. The following are some quotes from the book (emphasis mine).
[T]he reasons that the Tudors sponsored the new general privy search were as much political as social. Henry VII and his successors wanted to rid England not only of its human eyesores but of the “seditions and rebellions” that vagrancy bred. Making the houses of their subjects less secure against searches was the Tudors’ method of making their throne more secure.
The reasons that Englishmen waited nearly a century after the general search became routine to condemn it arose from the compartmentalized sequence in which they realized its impact and the social values by which they gauged that impact. Although privy searches were general, they aimed at lowly, illiterate beggars and vagrants. Articulate Englishmen, therefore, did not experience those searches or sympathize with those who did, especially because most privy searches occurred in barns, lodging houses, and inns rather than in private dwellings. Social attitudes toward vagrants as an unworthy underclass delayed resentment of such searches until other more esteemed members of society were subjected to them. […]
Mass protests against the mechanism of search and seizure started in the 1580s because of further changes in it and in the identity of the types of persons that it affected. Having demonstrated its efficacy as an antivagrancy device, the general search was an obvious candidate for extension to the further purpose of crushing religious and political dissent, and there too, it produced spectacular results against Jesuits and Protestant radicals alike.
The present-day counterpart [of vagrancy searches] might be weekly, nocturnal inspections of all liquor stores, rooming houses, and video-game parlors for drunks, drug-addicts, derelicts, and runaway adolescents.
(I would add street-checks to this list.)
The nobility and dissenting clerics, the articulate and educated groups most capable of protesting violent searches began to experience them routinely for the first time in the last two decades of the sixteenth century.
It’s important to understand what the law allows our government to do, regardless of whether actual practice is currently constrained by policy, practicality, or discretion.
Notes
1. ↑ Customs Act, RSC 1985, c 1 (2nd Supp), s 99(1)(a).
2. ↑ Ibid, s 2(1).
3. ↑ See e.g. R v Gibson, 2017 BCPC 237 at paras 95–98. See also R v Moroz, 2012 ONSC 5642 at paras 20–21.
4. ↑ William J Cuddihy, The Fourth Amendment: Origins and Original Meaning 602–1793 (Oxford University Press, 2009).
5. ↑ Ibid at 94.
6. ↑ Ibid at 95–98.