Access to information research in the digital era

Abstract

In the four decades since Canada's Access to Information Act (1983) came into force, the massive proliferation of digital technologies has prompted significant transformations in the operations of modern governments. The impacts of this digital era on access to information (ATI) research in Canada—by which we mean both research on ATI law (and its administration) as well as research using ATI law (to generate data)—have not received the scholarly attention they deserve. In this short article, we focus on describing the pervasive digital environment that Canada's ATI system operates in and what this means for future ATI research and practice. We begin by tracing how the digitization of government has caused shifts in the administration of Canada's ATI regime before reflecting on some of the implications of these changes for the current and future state of ATI research. A core point of contention is that the future of ATI in Canada and elsewhere is productively understood through the lens of data politics—the notion that all decisions to collect, share, or use data are intractably political (Bigo et al., 2019). We conclude by proposing several research questions that can help guide future ATI research in the era of digital government. The production and control of information is tantamount to the everyday work of governance (Pettigrew, 1972). In the era of “digital government” (Clarke et al., 2017; Lindquist, 2022), it is increasingly difficult if not impossible to think of a single governance function that is not digitized in some way. As many early proponents of “e-government” argued (Silcock, 2001), the adoption of digital technologies in the public sector can enable more effective, democratic, and participatory bureaucratic processes. At the same time, trends in digital government “challenge traditional notions of administration, management, organization, accountability, and engagement” in ways not yet fully examined and understood by academic researchers (Gil-Garcia et al., 2018: 633). The digitization of Canadian governance is reshaping the use and administration of federal ATI law in at least two ways. First, the information that governments produce, both for purposes of internal and external communication, is increasingly “born-digital” (DeLuca, 2020: 5). Regarding ATI, this has implications for how government information is preserved, retrieved, and processed for disclosure. To reduce burdens to the ATI system, it is now more feasible to release large volumes of information through proactive disclosure. The Trudeau government's most recent Access to Information Act reform bill, which achieved royal assent in 2018, introduced into Canadian law “the principle of ‘open by default’ in the digital age by making key information available proactively, without the need to make a request” (Canada, 2019; Duncan et al., 2023). A growing digital repository of information, under the umbrella of “Open Government” (Canada, 2022), is now publicly available on the Internet. But if digitization is expanding the amount of information accessible to citizens through proactive disclosures and by request, this is not true of all forms of information. Digitization makes it easier than ever to efficiently destroy government records. Take the Government of Canada's “Completed Access to Information Requests” portal as an example. This system, one of the most praised ATI modernization efforts enacted by the former Harper administration, allows users to informally request copies of the results of ATI requests completed by others. But only up to a point. The information that can be requested through the portal is limited to requests processed within the past two years. In most federal agencies, ATI disclosure packages older than two years can be destroyed in line with the minimum retention period recommended by Library and Archives Canada's Generic Valuation Tool (Canada, n.d.). Whereas the availability of affordable digital storage and computing power has made it feasible to collect and release more information than ever, this has not precluded policies of information control through deletion. The increasingly digital format of government information also affects how those on the receiving end of an ATI disclosure make sense of the information they obtain and what they choose to do with it. Many ATI users have a high level of digital literacy, a competency the Government of Canada actively promotes through programming and strategic messaging (Shepherd & Henderson, 2019). Even where users lack advanced technical skills, off-the-shelf digital tools are increasingly accessible to non-technical users. One example is Google's Pinpoint tool for investigative journalists, which uses artificial intelligence to help users quickly sort and search through large collections of text. Digital tools like Pinpoint allow ATI users to handle much larger amounts of information pursuant to their request than they could in the past. This is partly reflected in official statistics: between 2015–2018, the amount of information released under federal ATI law increased by 260% (Canada, 2019). There is also the matter of how widely the information is shared once it is released. The fact that information is often released in a digital format makes it easy for those on the receiving end to share it with others using platforms like Archive.org or Dataverse. Second, digital government is changing how—and by who—ATI disclosure is managed and controlled. Federal employees use proprietary software to track and manage information about the disclosure process (AccessPro Case Management is the most commonly used in the federal government). These private software solutions shape and mediate the disclosure process and outcome in ways deserving of future study (Stratton & Carter, 2023). In a digital-first organization, it is becoming less common for ATI officers to retrieve information from a physical archive. Instead, ATI officers—or other federal employees on their behalf—commonly identify, locate, and retrieve information by conducting broad and targeted searches from a computer. This new digital workflow is faster, cheaper, and more efficient. It is also more likely to turn up larger volumes of information with potential relevance to a request. Government information is increasingly acquired, produced, stored, and analyzed by private contractors (Luscombe et al., 2022). Private entities, even when they enter into partnership with the Government of Canada, are exempt from the Access to Information Act, putting a great deal of information which otherwise would have been subject to disclosure laws beyond the reach of civil society. This can also work in the other direction: when private data is purchased from data brokers by the Government of Canada (Boutilier, 2022), this information enters the jurisdiction of the Access to Information Act. ATI officers in Canada have long employed a range of informal strategies to direct, minimize, deter, block, or otherwise prevent meaningful disclosure from taking place (Larsen & Walby, 2012; Luscombe et al., 2017; Piché & Walby, 2021; Roziere & Walby, 2020). These have included using exorbitant fee estimates and lengthy estimated processing times as a deterrent, providing large quantities of information in hardcopy or digital, non-machine-readable formats (e.g., printed scans of digital spreadsheets), and simply failing to offer reasonable assistance throughout the request process. Reforms to the Access to Information Act in 2016 required federal agencies to waive all fees associated with processing an ATI request except for the initial $5 submission charge. Federal agencies were also directed to grant the requester decision-making power over the format of the final disclosure. Those filing requests under federal ATI law can now ask for the information to be provided in a digital, machine-readable format. Claims about the laborious nature of sifting through bankers' boxes in distant off-site archive facilities and other informal tactics of information control are less convincing to ATI users than they once were. This changes not only how ATI officers engage in information control, but also the kinds of “access brokering” (Larsen & Walby, 2012) tactics used by requesters to navigate and overcome the barriers that ATI officers may attempt to create. The digitization and consequent datafication of governance is creating new opportunities and challenges for ATI users. By datafication, we mean the process of creating value by taking “information about all things under the sun… and transforming it into a data format to make it quantified” (Mayer-Schönberger & Cukier, 2013: 42). Artificial intelligence and data-driven decision-making in public administration have taken hold (Auld et al., 2022). Datafication presents many new avenues for public administration scholars to explore including assessing the effects of automation or advancing understandings of algorithmic fairness and accountability (Ávila et al., 2020; Eubanks, 2017). People are increasingly aware of how different aspects of their lives are datafied with implications for how they are governed (Aradau & Blanke, 2022). While access to digital tools and training remain pressing equity issues in Canada (Haight et al., 2014), the tools and skills required to access, manage, and analyze large data sets are becoming more accessible. Governments are unable to take advantage of the opportunities presented by new digital technologies without generating masses of data. The explosion of data within public administrations and the democratization of data analysis are pushing ATI use beyond the search for a “‘smoking gun’ document” (Walby & Larsen, 2011: 524). A small but growing number of researchers, activists, and civil society organizations are beginning to use federal ATI and regional freedom of information laws to obtain quantitative digital data to uncover more systemic institutional patterns and challenges (Browne, 2022; McClelland, 2022; Mummolo, 2018; Owusu-Bempah & Luscombe, 2021; Piché et al., 2022; Reeves-Latour & Morselli, 2017). For example, investigative journalist Tom Cardoso has spent years negotiating access to large datasets on prisoner risk assessment instruments and analyzing them to uncover systematic biases against Black and Indigenous people (e.g., Cardoso, 2020). The datasets Cardoso generates through ATI and shares online are often subsequently analyzed by academic researchers that benefit from access to data they might not have otherwise obtained (see e.g., O'Connell & Laniyonu, 2023 for a scholarly analysis of a federal risk assessments dataset Cardoso acquired under ATI and published online). Researchers have also begun applying computational techniques to render documents searchable and extract valuable insights. In Canada, ATI records are often released in non-searchable digital formats. The growing accessibility of free and open source optical character recognition algorithms has made it easier to search through releases by making image format text searchable. The capacity to convert image format records into machine readable texts presents opportunities to analyze government releases at scale using increasingly popular “text-as-data” (Grimmer et al., 2022) techniques that rely on machine learning. Although many public administration scholars do not yet possess training in text-as-data techniques and research designs, there are growing numbers that do as these methods become popularized in disciplines like sociology and political science. For example, Berliner and colleagues (Berliner et al., 2021; Berliner et al., 2018) use an unsupervised machine learning approach called topic modeling to analyze over 1 million requests for official information in Mexico. They find that system users from regions that are more pro-government receive higher quality responses to information requests. Although currently few in number, we anticipate that more public administration scholars will develop competence with such methods as they become more mainstream in social sciences training at the graduate and undergraduate levels. The goals of open scholarship—“to improve openness, integrity, social justice, diversity, equity, inclusivity and accessibility in all areas of scholarly activities” (Parsons et al., 2022)—are changing how many social scientists, journalists, and other researchers use ATI and what they do with the disclosure packages they receive. Compared to a decade ago, it has become much more common for both the government and requesters to proactively share material. The emergence of “open by default” proactive disclosures is evidence of how open scholarship and open government norms are emerging concurrently. Parallel to the government's own proactive disclosure initiatives, many ATI users are actively sharing the results of their requests with others using a variety of file sharing platforms. Requesters are making their disclosures public because they believe that making information publicly available is good for science, for democracy, and for civil society. Depending on the project and the data, sharing the results of an ATI request may not only be viewed as a good thing to do but a necessity. When conducting a quantitative analysis of data obtained under ATI, sharing the “raw” data is an important part of fostering trust and enabling replication (Freese & Peterson, 2017). Alongside the publication of quantitative datasets, others have created publicly available digital document repositories. In collaboration with Kevin Walby, Director of the Centre for Access to Information and Justice at the University of Winnipeg, we recently launched a dashboard that keeps a running log of previously completed ATI requests and provides access to over 1000 disclosure packages that users can either download or explore using Google Pinpoint and the Internet Archive (Duncan et al., 2023). Canada Declassified (Sayle & Wiseman, 2022) is a digital repository run by historians at the University of Toronto composed of Canadian Cold War-related records declassified under the Access to Information Act. Civic tech advocacy is similarly driving innovation in government transparency. Volunteer “data activists” and nonprofit organizations are filling in gaps where governments are unable or unwilling to provide services that facilitate citizens' demand for information (Milan & Van der Velden, 2016). For example, Alivateli is an open-source platform created by the UK-based not-for-profit, mySociety. The platform serves as an intermediary between people and governments by facilitating requests for information. The platform maintains a public repository of previous requests made through the platform with all communications between requesters and the government that anyone can search and download. Alivateli's platform first launched in the UK as WhatDoTheyKnow.com and has since been adopted in over 25 jurisdictions. Data activists around the globe are creating public interest datasets (often using maps) designed to provide access to information that is otherwise not available. In the US, the advocacy organization Data for Black Lives aggregated state-level public health data to publish data on race-based disparities in COVID-19 cases and deaths (Data for Black Lives, 2022). Mexican engineer and activist, María Salguero systematically tracked and mapped media coverage of femicides (murders targeting women and girls) in Mexico because government figures, as demonstrated by her project, vastly undercount this type of violence against women (Sim, 2018). In Canada, the Policing the Pandemic Project used a similar “counter-mapping” (Kidd, 2019) technique relying on media coverage and information requests to track instances of police enforcement of COVID-19 mandates and restrictions as Canadian police were not proactively releasing this information (Luscombe & McClelland, 2020). These activist-generated datasets are created for different reasons, are born out of varying social and political contexts, and are not always accompanied by a level of methodological transparency that would allow one to properly validate them. This is not to say that such datasets should not be used for academic research, but that they should be approached critically with an understanding of what motivated their creation and investigation into how they were assembled. As part of any research project using data generated through ATI, researchers must always carefully reflect on what the data can and cannot tell them about the work of government while always keeping in mind that the very creation and disclosure of data itself can be and often is leveraged by government officials as a strategic means of securing legitimacy (Rappert, 2012). The values of openness and accountability that motivate data activism should be understood as models for open government rather than replacements for government intervention. An open-source mentality has many strengths. As demonstrated by successful software projects like the Linux operating system or the R programming language, open-source initiatives encourage the formation of active communities. Projects that attract diverse and engaged stakeholders often produce more robust and flexible systems than their proprietary counterparts (Weber, 2004). While often developed for a niche community of users (like government transparency wonks), such projects are frequently sources of innovation that have effects far beyond the core community of project stakeholders. Nonetheless, relying purely on data activists' altruistic endeavors is both insufficient and unsustainable in the long run. While the examples we draw upon in this article are full of promising ideas, relatively few are as effective as they could be and many fizzle out as their creators and the communities around them turn to other pursuits. Data activism initiatives highlight gaps in official information. They should be understood not merely as supplementary to open government initiatives but reflective of information that governments could be making available to citizens using government resources. What is the role of scholars in engaging with ATI as a form of data politics? How might academics balance the goals of producing rigorous knowledge about ATI while promoting government policies that enable high impact, high quality public administration research through access to data? How and to what extent is the datafication of governance creating new avenues to study how governments work? In what ways is datafication foreclosing or reshaping the traditional concerns of public administration scholarship? In what ways might norms of open scholarship and open government enable more robust forms of collaboration between social scientists and the government institutions that they study? How and to what extent are governments actively learning from civic tech and data activism initiatives? How does or how could civic tech advocacy translate into core government programs, services, or policies supported by public rather than private resources? What new research questions can access to government databases open up for public administration scholars? What challenges and opportunities currently exist for researchers to obtain access to large data sets or databases from federal institutions under ATI law? What kinds of established and innovative (e.g., text-as-data) methodologies and skill sets are required of researchers who hope to maximize the value of data accessed through ATI? The digitization of government and broader datafication of society have shifted how ATI is understood both by governments and their stakeholders. The norms and practices of open scholarship are frequently complementary to those of civic technologists and data activists, giving common cause to calls for more effective forms of open government to enrich both scholarship and advocacy. ATI implicates researchers in the conduct of data politics as brokering access is often a fraught process of negotiation. Even for public administration scholars who would wish to distance themselves from claims that their work is politically motivated, we contend that ATI scholarship is inseparable from the datafication of governance. Thus, even ATI research aspiring to political neutrality is embedded in the broader context of data politics. It is these transformations that we suggest require a conceptual shift toward data politics in ATI research and practice.