On November 23, 2018, Waterfront Toronto hosted a Civic Labs workshop in Toronto. The theme of the workshop was Smart City Data Governance. I was asked to give a 10 minute presentation on the topic. What follows is a transcript of my remarks.
Smart city governance relates to how smart cities govern themselves and their processes; how they engage citizens and how they are transparent and accountable to them. Too often the term “smart city” is reduced to an emphasis on technology and on technological solutionism – in other words “smart cities” are presented as a way in which to use technology to solve urban problems. In its report on Open Smart Cities, Open North observes that “even when driven in Canada by good intentions and best practices in terms of digital strategies, . . . [the smart city] remains a form of innovation and efficient driven technological solutionism that is not necessarily integrated with urban plans, with little or no public engagement and little to no relation to contemporary open data, open source, open science or open government practices”.
Smart cities governance puts the emphasis on the “city” rather than the “smart” component, focusing attention on how decisions are made and how the public is engaged. Open North’s definition of the Open Smart City is in fact a normative statement about digital urban governance:
An Open Smart City is where residents, civil society, academics, and the private sector collaborate with public officials to mobilize data and technologies when warranted in an ethical, accountable and transparent way to govern the city as a fair, viable and liveable commons and balance economic development, social progress and environmental responsibility.
This definition identifies the city government as playing a central role, with engagement from a range of different actors, and with particular economic, social and environmental goals in mind. This definition of a smart city involves governance in a very basic and central way – stakeholders are broadly defined and they are engaged not just in setting limits on smart cities technology, but in deciding what technologies to adopt and deploy and for what purposes.
There are abundant interesting international models of smart city governance – many of them arise in the context of specific projects often of a relatively modest scale. Many involve attempts to find ways to include city residents in both identifying and solving problems, and the use of technology is relevant both to this engagement and to finding solutions.
The Sidewalk Toronto project is somewhat different since this is not a City of Toronto smart city initiative. Rather, it is the tri-governmental entity Waterfront Toronto that has been given the lead governance role. This has proved challenging since while Waterfront Toronto has a public-oriented mandate, it is not a democratically elected body, and its core mission is to oversee the transformation of specific brownfield lands into viable communities. This is important to keep in mind in thinking about governance issues. Waterfront Toronto has had to build public engagement into its governance framework in ways that are different from a municipal government. The participation of federal and provincial privacy commissioners, and representatives from federal and provincial governments feed into governance as does the DSAP and there has been public outreach. There will also be review of and consultation of the Master Innovation Development Plan (MIDP) once it is publicly released. But it is a different model from city government and this may set it apart in important ways from other smart cities initiatives in Canada and around the world.
Setting aside for a moment the smart cities governance issue, let’s discuss data governance. The two are related – especially with respect to the issue of what data is collected in the smart city and for what purposes.
Broadly speaking, data governance goes to the question of how data will be stewarded (and by whom) and for what purposes. Data governance is about managing data. As such, it is not a new concept. Data governance is a practice that is current in both private and public sector contexts. Most commonly it takes place within a single organization which develops practices and protocols to manage its existing and future data. Governance issues include considering who is responsible for the data, who is entitled to set the rules for access to and reuse of it, how those rules will be set, and who will profit/benefit from the data and on what terms. It also includes addressing issues such as data security, standards, interoperability, and localization. Where the data include personal information, compliance with privacy laws is an aspect of data governance. But governance is not limited to compliance – for example, an organization may adopt higher standards than those required by privacy law, or may develop novel approaches to managing and protecting personal information.
There are many different data governance models. Some (particularly in the public sector) are shaped by legislation, regulations and government policies. Others may be structured by internal policies, standards, industry practice, and private law instruments such as contracts or trusts. As the term is commonly used, data governance does not necessarily implicate citizen involvement or participation in the same way as “smart city governance” does – it is the “city” part of “smart city governance” that brings in to focus democratic principles of transparency, accountability, engagement and so on. However, where there is a public sector dimension to the collection or control of data, then public sector laws, including those relating to transparency and accountability, may apply.
With the rise of the data economy, data sharing is becoming an important activity for both public and private sector actors. As a result, new models of data governance are needed to facilitate data sharing. There are many different benefits that flow from data sharing. It may be carried out for financial gain, or it may be done to foster innovation, enable new insights, stimulate the economy, increase transparency, solve thorny problems, and so on. There are also different possible beneficiaries. Data may be shared amongst a group of entities each of which will find advantages in the mutual pooling of their data resources. Or it may be shared broadly in the hope of generating new data-based solutions to existing problems. In some cases, data sharing has a profit motive. The diversity of actors, beneficiaries, and motivations, makes it necessary to find multiple, diverse and flexible frameworks and principles to guide data sharing arrangements.
Open government data regimes are an important example of a data governance model for data sharing. Many governments have decided that opening government data is a significant public policy goal, and have done tremendous amount of work to create the infrastructure not just for sharing data, but for doing it in a useful, accessible and appropriate manner. This means the development of standards for data and metadata, and the development of portals and search functions. It has meant paying attention to issues of interoperability. It has also required governments to consider how best to protect privacy and confidential information, or information that might impact on security issues. Once open, the sharing frameworks are relatively straightforward -- open data portals typically offer data to anyone, with no registration requirement, under a simple open licence.
Governments are not the only ones developing open data portals – research institutions are increasingly searching for ways in which to publicly share research outputs including publications and data. Some research data infrastructures support sharing, but not necessarily on fully open terms – this requires another level of consideration as to the policy reasons for limiting access, how to limit access effectively, and how to set and ensure respect for appropriate limits on reuse.
The concept of a data trust has also received considerable attention as a means of data sharing. The term data trust is now so widely and freely used that it does not have a precise meaning. In its publication “What is a Data Trust”, the ODI identifies at least 5 different concepts of a data trust, and they provide examples of each:
· A data trust as a repeatable framework of terms and mechanisms.
· A data trust as a mutual organisation.
· A data trust as a legal structure.
· A data trust as a store of data.
· A data trust as public oversight of data access.The diversity of “data trusts” means that there are a growing number of models to study and consider. However, it also makes it a little dangerous to talk about “data trust” as if it has a precise meaning. With data trusts, the devil is very much in the details. If Sidewalk Labs is to propose a ‘data trust’ for the management of data gathered in the Sidewalk Toronto development, then it will be important to probe into exactly what the term means in this context.
What Sidewalk Labs is proposing is a particular vision of a data trust as a data governance model for data sharing in a smart cities development. It is admittedly a work in progress. It has some fairly particular characteristics. For example, not only is it a framework to set the parameters for sharing the subset “urban data” (defined by Sidewalk Labs) collected through the project, it also contemplates providing governance for any proposals by third parties who might want to engage in the collection of new kinds, categories or volumes of data.
In thinking about the proposed ‘trust’, some questions I would suggest considering are:
1) What is the relationship between the proposed trust and the vision for smart city governance? In other words, to what extent is the public and/or are public sector decision-makers engaged in determining what data will be governed by the trust, on what terms, for whose benefit, and on what terms will sharing take place?
2) A data governance model does not make up for a robust smart city governance up front (in identifying the problems to be solved, the data to be collected to solve them, etc.). If this piece is missing, then discussion of the trust may involve discussing the governance of data where there is no group consensus or input as to its collection. How should this be done (if at all)?
3) A data governance model can be created for the data of a single entity (e.g. an open government portal, or a data governance framework for a corporation); but it can also be developed to facilitate data sharing between entities, or even between a group of entities and a broader public. So an important question in the ST context is what model is this? Is this Sidewalk Labs data that is being shared? Or is it Waterfront’s? Or the City’s? Who has custody/control or ownership of the data that will be governed by the ‘trust’?
4) Data governance is crucial with respect to all data held by an entity. Not all data collected through the Sidewalk Toronto project will fall within Sidewalk’s definition of “urban data” (for which the ‘trust’ is proposed). If the data governance model under consideration only deals with a subset of data, then there must be some form of data governance for the larger set. What is it? And who determines its parameters?