The technical architecture defines how a civic digital trust works. A legal agreement is only as good as the trust's ability to allow the secure collection, storage and sharing of data, and to have oversight and insight into the algorithms that transform the data into real world uses.
The most significant choice of technical architecture is how centralized or decentralized the digital assets are held across the network. In this section we summarize five distinct architectures: centralized, semi-centralized, decentralized, open data, data marketplace, and data sharing agreements. This is meant to illustrate the variety of options, rather than be an exhaustive list.
Different technical architectures make sense for different purposes. In addition, different technical architectures will solve for different legal requirements, particularly where stakeholders who may contribute or use data are under different legal requirements. Selecting the best architecture requires clarity and precision about the specific purpose of the trust. For example, to achieve the purpose of enhancing public services, an open data architecture may best enable the local technology community to develop new insights, products and services. However, enhancing a subset of public services like health care delivery may not be achievable through open data architecture if there are legal restrictions on how hospitals can collect, use or contribute information. In these cases, a centralized architecture that reconciles legal requirements may provide the best technical architecture.
Types of Data Sharing Models
Centralized Architecture
Description: With a centralized architecture, the governing body creates the database, standards, platforms and holds them locally. With a centralized platform the governing body has the greatest control of the management and enforcement of the assets held within. This is due to the fact that the infrastructure in place was built by the organization, creating ownership of the assets.
Data Access: The data within this centralized method is stored in one place, with the governing body granting access through a central point of access.
Data Analytics: The governing body creates unified standards that the data and platform utilize. This allows for the most powerful search, analysis and quality assurance of aggregated data
Costs: Heavy upfront costs building and maintaining the centralized repository. Low ongoing costs relating to maintaining the repository.
Change Management: The centralized architecture provides the greatest control of change management, but may also inhibit innovation.
Semi-Centralized Architecture
Description: The semi-centralized data sharing platform is a hybrid between a centralized and distributed system. In practice, we have seen centralized platforms and infrastructure built by a governing body, with public and private institutions creating and maintaining their own sharable repositories of data which adhere to the governing body's principles and standards.
Data Access: A central portal or platform grants access to the multiple repositories of data.
Data Analytics: Cross-repository searching and analytics, metadata and aggregate statistics can be developed by the central authority.
Costs: There are costs associated with developing data interoperability mechanisms and common usage policies. Ongoing costs include operating and maintaining the portal and administering policies.
Change Management: The semi-centralized architecture provides moderate control measures, as the governing body can make adaptations and improvements to platforms and standards, but will require some co-ordination to facilitate change across any decentralized repositories.
Decentralized Architecture
Description: In a decentralized system, the nodes of information are held with the various participating entities, and are all interconnected to encourage the sharing of their repositories for approved uses. In this system, the governing body creates standards and policies for all partnering entities to follow to ensure ease of access to information and the ability to utilize them. Each entity creates and manages their own repositories, and may provide their own individual platforms for data access.
Data Access: Access to each repository separately, but under a common usage or access policy and single approval.
Data Analytics: Because the data all held by various repositories, an index or catalogue is the only method to obtain data.
Costs: Initial costs are for the process to develop common usage and ontology. Ongoing costs include the management of the distributed ecosystem, the administration of policies and standards and maintenance of the repositories.
Change Management: The decentralized architecture provides some control measures, as governing body can create new standards, but will require substantial co-ordination to facilitate change across the repositories. Local innovation is easiest in a decentralized model.
Open Data
Description: Common standards are created by an entity, collaboration, or group to create a repository of shared data. This method requires the exclusive use of non-personally identifiable information.
Data Access: Access to central repository with common usage, standards, access policy and single approval.
Data Analytics: Powerful search, analysis and high-quality assurance of aggregated data.
Costs: Initial costs to index and catalogue data and repositories. Little centralized costs and maintenance.
Change Management: Open data will require strong coordination across the system to have data sets in readable and usable formats. Open data also allows for open innovation. With open data, there is no ability to restrict or control its use.
Data Marketplace
Description: Neutral legal, tax entity, and platform that brings together buyers and sellers of data.
Data Access: Central database of repositories.
Data Analytics: Cross-repository searching and analytics.
Costs: Platform as a service: Build a central platform and point of access. Monitoring Service Level Agreements that achieve data quality.
Change Management: The data marketplace provides control of change management, as establishing entity can dictate terms and conditions. Data may require cleaning and adaptation by contributing entities.
Data Sharing Agreements
Description: An agreement between multiple institutions to share data according to certain terms and conditions. Data sharing agreements identify the standards which govern the collection, storage, security, analysis, re-use, and destruction of data.
Data Access: There is granted access to repositories through the agreements of each institutions with dictated terms and conditions around the use.
Data Analytics: Search and analytics of the repositories are limited to the agreed upon data sets. Requires careful consideration and strategy to compile complementary data sets to extract value.
Costs: Low costs relative to other models, as parties are granting access to data repositories. Costs include staffing and expertise to clean and make use of the data.
Change Management: Low change management ability as agreements are typically narrow in scope for a specific purpose. Flexibility is inherent as new agreements can be created to better suit needs of an organization.
