What if a portal had just one job to do? And could do it well? Speculations on portal futures...
This is the last (for now) in a short series of posts exploring the past, present and future of (open) data portals, as part of an Open Data Institute project on Data Platforms and Civic Engagement. So far we’ve looked at the (1) terminology of portals; (2) the evolution of different technical portal tools; (3) a framework for thinking about the different layers of a portal in practice; (4) the explosion of academic literature on portals; (5) some examples of portal experiments and innovations; and (6) some of the different organisational dynamics around data portals. In this final post of the series, I look at some potential future visions of the data portal. I’m doing this with a five potential perspectives on the portals of the future.
As we saw in previous posts, numerous early academic, and perhaps even activist, visions of the data portal made the assumption that the end-point of a mature open government data portal was to be a platform where transparency and citizen participation were directly realised. That is, portals as interactive, social spaces where datasets were not only described, but visualisations were shared, policies discussed, and citizen-government interactions explored. In this vision, the goal of portal development is to make the engagement layer integral to the portal. Experience from the last decade suggests these maturity models are likely wrong.
So, if the portal is not the place to host citizen participation, what role can it play as an enabler?
Each of the sections below offers a claim for the focus that a ‘minimal’ modern portal should have. Imagine, if you will, that the decision is made that the portal will just do this one thing, and do it well. It then offers some initial thoughts on what it would mean to design portals in this way, and checks whether such a portal would have a role to play in enabling citizen participation.
Datasets need to be described. The main and only essential function of a portal is to help governments to create, quality assure, and publish, standardised structured meta-data about datasets.
Modern meta-data tooling should help data owners to create meaningful descriptions of their datasets, using carefully designed forms and workflows for preparing and publishing meta-data (including handling approvals).
Advanced meta-data factories might provide tools that can look inside datasets to suggest geographic or thematic tags, or dataset descriptions. Plugins might provide support for richer annotation of dataset context as part of the publishing workflow, including applying ideas from ‘Structured Context’ and ‘Datasheets for Datasets’, or building on Open Data Certificates.
The portal as a meta-data factory faces inwards to the organisation, and is agnostic about how the structured meta-data is then made available to search or browse. On one pathway, the resulting standardised meta-data (DCAT, Schema.org) could be integrated into existing Content Management Systems, offering the ability to manage data, pages and documents side-by-side on government websites, whilst still making data available to tools like Google Dataset Search.
What does this contribute to citizen participation?
Tooling that is focussed on describing datasets, rather than listing them, could encourage a quality-over-quantity approach to public data. It could improve the chance that data will be published and maintained alongside other contextual information, rather than being shunted into a ‘data silo’ on the portal - whilst making sure data remains discoverable through federated dataset search tools.
By placing the emphasis on rich descriptions of data, citizens can more quickly understand whether the data will be of use to them (a good description is so much more useful than a mediocre ‘Preview this data’ tool!).
Citizens need to be able to trust that public data is governed in the public interest. Data portals should list all the data a public organisation holds or has access to, and should help citizens understand what it is, how it is managed, and who has access to it.
Transparency isn’t just about making selected datasets public. It’s also about letting citizens know about the ways in which data is used to make decisions that affect them. That might involve providing a register of all the data sharing agreements a public body has entered into, or detailing the specific services that are accessing and using particular datasets for automated decision-making.
The portal as a data governance hub builds on ideas from enterprise meta-data platforms, which document not only datasets, but also the data views, machine-learning models, and internal or external stakeholders who are accessing given data resources.
The portal as a data governance hub responds to current concerns about data risks, and reframes portals for a ‘full spectrum’ world.
What does this contribute to citizen participation?
It removes the current opacity around data-sharing, where private firms or other public agencies may have bilateral data sharing arrangements that are influencing policy making or implementation, but are more-or-less invisible to the public.
It supports discovery of how data is already being used, and allows citizens (and other internal and external stakeholders) to engage with and learn from this, rather than starting from scratch.
Open data is more useful when it is published according to thematic standards. Data portals should focus on Quality Assurance, and validate datasets against relevant domain-specific standards, monitoring and reporting on data quality and uptime.
Whilst portals might host a long-tail of one-off datasets, there is a discrete set of ‘high value datasets’ that most public portals might contain, from public spending records and procurement data, to service locations and land assets. Fortunately, standard schema exist for most of these and portals could be providing automated validation of data against these.
By reporting on data quality both to public users, and back into the organisation, the portal as a quality assurance tool drives alignment of data, and reduces the friction to data use.
What does this contribute to citizen engagement?
Local portals that can provide automatic reporting on whether mandated publication or high-value datasets are available allow citizens to get on with analysing the data - building shared tools that can work across different areas to spot patterns, explore trends and use data productively. The shift the burden of reporting poor data quality from the users onto the management layer of the portal team.
Portals should focus on data demand, not supply. A portal as switchboard surface focusses on capturing both internal and external data needs, and then identifying how to meet them using existing, or new, data resources.
In “From Repositories to Switchboards: Local Governments as Open Data Facilitators” the authors argue for “a novel role of the custodianship and facilitation of Open Data exchange”- pointing towards a model in which portals might surface both government questions that could be answered with data from outside the organisation, and citizen or company needs that could be served by opening or sharing data from inside the state. This positions the portal as a platform that can allow data to flow into government, not just out of it.
The portal as a switchboard is primarily a service made of people empowered and equipped to engage internally and externally, though it might also be supported by technical workflows.
What does this contribute to citizen engagement?
This gives citizens the chance to create and contribute data, rather than just access it. It creates opportunities to create broker more human connections between data owners inside government and potential external data users.
Every dataset tells a story. Portals should focus on providing data-specific visualisations and exploration tools for priority datasets.
This model is already evident in a number of places, such as the curated views for certain data packages on DataHub.io, selected datasets on the London Datastore, or OpenDataSoft’s Key Performance Indicator (KPI) dashboards. However, dataset narratives and tailored exploration tools are generally bolted-on, rather than built into, the way platforms work.
A portal as a narrative platform would focus on giving data publishers the tools to highlight the most common uses of their data, providing users with faster ‘data->fact’ access. It would place the emphasis on presenting interactive data access, rather than raw data, or fixed analysis. It rejects the idea that all data interfaces will be built by external intermediaries, and establishes the important role of data platforms in mediating access to data.
What does this contribute to citizen engagement?
It provides shared tools for use by internal and external stakeholders.
It gives individuals direct access to data points they can use to engage with government: asking questions, contesting decisions, or proposing solutions.
The sketches above are, of course, not mutually exclusive. And there are obviously many more ‘The portal as…’ visions that might be outlined.
These are just the five that seemed most relevant to me to capture some of the exploration I’ve been on over the last few weeks, and to contribute to conversations about what role portals could or should play in future. At this stage, I’m not at all sure which, if any, I think is the most promising direction for travel. However, I do think that, just as earlier data portals were the product and tools of a movement for open data, refocussing and revitalising open data portals (and open data itself) will take movement-building narratives, energy and action: and so the choice is not just a technical one.
I am, however, sure that the answer is not the “Mrs Armitage’s Biycle” solution of adding in everything. What this portal needs, I said to myself, what this portal really needs…
Early in 2022, we’ll be taking some of the learning and ideas from this deep dive, as well as a set of parallel interviews and outreach with portal managers and users, into some form of design sprint. Format tbc. - but given current work from home rules in the UK, likely to involve considerable online component over 1 - 2 weeks in early Feb. We’re hoping to come out with some artefacts that can feed into future developments of the portal landscape.
If you would like to be part of that process, please do get in touch. You can drop me a line to [email protected] or reach out on Twitter @timdavies.