Pol.is, Official Languages and a Shift Towards ‘People-First’ Policy Development
Open policy-making is an opportunity for government and stakeholders to move from linear, polarized, single-issue, interest-based considerations to interactions that are networked, collaborative, opportunity-based and where complexity is viewed as an asset. Adapted for the Government of Canada context, Pol.is is a cost-effective and highly scalable, digital engagement platform that can be used as part of broader strategies to put people and robust evidence at the heart of government decisions.
CORE PROBLEM: There is an ever-increasing awareness that government policies, programs and services are developed in environments that are as complex as public challenges the seek to address. Diverse perspectives are often acknowledged as potential assets in addressing complex problems in novel ways. But tensions can are easily exacerbated in the digital era, as affinity groups around wide ranges of perspectives can quickly coalesce and, more often than not, collide. The resulting tensions create all sorts of challenges for governments, which struggle to know how to engage with stakeholders and citizens and refresh policy approaches to complex domains given with the volume, velocity and variety of standpoint in a marketplace of competing ideas. In the Canadian context, public engagement efforts are further compromised because many tools, methods and approaches cannot be successfully adapted to accommodate the needs of a highly heterogenous population, including linguistic diversity.
OPPORTUNITY: Pol.is is an artificial intelligence (AI) powered, open source, engagement platform that identifies areas of consensus (and lack thereof) from across stakeholder groups within large, heterogenous populations. Pol.is quickly and effectively generates the sort of rich information usually collected through time-consuming and expensive community forums, focus groups or interviews. It accomplishes this by using machine-learning algorithms that extract patterns from data produced by respondents. Adapted for use in the Government of Canada (GC) context, Pol.is was adapted to meet all compliance requirements (including official languages) for deployment the Canadian federal context. Pol.is facilitates open-policy-making by letting individuals do what individuals are good at— sharing perspectives and experiences by submitting tweet-length statements, as well as reading and voting on the statements submitted by other participants. As a conversation unfolds, Pol.is allows computers do what computers are good at—finding patterns and visualizing the results. Used as tool within stakeholder engagement initiatives, Pol.is allows governments to demonstrate commitment to building stronger relationships across diverse stakeholders as part of building a stronger evidence base to address public goods issues.
HOW DOES A POL.IS ENGAGEMENT WORK? A cloud-based engagement platform that and can be accessed from any device, Pol.is is a tool offers the opportunity to enhance openness and participation, placing much of the control into the hands of participants. Participants build a conversation with each other when individuals submit tweet-style statements that reflect their experiences and perspectives. The accumulation of statements creates a dynamic, virtual conversation. The more diverse the participants, the broader the conversation. To add a quantitative layer to this qualitative (or anecdotal) exploration of an issue, Pol.is facilitates participants to ‘vote’ (i.e. agree, disagree or pass) on each statements, thereby deepening the conversation. Participants can start anytime during an engagement. They can visit the conversation as often and for as long as they want, contribute as many statements as they wish. The platform efficiently tracks voting by participants, so that on each subsequent visit participants only see a stream of statements upon which they have not yet cast a vote. In an open engagement, the Pol.is link can be shared by participants with anyone they think might be interested in the topic under discussion. And with the GC adaptation, participants are able to choose whether and when to apply translation. In addition analyzing and visualizing the data and reporting results of the conversations themselves, Pol.is also provides data regarding participant practices, which can be explored to ensure strengthen government approaches to stakeholder engagement.
PILOT PHASE OF POL.IS IN A GC CONTEXT: Pol.is first use in the GC was in an open conversation around the changing nature of the visual arts marketplace in Canada. The goal: to host a conversation on a single digital channel, in which participants could participate in the official language of their choice (i.e. English or French). Usually, this would require two separate channels, one per official language. This might further result in costly complaints to the Canadian Official Languages Commission, as well as less trust and/or understanding around a shared public goods issues. By committing to an open engagement with stakeholders, the government team was able to invest its time into efforts to adapt Pol.is to deliver a frictionless experience to participants while being compliant with all GC requirements, including official languages policy. Five subsequent tests Pol.is deployments have been undertaken at Canadian Heritage. Data and insights from the pilot phase are collated into a document that discusses scaling open engagement. Subsequent phases are being scoped.
What Makes Your Project Innovative?
Through agile process, Pol.is was adapted in six months as a digital engagement to comply with all GC requirements related to data privacy, security, accessibility and official languages. Pol.is was deployed six times in 2018 to identify the conditions under which this digital tool addresses the following long-standing stakeholder engagement challenges in a Canadian open government context.
- How to support both Canadian official languages in a single conversation without the lag-time or costs associated with traditional translation services?
- How to accommodate any number of participants without any increase in the time or costs for data analysis?
- How to combine qualitative and quantitative data as relevant policy evidence?
- How to host digital engagements that do not devolve into combative monologues where the loudest voices dominate, and long-held impasses prevent groups from moving forward on complex policy issues?
What is the current status of your innovation?
Implementation (Pilot Testing and Refinement): Public service experts from IT, security, privacy, accessibility, official languages, communications, public opinion research, policy research, policy development, public engagement, open government, procurement and legal services teamed up to pitch, adapt and deploy Pol.is in under six months. We worked in deep collaboration with external partners. For instance, on the technical side, the small technology startup company that developed Pol.is supported the platform’s adaptation. During the same six-month period, we worked with 25 stakeholders groups and prepared for the pilot deployment of Pol.is in a national engagement, which explored the impact of digital disruption on the marketplace dynamics experienced by visual artists. Five subsequent deployments took place in 2018. The data from this pilot phase is being used to develop a plan to sustainably scale the use of Pol.is across the GC.
Collaborations & Partnerships
Pol.is was developed by a technology start-up based in San Francisco, with which the government team worked closely to adapt Pol.is an digital engagement tool to fit the open government context. Using agile process, specialists articulated and tested versions against GC requirements, while the start-up made changes keeping user-experience at the heart of its design. The start-up continues to learn alongside us, which has led to transformations in its own business model.
Users, Stakeholders & Beneficiaries
We used Pol.is in 3-week open conversation with a broad and diverse cross-section of stakeholders in the Canadian visual arts marketplace, including artists, wholesalers, institutional collectors, galleries, museums, schools, artist-supporting organizations. We supported stakeholder learning and preparation to use a new digital tool. In a data-poor issue area, stakeholders were eager to generate data, even asked to extend the conversation by one week once it was underway.
Results, Outcomes & Impacts
FROM DIGITAL SURVEY TO ‘PEOPLE-FIRST’ DIGITAL ENGAGEMENT: Before Pol.is, surveys developed by analysts were deployed for a 4-week periods, generating between 30 and 75 responses. During a 4-week period, Pol.is drew 752 participants, 200 of whom submitted statements. The total number of statements was 577. The total number of votes cast was 43, 968.
LANGUAGE NO LONGER A BARRIER: The adaptation of Pol.is put in-real-time translation capacity (via Google Translate) at the hands of engagement participants, 87% of whom participated in English and 23% in French. Google Translate proved to be 77% as effective as paid-translation services, with huge time- and cost-savings (i.e. in-real-time and free) to Canadian tax-payers. This solution can be scaled up to unlimited numbers of participants, which allows for larger data captures. Moreover, it is linguistically-scalable and can accommodate any language, including Indigenous ones, that are offered on web-based translations sites.
Challenges and Failures
LANGUAGE: Language politics is very sensitive in the Canadian federal context. Traditionally, we have not had the tools to host effective, cohesive, participant-driven conversations as part stakeholder engagement and open dialogue. Previous attempts to use online tools have yielded linguistically-segregated and/or –marginalized results. While this pilot benefitted from the time- and cost-savings associated with artificial intelligence and machine-translation, translation accuracy rate hovered around 77%, eliciting a single complaint to the Office of the Commissioner of Official Languages (OCOL) and triggering a project review. We anticipated (indeed invited) such an occurrence in order to extend the dialogue we had started with the OCOL prior to the deployment of Pol.is. The dialogue with OCOL is an on-going, productive one. They remain interested in the use of technology to foster language duality, but would like to be kept closely briefed on our activities.
Conditions for Success
There needs to be sufficient political will and public service know-how to put people at the centre of policy development. We have learned that a tool is just the start of ther journey. Successfully and sustainably scaling of Pol.is across the GC requires increased capacity building in the following areas: open policy-making, digital literacy, public engagement. Now that we have a well-functioning tool that has proven to be compliant across GC requirements, we have started to scope out a new phase of work, and plan to embed a data scientist and a design professional onto an engagement/policy development team, so that we can use the tool across an entire policy development cycle. We continue to stay connected with the GC communities of practice and secretariats that have supported us to this point. We are also on-boarding new partners, as well as connecting with external specialists that can help inform our work as it moves forward.
As part of the pilot phase, 5 subsequent deployments of Pol.is occurred throughout 2018, as Canadian Heritage explores the use of Pol.is in different stakeholder relations context. These cases have helped us develop ‘use-when’ scenarios, as well as a guidebook to help other teams across the GC implement their own Pol.is open engagement processes. We have found that the tool functions as desired in terms of opening (and increasing) participation, generating data and ensuring linguistic and other forms of inclusion (i.e. accessibility, etc.). We have also learned that the capacity required to surface the nuances in the data requires a level of data literacy beyond that found in the average policy, program or research analysts. In terms of next phases of work on Pol.is, we are seeking to build strategic partnerships with organizations that have it as their mandate to increase the data capacity across the GC.
OPEN POLICY-MAKING IN THE DIGITAL ERA: Digital technology can build relationships, as well as enhance trust and understanding across diverse standpoints. Pol.is effectively collates diverse perspectives and sorts through areas of consensus and lack thereof. As such, Pol.is is at its best when combined with other tools and methods to create a cohesive engagement strategy. For open policy-making, Pol.is data requires stakeholders to be involved in co-interpreting results.
DATA LITERACY & EVIDENCE: Pol.is yields AI data that surfaces nuances and patterns across the many issue and stakeholders that combine to make up a policy challenge. Advice from data scientists and specialists throughout a stakeholder engagement project’s life cycle is required, a but such capacity is limited within Government of Canada context. Opportunities for Phase II Pol.is work is currently being scoped, particularly collaborations with internal partners (i.e. Statistics Canada, specialists working in the public opinion research domain, etc.). Meanwhile, our start-up partner has come to realise that the introduction of a software-as-a-service (SaaS) solution was simply a starting point for their journey. Their business model is now evolving from a private company to a benefit corporation model (B-corp, a for-profit company with clear commitments and priorities societal and environmental agendas) whereby additional services and tools can be developed to support and enhance data literacy and capacity.
LANGUAGE: The adapted Pol.is platform offered a highly effective and exciting remedy to some of the quality, cost and time challenges related to official language use within stakeholder engagements. Any limits to the effective exploitation of machine translation can be addressed through clear user-guidelines around submission of participant statements.