In our recent essay, Reconfiguring the State for the Internet Era, we argued that government needed to fundamentally change to meet people’s evolving needs and its own responsibilities. This pandemic has shown the importance of a strong state, yet also exposed significant weaknesses, with public institutions only bleeding legitimacy each time they fall short of the expectations and needs of the networked public. But merely reinforcing the existing operating model of government with new technology is never going to be enough. Instead we need deep, foundational reform across (at least) four categories: infrastructure, organisation, competition, and civic engagement.
To inform our research, we collected some short case studies of leading operating models across the public and private sectors:
Public sector: Estonia, India, Taiwan, London Office of Technology & Innovation (LOTI)
Private sector: Amazon, Spotify, Basecamp
While no country or organisation has mastered the tech revolution nor provides a ready-made blueprint for others to simply follow, this exercise helped to illustrate ways of materialising the high-level goal of government reform. We also recognise there are other aspects to each of these case studies not covered in this analysis.
We’re keen to hear of any other examples or insights we’ve missed here, so do get in touch: @BeaconRosie & @andrewjb_
Estonia operates as a digital government founded on two foundational technologies: a national, digital identity and X-Tee (formerly known as X-Road), a data infrastructure platform. Combined with governance rules that require data to be collected only once, the Estonian government is based on an digital operating model that is internationally renowned.
What this means for people:
Investing in modern data infrastructure and secure, user-centric digital identity has allowed for new models of public services, such as proactive child benefit payments: parents don’t need to apply for financial support, since the state already has the necessary information to anticipate needs.
The Estonian government estimates that its approach to digital government, including e-signatures underpinned by a digital identity, has generated annual savings equivalent to 2% of its GDP and given back 5 days to every adult, every year, saved from avoiding paper-based bureaucracy.
99% of public services are online, 99% of Estonian residents have an ID card, and completing taxes online typically takes only 3-5 minutes.
X-tee data exchange infrastructure prevents the need for a single, centralised database and embeds an audit trail to promote transparency. Used by 1000+ organisations in Estonia daily, data is only stored where it’s created and accessed by other services over an API. The system also creates an indelible record each time anyone accesses a users’ information, and there are penalties for doing so without consent.
Estonia, Finland and Iceland are working to expand this infrastructure internationally by creating a federated data exchange layer across the two countries
Diagram: Estonia X-Tee Data Exchange Layer
E-Estonia – Estonian government
Parents no longer have to apply for family benefits – Estonian Public Broadcasting
Data Exchange Layer X-tee – Estonia Information System Authority
Estonia, The Digital Republic – The New Yorker
India has created a National Open Digital Ecosystems (NODE) strategy, or the ‘India Stack’: a set of interoperable software services supporting digital payments, authenticated paperless documents, and identity verification. This has improved access to many public, financial and other private services, but inadequate privacy protections and data security, as well as centralised government control, have undermined public trust.
What this means for people:
99% of India’s adult population, 1.064 billion people, are enrolled in Aadhaar, the biometric identity which is foundational for the entire ‘India Stack’ and which provides access to public, financial and other services. In 2015, the country’s unbanked population was 233 million, half the number it was in 2011 at 577 million.
People can benefit from direct financial support from government, which is underpinned by Aadhaar-based identity verification.
16.96 billion e-KYC (know your customer) checks have carried out, saving more than $7 billion.
Aadhaar, the foundational biometric identity, is part of a stack of services to enable presence-less, paperless and cashless service delivery. Individuals receive a card with a 12-digit ID number which is linked to their fingerprints and an iris scan, held in a central database. This identity layer allows for e-KYC checks and e-signatures, as well as authentication for payments via another part of the stack, the Unified Payments Interface (UPI).
Open APIs were important in ensuring take-up. In order to facilitate take-up from businesses and citizens alike, all of these services were launched as application programming interfaces (API), which enable services to interact and share data. This approach allowed the government to develop the India Stack in phases, starting with the Aadhaar identity layer and building other layers (e.g. e-signatures, UPI, document storage) on top.
Privacy, security and inclusion concerns have been routinely relegated. Without strong privacy legislation in place from the beginning, Aadhaar became a de facto requirement for using many services, including opening a bank account. The Supreme Court banned this practice in September 2018 because of privacy concerns, included insecure private-sector data infrastructures which had been allowed data to be illegally sold on WhatsApp. Strict central government control has also raised concerns about intrusive surveillance and exclusion from accessing public services, undermining public trust.
Diagram: India Stack
vTaiwan demonstrates how digital tools can enable constructive, civic engagement and public deliberation without overwhelming government resources. These processes can also be tightly integrated into policymaking, making legislation more legitimate, representative and effective.
What this means for people:
Thousands of people can now participate in a rolling programme of civic discussion and policymaking in Taiwan. For example, 9,700 people mobilised to digitize 30,000 campaign finance records in 24 hours, more than 4,000 people helped solve the complicated issue of Uber regulation, and 2000 online participants and 20 face-to-face contributors discussed 200 proposals to achieve a consensus policy position on regulation of ‘shell companies’.
This improved transparency and political trust is also a potential contributor to Taiwan’s effective response to Covid-19.
vTaiwan uses digital tools to communicate with stakeholders precisely and enable personal expression that can materially impact policy. This began with Discourse, a forum-based technology, but the results were mixed. They then deployed Pol.is, an AI based tool that prompts people to agree, disagree or skip several statements, as well as offer their own statement. Pol.is then clusters users who voted similarly into opinion groups. Surveys are shared organically and advertised via Facebook to improve reach and representation. The Social Innovation Lab in Taipei also hosts regular hackathons to rethink how to deliver many government services or policies.
vTaiwan is radically transparent and open. The Taiwanese government had been criticised previously for its lack of transparency. When g0v (govzero), the civil society group that co-ordinates vTaiwan, formed in 2012, this civic technology community gained significant public credibility after successfully showing how to conduct transparent democratic processes at scale. Audrey Tang, g0v member and Digital Minister for Taiwan, routinely publishes meeting and office hour minutes online, and all dialogues that use vTaiwan facilitators and agree to open transparency are live streamed, transcribed and posted on vTaiwan’s website. Minister Jaclyn Tsai, who manages national government policy around technology, also ensures that for every policy that uses vTaiwan all background material is available online. This transparency radically reduces the traditional information asymmetry between policymakers and citizens and improves public trust.
vTaiwan is embedded into government processes. Mass, civic engagement technologies are only one part of vTaiwan. AI-facilitated conversation tools are often the starting point for public meetings with scholars and officials. The Government also commits either to act on points that achieve consensus or to provide a detailed, justified explanation of why those consensus points are not yet feasible to ensure accountability. If there is no consensus on an issue following discussions in vTaiwan, ministries hold livestreamed consultation meetings that are subsequently posted online.
Diagram: Pol.Is Opinion Clustering System
vTaiwan: Public Participation Methods on the Cyberpunk Frontier of Democracy – Liz Barry
Notes on Aspects of the vTaiwan Phenomenon – Tom Attlee
Uber responds to vTaiwan’s coherent blended volition – Audrey Tang
LOTI is a membership organisation for the London Boroughs. Recognising that many technology challenges are shared across boroughs, it works on building digital capability, coordinating projects, sharing insights and collaborative problem solving. LOTI also illustrates the power of working in the open for collaboration between different public sector bodies, making service delivery more effective.
What it means for people:
LOTI helped coordinate free school meals provision and support for vulnerable residents across London during the Covid-19 pandemic by improving data sharing and collaboration between boroughs: since some people would live and work or go to school in different areas, no single borough had all the information they needed to deliver support. This is part of broader work on improving data management and collaboration for many issues that transcend any one area.
LOTI used its unique position as the collective voice of boroughs’ data and policy teams to campaign for improved public health data on Covid-19, in turn facilitating more tailored support, and launched 7 different proposals to improve digital inclusion.
Developed a toolkit for launching a digital apprenticeship programme.
Launched City Tools: London. An interactive dashboard that maps technologies, contracts and skills across London’s boroughs.
Developed a standardised process for Information Governance.
Developed some smart street prototypes.
Provided a wealth of guidance, supported data transfers and informed the choice of new technology solutions to support borough priorities in response to Covid.
All of LOTI’s work and insight is open to all. Although only 15 of the 33 London boroughs (inc. City of London) are members, work carried out is made available to all. They publish week notes, reusable resources, guides and toolkits. In their 2020 Annual Report, they also record projects that were less successful, demonstrating that agile working in the open means embracing successes as well as failures in order to improve long term. These provide important lessons: one example was a project to implement seamless wifi across London’s public sector estate, which on reflection needed a discovery phase and greater capability to deliver from all members.
They support integration across the boroughs through shared standards and data collaboration. Part of LOTI’s work is improving collaboration between member boroughs, in an attempt to make co-operation more seamless. When starting out in June 2019, they identified three core areas where they needed to ‘fix the plumbing’: improving digital skills, technology procurement and data collaboration. The latter was particularly important for some of their projects during Covid-19, such as improving data sharing to assess who was entitled to free school meals. Another example is their establishment of a LOTI data analysts network for boroughs to exchange insights on Covid-19 challenges.
They focus on iterative processes and prototypes. LOTI is also interesting for modelling ways of working which are found in the tech and start up sector in a local government context. Rather than setting up a working group on Smart Street Infrastructure, they did a week-long sprint which generated the seeds of three prototypes. They are also hosting a series of seven short pitches by digital leaders where they showcase their innovative solutions to improving digital inclusion. This kind of competition and prototyping means that the innovations have been researched and tested, accelerating the adoption process.
Diagram: LOTI's Outcomes-Based Methodology
Personal reflections on a year of LOTI – Eddie Copeland
LOTI: Year One Weeknotes – Onyeka Onyekwelu
COVID-19: Borough Challenges with PHE and NHS datasets - LOTI
Let’s finally fix data sharing in London - Eddie Copeland
Massive investment in scalable, microservices infrastructure allows Amazon to deliver effectively due to minimal interdependencies while reducing the costs and risks of experimentation.
What governments could learn from this model:
Investing in this infrastructure would give public institutions far greater ability to respond to people’s evolving needs more quickly, e.g. designing and iterating new public services at scale.
APIs also allow other organisations and services to focus on users’ various needs – e.g. registering a business, planning a route, or registering for a training course – and remove the bottleneck of relying on a single public body to meet all these needs.
Shifting to an API-based, microservices infrastructure was necessary for Amazon to adopt a scalable, platform business model. Since 2002, when it made this shift, Amazon’s market capitalisation has grown from $8bn to $1.6tn (20000%).
By 2014, this model allowed 50m+ code deployments every year, or 1.5 every second. This improves on every 11.6 seconds in May 2011.
This model has since been adopted by thousands of technology companies and some governments, most notably Estonia.
In 2002, Jeff Bezos required every team to expose their service and data via APIs, prohibiting all other ways of linking to or sharing data. By default, all services were designed for others to be able to integrate them, so that no service would end up as a large monolith. This approach also enabled Amazon to organise in lots of small teams who could move at speed, rather than large groups with lots of interdependencies.
Microservices have many advantages:
They allow for cheap and modular iteration, rather than expensive, time-consuming wholesale upgrades;
New services can be created quickly and tested cheaply, building on existing functionality rather than starting from scratch;
It is easier to innovate, since microservices are agnostic about which applications they integrate with so future needs or functions don’t need to be anticipated up-front;
Stability is improved since faults in each microservice don’t bring down the whole system.
This modern infrastructure has also allowed Amazon to solve the ‘calculation problem’ of large ecosystems: it has enough data about purchasing behaviour, warehouse space, packing efficiency, logistics resource and cashflow to predict supply and demand curves and set prices dynamically.
Diagram: From Monolithic Software to Microservices
11 Notes on Amazon – The Family
Amazon Sees Like a State – The Diff
Next Generation Digital Government Architecture – Kristo Vaher, Estonia Chief Technology Officer
Spotify can deliver at pace because it is organised around multidisciplinary teams which are highly autonomous yet also highly aligned to a common, global vision.
What governments could learn from this model:
Public services are often confusing or unwieldy because they’re delivered in different silos, and teams find it hard to work across this barriers. Organising in multidisciplinary teams with high autonomy could remove many of these barriers and would provide a more joined-up experience for people – reducing the burden on users to connect up different bits of government.
It could also allow Ministers to focus on longer-term issues and focus less on day-to-day delivery, where that’s helpful.
Daniel Ek, CEO: “I trust my team to manage the day-to-day, shorter-term initiatives and iterate as needed based on data and insights…this then frees me up to think about the long term…very few things…make their way up to me.”
“Squads” have end-to-end responsibility for specific services, covering “design, commit, deploy, maintenance, and operations”. Teams must therefore be multidisciplinary to include expertise + experience of all parts of this cycle, but other groups – Chapters, Guilds and Tribes – help specific disciplines and others to convene, share insights and plan effectively.
Focus areas are mutually exclusive to prevent interdependencies and delays. But global alignment enables effective delivery without local sub-optimisation.
Increasing complexity is solved by splitting up teams, rather than adding more people to the team, to retain agility. Some common standards are required between teams to minimise the start-up costs for new teams and promote broader collaboration.
Diagram: Spotify's Multidisciplinary Organisational Structure
Daniel Ek – The Observer Effect
Spotify Engineering Culture: Part 1 – Spotify Engineering
Spotify Engineering Culture: Part 2 – Spotify Engineering
Basecamp is a software company providing project management and collaboration tools. These tools are especially designed to enable remote working, which Basecamp implement themselves. The company demonstrates how technology can enable productive, remote working via open communication and collaboration, ensuring that organisations can recruit from anywhere and provide an improved work-life balance for employees.
What governments could learn from this model:
Civil servants could benefit significantly from public bodies using tools like Basecamp, as this could improve visibility of projects across government, improving coordination and reducing duplication.
Ministers and citizens could also benefit from the civil service recruiting and attracting more remote workers, which could help bring the very best people into the civil service and shift beyond a London-centric mindset. Both of these factors could improve public services and government decision-making.
Basecamp now has 3.3 million accounts set up in 2020. They grew from 45 in 2004, to 89,000 in 2006.
9 out of 10 Basecamp customers report having a better handle on their business. 8 out of 10 say their teams are more self sufficient. 6 out of 10 have fewer weekly meetings.
It is built on the idea of open, centralised communication. One advantage of remote work is the elimination of many office interruptions, and Basecamp try to mirror this in their culture and system design. Common, central tools are deployed to maximise effectivness, based on the idea that spreading work across emails, file services, task managers and spreadsheets is highly unproductive. 98% of internal communication happens inside the Basecamp system, and Basecamp themselves don’t use email internally and rarely have in-person meetings. Video calls are kept to a minimum to avoid disruption and text communication is preferred, so that colleagues can communicate asynchronously.
Universal transparency enhances project management and removes interdependencies. Although centralised, the Basecamp system is built on openness: the company thinks it should be easy to see the bigger picture as well as the finer detail, irrespective of your position in the company. Each day and week the tool asks for a short update on what people are doing to improve visibility across the company and ensure nothing falls through the cracks. This reflects a view that most teams spend too much time on status meetings, when tools can handle this need more effectively and without meetings. Work is also broken up into separate projects, with each project containing everything related to the work: the people involved, every discussion, document, file and task.
Basecamp: Before and After - Basecamp
They Led the Cult of Remote Work. Now We’re All Members. – Adam Bluestein, Medium
Basecamp’s Jason Fried on the Learning Curve of Remote Work – Chrysanthe Tenentes
The Five Levels of Remote Work — and why you’re probably at Level 2 – The Startup