Government as a cognitive system has been saved
The authors would like to thank Thirumalai Kannan D and Dimple Jobanputra for their research contributions.
Cover image by: Lucie Rice
United States
United States
India
United States
Data and information have been established as the fuel for the global economy. Now governments are developing new capabilities to exploit the power of data for social good. Governments are learning and evolving—just as cognitive systems do.
Information, whether about the past and present and future scenarios are augmenting decision-making and creating immense value by providing actionable insights. We are witnessing the emergence of government as a system of cognition that:
• Uses hindsight, in that it leverages advanced analytics and machine learning capabilities to access evidence from the past on what has worked and what hasn’t.
• Monitors real-time information on what is actually happening to inform tactical decisions.
• Builds foresight by applying predictive analytics and conducting simulation exercises to anticipate events before they occur and implementing preventive measures.
The cognitive systems approach represents a fundamental advancement in the traditional form of the “predict-prevent-evaluate” cycle. Now governments can design programs with an intelligence architecture in mind: shaping or even changing future outcomes by introducing scalable, real-time information to adapt predictive models that are founded on the hindsight of past performance.
COVID-19 and its aftermath have shown how important it is to be armed with the right information to guide decisions at the right time. For instance, Taiwan’s successful handling of COVID-19 showcases how a government can couple evidence from the past with information about the present and futures thinking to build an effective response to the pandemic (figure 1). When a whistleblower report warning of a new SARS-like virus in Wuhan began circulating on internet message boards in December 2019, Taiwan's medical officers took note and snapped into action. Their swift response was due to previous experience with the SARS virus (hindsight). Informed by years of post-SARS scenario planning (foresight), Taiwan rapidly implemented travel restrictions and health screenings for people traveling from Wuhan.1 Then, to prevent panic-buying early in the pandemic, Taiwan published real-time stock levels of masks at all 6,000 pharmacies in the country.
Countries such as the United States, Australia, Peru, and India have institutionalized evidence-based decision-making, both governmentwide and within specific departments,2 buttressed by the proliferation of data and the advent of emerging technologies. Public workers are increasingly assessing their policy decisions and approaches to evaluate what works and what doesn’t.3 For instance, the United Kingdom’s What Works initiative is a network of research centers and affiliates focused on finding “the best available evidence” to inform government decision-making.4 In the United States, the What Works Cities program encourages local government to use data to tackle pressing challenges faced by residents.5
Also, instead of relying solely on randomized controlled trials and other traditional evaluation methods, policymakers can now deploy artificial intelligence (AI) and machine learning to analyze vast and complex data sets. As technology expands the possibilities for data analysis, government agencies worldwide are launching specialized units in-house or partnering with industry, universities, and nonprofits to bolster evidence-based decision-making.
From guiding budgetary decisions to assessing regulatory burden to improving education outcomes, governments are beginning to gather evidence from the past to course correct in the present and to inform future decisions. For example, the What Works State Standard of Excellence helps US states use an evidence-based approach to efficiently manage taxpayer dollars.6 States such as Colorado, Minnesota, New Mexico, and Rhode Island require agencies to identify and highlight evidence while formulating budget proposals, changing existing programs, or starting new ones. In Minnesota, such evidence-focused funding decisions led to a US$87 million investment in new and expanded evidence-based programs in the FY2020–2021 budget.7
At the other end of the globe, the Office of Best Practice Regulation evaluates Australia’s most complex policy problems using an evidence-based approach. The agency’s Regulatory Impact Analysis system provides a framework for creating clear, evidence-based policy that reduces regulatory burden.8
The Danish Agency for Labor Market and Recruitment, a unit of Denmark’s Ministry of Employment, evaluates and communicates the impact of active labor market policies by collecting existing evidence about what works and building new evidence through randomized controlled trials.9 Also, the European Commission has established a special unit called the Joint Research Centre to provide scientific support for European policies. The center hosts its own research facilities and specialist laboratories, and boasts of thousands of scientists, who provide evidence throughout the entire policy cycle.10
The biggest enabler of the cognitive government systems approach is the ability to make sense of the explosion of real-time data. While historical data can help with impact evaluation, real-time information can provide the missing link in the traditional evidence-outcome value chain. Government agencies are cognitively navigating vast troves of messy (incomplete, inconsistently defined and collected) and fast-moving real-time data to derive quick insights and develop a time-critical response. Operational applications of emerging technologies such as AI, which can generate insights and identify patterns within minutes, have made such data more useful than ever before.11
From managing defense and national security, to addressing environmental concerns, to meeting crucial public health demands, real-time information is helping public agencies stay on top of their game. For instance, Palanterra, an operational capability of the Australian Geospatial-Intelligence Organization, facilitates situational awareness by providing access to a strategic common operating view. It permits multiple Australian Commonwealth and jurisdictional organizations to view and share near-real-time data, which helps in supporting special security events as well as in managing disasters such as the Queensland floods and the Japanese tsunami.12
In another example, the US Geological Survey (USGS) partnered with NASA to use the latter’s observation satellites to capture real-time images of the Earth’s surface. The resulting data can help governments and policymakers make informed choices about natural resources and the environment.13
Governments are using real-time information as an anchor to quickly pivot their response should the situation demand, especially in the wake of the pandemic. They have followed data-intensive approaches to track and model the spread of the virus, develop therapeutics and vaccines, and manage health care capacity.14 The United Kingdom created a National Health Service COVID-19 Data Store to collect real-time information and drive the virus response at the national and regional levels. Such data can be used to track hospital bed capacity or ventilator supplies in a particular area, for example.15
In the United States, the Department of Veterans Affairs launched its National Surveillance Tool (NST) in June 2020 to keep track of ground-level COVID-19 developments and manage resources accordingly. The NST is also capable of performing predictive analytics to help the department anticipate future coronavirus hotspots and take pre-emptive action.16 Similarly, the Indian Council of Medical Research launched the National COVID-19 Registry in September 2020 to capture real-time data on COVID-19 patients. The information is used to support evidence-based clinical decisions, research, and policymaking.17
The education sector is relying on real-time data to improve student outcomes. In the United States, for instance, the University of South Florida uses a predictive analytics platform that feeds on real-time data that includes grades and class participation to highlight which students are facing challenges and develop appropriate interventions to support them.18 This has become all the more important given the sudden pivot to remote learning post the pandemic.
To prepare for future uncertainty, governments are establishing or expanding their sensing capabilities to better understand how long-term trends might play out. By analyzing multiple scenarios and running simulations, governments can assess the implications for current and potential decisions to form a long-term view. Moreover, COVID-19 has forced multiple government agencies to build better preparedness and take decisive action by anticipating what the future holds.
COVID-19 has also shown the power of informed foresight (“flattening the curve”) while also underscoring the importance of taking sound and timely decisions based on such insights. For instance, the applicability of foresight requires a clear understanding of the assumptions underlying the analysis–and dangerous extrapolations of results or misinterpretation are both real threats to effective use of foresight. So, new tools are placing requirements on decision-makers to raise their game if they want to effectively take advantage of a cognitive system.
Scenario planning is gaining ground in multiple public entities. For example, the Queensland government, together with Australia’s Q-Foresight program, examines long-term trends and risks pertinent to the state, including in transport, health, science, innovation, and environmental policy.19 Also, to help guide decisions in the postpandemic world, the US Department of Health and Human Services developed five pandemic planning scenarios. Public health officials used the resulting data to explore the potential effects of strategies such as social distancing, while hospital administrators were able to assess and plan for resource needs.20 Meanwhile, the Canadian government assessed three scenarios around various levels of COVID-19 control to guide their pandemic response. The country evaluated potential futures for “no control,” “weaker controls,” and “stronger epidemic control,” to estimate the duration and infection rate of the pandemic.21
Governments are also simulating different future possibilities to check their preparedness in crucial areas, including resilience to cyberattacks, disasters, and pandemics. For instance, European Union (EU) defense ministers participated in an exercise held in Estonia in 2017 to assess their ability to respond to potential cyberattacks against EU maritime forces and military headquarters. Also, the EU and NATO collectively conduct tabletop exercises to stay coordinated on hybrid warfare scenarios.22
In a similar vein, the Security Bureau in Hong Kong conducted an interdepartmental simulation of a super typhoon scenario in 2019. The tabletop exercise helped check for the contingency-handling capabilities and interoperability of the participating bureaus and departments.23 In April 2020, Taiwan’s New Taipei City conducted an hour-long tabletop exercise to simulate potential government responses to a second wave of coronavirus infections. Officials rehearsed interdepartmental coordination and practiced implementing measures such as travel restrictions, business closures, and supply rationing.24
Mathematical and statistical models are other tools that public agencies are leaning on for decision-making. While mathematical modeling has been prominent in research and academia, COVID-19 has brought it to the center of policymaking. Agencies are developing and using models to forecast the number of COVID-19 infections and even deaths. For instance, the Centers for Disease Control and Prevention (CDC) along with its partners leverages statistical or mathematical models to predict deaths and cases per week for the next four weeks. This helps the CDC make better decisions on resource allocation and implementation of social distancing measures.25
The Public Health Agency of Canada set up a Canadian COVID-19 modelling network comprising of experts from federal, provincial and territorial governments and universities. The predictions help guide public health measures while evaluating the impact of the existing ones.26
• As of June 2020, there were 169 examples of data-driven practices and evidence-based policies in place in 35 states across the United States.27
• The UK What Works Network encompasses policy areas that constitute more than 250 billion pounds of public spending.28
• A national poll conducted by Results for America, in collaboration with the University of Chicago, suggests that close to 92% of Americans believe policymakers should support their decisions with the best available evidence and data.29
We are witnessing the introduction of augmented intelligence into the public policy landscape. Leading governments are building on the lessons and successes of digital government to pursue a cognitive systems approach that embraces the next generation of information, data, and insight management.
• Be led by the science. Shift from instinct-based decision-making and rewire the mindset and cultural DNA of organizations to be data- or science- driven.30
• Operationalize cognitive decision-making. The value of an insight is a function of how actionable it is and what impact the actions have on the ground. So, apart from ingesting the information, it’s critical to react in a timely fashion. In the absence of a timely reaction or course correction, cognitive abilities to assess, sense, and adapt will fail to provide real value.
• Be proactive. Anticipate events before they occur, but also develop a timely, decisive response and not spend too much time deliberating on the probable outcomes.
• Become an agile organization. Shorten feedback loops (through sensing and monitoring) to enable more frequent, incremental improvements.
• Create a culture of continuous learning. Provide training and build capacities for conducting tabletop and simulation exercises.
• Encourage citizen participation and cocreation to build evidence. Institute frameworks, tools, and approaches that expand the traditional methods of forming an evidence base.
• Be a tech-instinctive organization. Establish a robust data collection mechanism. Ensure that the data collected is standardized and can be used by multiple agencies.
• Reduce barriers to data-sharing and revisit data interoperability norms.
• Adopt transparency tools. Utilize dashboards, open data, social media, and other tools to reveal how systems are working.
• Harness the power of AI to analyze all kinds of data, both structured and unstructured.
• Develop the skills to use these new tools. Real-time data can be messy. Decision-makers need the skills to understand the messy data and carefully interpret the implications of potential future outcomes.
• Ensure diversity. Formalize approaches to ensure diversity, equity, and inclusiveness in policy design, service delivery, and program evaluation.
The authors would like to thank Thirumalai Kannan D and Dimple Jobanputra for their research contributions.
Cover image by: Lucie Rice