How to Hold Algorithms Accountable


Nicholas Diakopoulos and Sorelle Friedler at MIT Technology Review:  Algorithms are now used throughout the public and private sectors, informing decisions on everything from education and employment to criminal justice. But despite the potential for efficiency gains, algorithms fed by big data can also amplify structural discrimination, produce errors that deny services to individuals, or even seduce an electorate into a false sense of security. Indeed, there is growing awareness that the public should be wary of the societal risks posed by over-reliance on these systems and work to hold them accountable.

Various industry efforts, including a consortium of Silicon Valley behemoths, are beginning to grapple with the ethics of deploying algorithms that can have unanticipated effects on society. Algorithm developers and product managers need new ways to think about, design, and implement algorithmic systems in publicly accountable ways. Over the past several months, we and some colleagues have been trying to address these goals by crafting a set of principles for accountable algorithms….

Accountability implies an obligation to report and justify algorithmic decision-making, and to mitigate any negative social impacts or potential harms. We’ll consider accountability through the lens of five core principles: responsibility, explainability, accuracy, auditability, and fairness.

Responsibility. For any algorithmic system, there needs to be a person with the authority to deal with its adverse individual or societal effects in a timely fashion. This is not a statement about legal responsibility but, rather, a focus on avenues for redress, public dialogue, and internal authority for change. This could be as straightforward as giving someone on your technical team the internal power and resources to change the system, making sure that person’s contact information is publicly available.

Explainability. Any decisions produced by an algorithmic system should be explainable to the people affected by those decisions. These explanations must be accessible and understandable to the target audience; purely technical descriptions are not appropriate for the general public. Explaining risk assessment scores to defendants and their legal counsel would promote greater understanding and help them challenge apparent mistakes or faulty data. Some machine-learning models are more explainable than others, but just because there’s a fancy neural net involved doesn’t mean that a meaningful explanationcan’t be produced.

Accuracy. Algorithms make mistakes, whether because of data errors in their inputs (garbage in, garbage out) or statistical uncertainty in their outputs. The principle of accuracy suggests that sources of error and uncertainty throughout an algorithm and its data sources need to be identified, logged, and benchmarked. Understanding the nature of errors produced by an algorithmic system can inform mitigation procedures.

Auditability. The principle of auditability states that algorithms should be developed to enable third parties to probe and review the behavior of an algorithm. Enabling algorithms to be monitored, checked, and criticized would lead to more conscious design and course correction in the event of failure. While there may be technical challenges in allowing public auditing while protecting proprietary information, private auditing (as in accounting) could provide some public assurance. Where possible, even limited access (e.g., via an API) would allow the public a valuable chance to audit these socially significant algorithms.

Fairness. As algorithms increasingly make decisions based on historical and societal data, existing biases and historically discriminatory human decisions risk being “baked in” to automated decisions. All algorithms making decisions about individuals should be evaluated for discriminatory effects. The results of the evaluation and the criteria used should be publicly released and explained….(More)”

New Data Portal to analyze governance in Africa


Governance and Service Delivery: Practical Applications of Social Accountability Across Sectors


Book edited by Derick W. Brinkerhoff, Jana C. Hertz, and Anna Wetterberg: “…Historically, donors and academics have sought to clarify what makes sectoral projects effective and sustainable contributors to development. Among the key factors identified have been (1) the role and capabilities of the state and (2) the relationships between the state and citizens, phenomena often lumped together under the broad rubric of “governance.” Given the importance of a functioning state and positive interactions with citizens, donors have treated governance as a sector in its own right, with projects ranging from public sector management reform, to civil society strengthening, to democratization (Brinkerhoff, 2008). The link between governance and sectoral service delivery was highlighted in the World Bank’s 2004 World Development Report, which focused on accountability structures and processes (World Bank, 2004).

Since then, sectoral specialists’ awareness that governance interventions can contribute to service delivery improvements has increased substantially, and there is growing recognition that both technical and governance elements are necessary facets of strengthening public services. However, expanded awareness has not reliably translated into effective integration of governance into sectoral programs and projects in, for example, health, education, water, agriculture, or community development. The bureaucratic realities of donor programming offer a partial explanation…. Beyond bureaucratic barriers, though, lie ongoing gaps in practical knowledge of how best to combine attention to governance with sector-specific technical investments. What interventions make sense, and what results can reasonably be expected? What conditions support or limit both improved governance and better service delivery? How can citizens interact with public officials and service providers to express their needs, improve services, and increase responsiveness? Various models and compilations of best practices have been developed, but debates remain, and answers to these questions are far from settled. This volume investigates these questions and contributes to building understanding that will enhance both knowledge and practice. In this book, we examine six recent projects, funded mostly by the United States Agency for International Development and implemented by RTI International, that pursued several different paths to engaging citizens, public officials, and service providers on issues related to accountability and sectoral services…(More)”

Digital Kenya: An Entrepreneurial Revolution in the Making


(Open Access) book edited by Bitange Ndemo and Tim Weiss: “Presenting rigorous and original research, this volume offers key insights into the historical, cultural, social, economic and political forces at play in the creation of world-class ICT innovations in Kenya. Following the arrival of fiber-optic cables in 2009, Digital Kenya examines why the initial entrepreneurial spirit and digital revolution has begun to falter despite support from motivated entrepreneurs, international investors, policy experts and others. Written by engaged scholars and professionals in the field, the book offers 15 eye-opening chapters and 14 one-on-one conversations with entrepreneurs and investors to ask why establishing ICT start-ups on a continental and global scale remains a challenge on the “Silicon Savannah”. The authors present evidence-based recommendations to help Kenya to continue producing globally impactful  ICT innovations that improve the lives of those still waiting on the side-lines, and to inspire other nations to do the same….(More)”

Make Democracy Great Again: Let’s Try Some ‘Design Thinking’


Ken Carbone in the Huffington Post: “Allow me to begin with the truth. I’ve never studied political science, run for public office nor held a position in government. For the last forty years I’ve led a design agency working with enduring brands across the globe. As with any experienced person in my profession, I have used research, deductive reasoning, logic and “design thinking“ to solve complex problems and create opportunities. Great brands that are showing their age turn to our agency to get back on course. In this light, I believe American democracy is a prime target for some retooling….

The present campaign cycle has left many voters wondering how such divisiveness and national embarrassment could be happening in the land of the free and home of the brave. This could be viewed as symptomatic of deeper structural problems in our tradition bound 240 year-old democracy. Great brands operate on a “innovate or die” model to insure success. The continual improvement of how a business operates and adapts to market conditions is a sound and critical practice.

Although the current election frenzy will soon be over, I want to examine three challenges to our election process and propose possible solutions for consideration. I’ll use the same diagnostic thinking I use with major corporations:

Term Limits…

Voting and Voter registration…

Political Campaigns…

In June of this year I attended the annual leadership conference of AIGA, the professional association for design, in Raleigh NC. A provocative question posed to a select group of designers was “What would you do if you were Secretary of Design.” The responses addressed issues concerning positive social change, education and Veteran Affairs. The audience was full of several hundred trained professionals whose everyday problem solving methods encourage divergent thinking to explore many solutions (possible or impossible) and then use convergent thinking to select and realize the best resolution. This is the very definition of “design thinking.” That leads to progress….(More)”.

We All Need Help: “Big Data” and the Mismeasure of Public Administration


Essay by Stephane Lavertu in Public Administration Review: “Rapid advances in our ability to collect, analyze, and disseminate information are transforming public administration. This “big data” revolution presents opportunities for improving the management of public programs, but it also entails some risks. In addition to potentially magnifying well-known problems with public sector performance management—particularly the problem of goal displacement—the widespread dissemination of administrative data and performance information increasingly enables external political actors to peer into and evaluate the administration of public programs. The latter trend is consequential because external actors may have little sense of the validity of performance metrics and little understanding of the policy priorities they capture. The author illustrates these potential problems using recent research on U.S. primary and secondary education and suggests that public administration scholars could help improve governance in the data-rich future by informing the development and dissemination of organizational report cards that better capture the value that public agencies deliver….(More)”.

Even in Era of Disillusionment, Many Around the World Say Ordinary Citizens Can Influence Government


Survey by Pew Global: “Signs of political discontent are increasingly common in many Western nations, with anti-establishment parties and candidates drawing significant attention and support across the European Union and in the United States. Meanwhile, as previous Pew Research Center surveys have shown, in emerging and developing economies there is widespread dissatisfaction with the way the political system is working.

As a new nine-country Pew Research Center survey on the strengths and limitations of civic engagement illustrates, there is a common perception that government is run for the benefit of the few, rather than the many in both emerging democracies and more mature democracies that have faced economic challenges in recent years. In eight of nine nations surveyed, more than half say government is run for the benefit of only a few groups in society, not for all people.1

However, this skeptical outlook on government does not mean people have given up on democracy or the ability of average citizens to have an impact on how the country is run. Roughly half or more in eight nations – Kenya, Nigeria, South Africa, the U.S., India, Greece, Italy and Poland – say ordinary citizens can have a lot of influence on government. Hungary, where 61% say there is little citizens can do, is the lone nation where pessimism clearly outweighs optimism on this front.

Many people in these nine nations say they could potentially be motivated to become politically engaged on a variety of issues, especially poor health care, poverty and poor-quality schools. When asked what types of issues could get them to take political action, such as contacting an elected official or taking part in a protest, poor health care is the top choice among the six issues tested in six of eight countries. Health care, poverty and education constitute the top three motivators in all nations except India and Poland….(More)

What We Should Mean When We Talk About Citizen Engagement


Eric Gordon in Governing: “…But here’s the problem: The institutional language of engagement has been defined by its measurement. Chief engagement officers in corporations are measuring milliseconds on web pages, and clicks on ads, and not relations among people. This is disproportionately influencing the values of democracy and the responsibility of public institutions to protect them.

Too often, when government talks about engagement, it is talking those things that are measurable, but it is providing mandates to employees imbued with ambiguity. For example, the executive order issued by Mayor Murray in Seattle is a bold directive for the “timely implementation by all City departments of equitable outreach and engagement practices that reaffirm the City’s commitment to inclusive participation.”

This extraordinary mayoral mandate reflects clear democratic values, but it lacks clarity of methods. It reflects a need to use digital technology to enhance process, but it doesn’t explain why. This in no way is meant as a criticism of Seattle’s effort; rather, it is simply meant to illustrate the complexity of engagement in practice. Departments are rewarded for quantifiable efficiency, not relationships. Just because something is called engagement, this fundamental truth won’t change.

Government needs to be much more clear about what it really means when it talks about engagement. In 2015, Living Cities and the Citi Foundation launched the City Accelerator on Public Engagement, which was an effort to source and support effective practices of public engagement in city government. This 18-month project, based on a cohort of five cities throughout the United States, is just now coming to an end. Out of it came several lasting insights, one of which I will share here. City governments are institutions in transition that need to ask why people should care.

After the election, who is going to care about government? How do you get people to care about the services that government provides? How do you get people to care about the health outcomes in their neighborhoods? How do you get people to care about ensuring accessible, high-quality public education?

I want to propose that when government talks about civic engagement, it is really talking about caring. When you care about something, you make a decision to be attentive to that thing. But “caring about” is one end of what I’ll call a spectrum of caring. On the other end, there is “caring for,” when, as described by philosopher Nel Noddings, “what we do depends not upon rules, or at least not wholly on rules — not upon a prior determination of what is fair or equitable — but upon a constellation of conditions that is viewed through both the eyes of the one-caring and the eyes of the cared-for.”

In short, caring-for is relational. When one cares for another, the outcomes of an encounter are not predetermined, but arise through relation….(More)”.

A Practical Guide for Harnessing the Power of Data


How does it do that? In a word: data.

Using a series of surveys and evaluations, Repair learned that once people participate in two volunteer opportunities, they’re more likely to continue volunteering regularly. Repair has used that and other findings to inform its operations and strategy, and to accelerate its work to encourage individuals to make an enduring commitment to public service.

Many purpose-driven organizations like Repair the World are committing more brainpower, time, and money to gathering data, and nonprofit and foundation professionals alike are recognizing the importance of that effort.

And yet there is a difference between just having data and using it well. Recent surveys have found that 94 percent of nonprofit professionals felt they were not using data effectively, and that 75 percent of foundation professionals felt that evaluations conducted by and submitted to grant makers did not provide any meaningful insights.

To remedy this, the Charles and Lynn Schusterman Family Foundation (one of Repair the World’s donors) developed the Data Playbook, a new tool to help more organizations harness the power of data to make smarter decisions, gain insights, and accelerate progress….

In the purpose-driven sector, our work is critically important for shaping lives and strengthening communities. Now is the time for all of us to commit to using the data at our fingertips to advance the broad range of causes we work on — education, health care, leadership development, social-justice work, and much more…

We are in this together. Let’s get started. (More)”

The case against democracy


 in the New Yorker: “Roughly a third of American voters think that the Marxist slogan “From each according to his ability to each according to his need” appears in the Constitution. About as many are incapable of naming even one of the three branches of the United States government. Fewer than a quarter know who their senators are, and only half are aware that their state has two of them.

Democracy is other people, and the ignorance of the many has long galled the few, especially the few who consider themselves intellectuals. Plato, one of the earliest to see democracy as a problem, saw its typical citizen as shiftless and flighty:

Sometimes he drinks heavily while listening to the flute; at other times, he drinks only water and is on a diet; sometimes he goes in for physical training; at other times, he’s idle and neglects everything; and sometimes he even occupies himself with what he takes to be philosophy.

It would be much safer, Plato thought, to entrust power to carefully educated guardians. To keep their minds pure of distractions—such as family, money, and the inherent pleasures of naughtiness—he proposed housing them in a eugenically supervised free-love compound where they could be taught to fear the touch of gold and prevented from reading any literature in which the characters have speaking parts, which might lead them to forget themselves. The scheme was so byzantine and cockamamie that many suspect Plato couldn’t have been serious; Hobbes, for one, called the idea “useless.”

A more practical suggestion came from J. S. Mill, in the nineteenth century: give extra votes to citizens with university degrees or intellectually demanding jobs. (In fact, in Mill’s day, select universities had had their own constituencies for centuries, allowing someone with a degree from, say, Oxford to vote both in his university constituency and wherever he lived. The system wasn’t abolished until 1950.) Mill’s larger project—at a time when no more than nine per cent of British adults could vote—was for the franchise to expand and to include women. But he worried that new voters would lack knowledge and judgment, and fixed on supplementary votes as a defense against ignorance.

In the United States, élites who feared the ignorance of poor immigrants tried to restrict ballots. In 1855, Connecticut introduced the first literacy test for American voters. Although a New York Democrat protested, in 1868, that “if a man is ignorant, he needs the ballot for his protection all the more,” in the next half century the tests spread to almost all parts of the country. They helped racists in the South circumvent the Fifteenth Amendment and disenfranchise blacks, and even in immigrant-rich New York a 1921 law required new voters to take a test if they couldn’t prove that they had an eighth-grade education. About fifteen per cent flunked. Voter literacy tests weren’t permanently outlawed by Congress until 1975, years after the civil-rights movement had discredited them.

Worry about voters’ intelligence lingers, however. …In a new book, “Against Democracy” (Princeton), Jason Brennan, a political philosopher at Georgetown, has turned Estlund’s hedging inside out to create an uninhibited argument for epistocracy. Against Estlund’s claim that universal suffrage is the default, Brennan argues that it’s entirely justifiable to limit the political power that the irrational, the ignorant, and the incompetent have over others. To counter Estlund’s concern for fairness, Brennan asserts that the public’s welfare is more important than anyone’s hurt feelings; after all, he writes, few would consider it unfair to disqualify jurors who are morally or cognitively incompetent. As for Estlund’s worry about demographic bias, Brennan waves it off. Empirical research shows that people rarely vote for their narrow self-interest; seniors favor Social Security no more strongly than the young do. Brennan suggests that since voters in an epistocracy would be more enlightened about crime and policing, “excluding the bottom 80 percent of white voters from voting might be just what poor blacks need.”…(More)”