➔ Phenomenal World

December 2nd, 2017

➔ Phenomenal World



The gray box of XAI

A recent longform piece in the New York Times identifies the problem of explaining artificial intelligence. The stakes are high because of the European Union’s controversial and unclear “right-to-explanation” law, which will become active in May 2018.

“Instead of certainty and cause, A.I. works off probability and correlation. And yet A.I. must nonetheless conform to the society we’ve built — one in which decisions require explanations, whether in a court of law, in the way a business is run or in the advice our doctors give us. The disconnect between how we make decisions and how machines make them, and the fact that machines are making more and more decisions for us, has birthed a new push for transparency and a field of research called explainable A.I., or X.A.I. Its goal is to make machines able to account for the things they learn, in ways that we can understand. But that goal, of course, raises the fundamental question of whether the world a machine sees can be made to match our own.”

Full article by CLIFF KUANG here. This page provides a short overview of DARPA's XAI (Explainable Artificial Intelligence) program.

An interdisciplinary group addresses the problem:

"Contrary to popular wisdom of AI systems as indecipherable black boxes, we find that this level of explanation should often be technically feasible but may sometimes be practically onerous—there are certain aspects of explanation that may be simple for humans to provide but challenging for AI systems, and vice versa. As an interdisciplinary team of legal scholars, computer scientists, and cognitive scientists, we recommend that for the present, AI systems can and should be held to a similar standard of explanation as humans currently are; in the future we may wish to hold an AI to a different standard."

Full article by FINALE DOSHI-VELEZ et al. here. ht Margarita For the layperson, the most interesting part of the article may be its general overview of societal norms around explanation and explanation in the law.

Michael comments: Human cognitive systems have generated similar questions in vastly different contexts. The problem of chick-sexing (see Part 3) gave rise to a mini-literature within epistemology.

From Michael S. Moore’s book Law and Society: Rethinking the Relationship: “A full explanation in terms of reasons for action requires two premises: the major premise, specifying the agent’s desires (goals, objectives, moral beliefs, purposes, aims, wants, etc.), and the minor premise, specifying the agent’s factual beliefs about the situation he is in and his ability to achieve, through some particular action, the object of his desires.” Link. ht Margarita

  • A Medium post with an illustrated summary of some XAI techniques. Link.
⤷ Full Article

November 18th, 2017

Duchamp Wanted


How to build justice into algorithmic actuarial tools

Key notions of fairness contradict each other—something of an Arrow’s Theorem for criminal justice applications of machine learning.

"Recent discussion in the public sphere about algorithmic classification has involved tension between competing notions of what it means for a probabilistic classification to be fair to different groups. We formalize three fairness conditions that lie at the heart of these debates, and we prove that except in highly constrained special cases, there is no method that can satisfy these three conditions simultaneously. Moreover, even satisfying all three conditions approximately requires that the data lie in an approximate version of one of the constrained special cases identified by our theorem. These results suggest some of the ways in which key notions of fairness are incompatible with each other, and hence provide a framework for thinking about the trade-offs between them."

Full paper from JON KLEINBERG, SENDHIL MULLAINATHAN and MANISH RAGHAVAN here. h/t research fellow Sara, who recently presented on bias in humans, courts, and machine learning algorithms, and who was the source for all the papers in this section.

In a Twitter thread, ARVIND NARAYANAN describes the issue in more casual terms.

"Today in Fairness in Machine Learning class: a comparison of 21 (!) definitions of bias and fairness [...] In CS we're used to the idea that to make progress on a research problem as a community, we should first all agree on a definition. So 21 definitions feels like a sign of failure. Perhaps most of them are trivial variants? Surely there/s one that's 'better' than the rest? The answer is no! Each defn (stat. parity, FPR balance, contextual fairness in RL...) captures something about our fairness intuitions."

Link to Narayanan’s thread.

Jay comments: Kleinberg et al. describe their result as choosing between conceptions of fairness. It’s not obvious, though, that this is the correct description. The criteria (calibration and balance) discussed aren’t really conceptions of fairness; rather, they’re (putative) tests of fairness. Particular questions about these tests aside, we might have a broader worry: if fairness is not an extensional property that depends upon, and only upon, the eventual judgments rendered by a predictive process, exclusive of the procedures that led to those judgments, then no extensional test will capture fairness, even if this notion is entirely unambiguous and determinate. It’s worth consideringNozick’s objection to “pattern theories” of justice for comparison, and (procedural) due process requirements in US law.

⤷ Full Article

June 30th, 2020



Brazil's Bolsa Familia is widely credited with lifting more than 20 million people out of extreme poverty, making it a global model for anti-poverty initiatives. Developed as part of a broader theory of equitable development, it serves as the basis for ongoing efforts to expand the social welfare system for the country’s poor and working class.

In a 2017 book, economist LENA LAVINAS takes a critical approach to Brazilian social policy. Examining the relationship between social policy and financial markets, Levinas argues that, despite its successes, the strategy of "social developmentalism" in Brazil unwittingly entrenched both unequal growth and the stagnation of social protection.

From the book:

"The twenty-first century seemed poised to pluck Brazil from its history of underdevelopment. After suffering through two decades (1980–2003) of low growth and considerable macroeconomic instability, Brazil—in step with the rest of Latin America—was ready to begin a series of rosy years. In the new developmental strategy, the missing link on the way to social cohesion, so the argument went, would emerge with the advent of mass consumption. In Brazil, as in the rest of Latin America, the core impediment to the expansion of a mass consumption society resided (above all else) in the absence of mechanisms for boosting consumption in the context of low productivity and the persistent oversupply of labor.

Performance in terms of the provision of public facilities has not tracked remotely close to the vitality of the market. It does, however, reveal welfare inequities that the market obscures. Through this prism, the upward social mobility observed in Brazil in the years spanning 2003–2014 failed to even come close to promoting a true expansion of the country’s middle classes. Social policy served as collateral to access financial markets through credit. In Brazil, the market has universalized access to color TVs and fridges among those in the lowest income quintile. Treated water, however, to say nothing of adequate sanitation, remains a luxury, the province of few."

Link to the publisher's page.

  • A 2018 by Lavinas details one of the book's arguments—"the collateralization of social policy." Link. And a 2013 paper by Lavinas examines the broad adoption of conditional cash transfer schemes throughout Latin America. Link.
  • In a 2014 paper, Michael McCarthy examines union attempts to control pension fund investment. Link. Another paper by Natascha van der Zwan on the financial politics of occupational pensions. Link. See also: McCarthy's book Dismantling Solidarity, on these same themes. Link.
  • Marie Gottschalk's book The Shadow Welfare State examines the American "private-sector safety net." Link. See also: Frank R. Dobbin's 1992 paper "The Origins of Private Social Insurance: Public Policy and Fringe Benefits in America, 1920-1950." Link.
⤷ Full Article

November 11th, 2017

The Hülsenbeck Children


Recommender systems power YouTube's controversial kids' videos

Familiar cartoon characters are placed in bizarre scenarios, sometimes by human content creators, sometimes by automated systems, for the purpose of attracting views and ad money. First, from the New York Times:

“But the app [YouTube Kids] contains dark corners, too, as videos that are disturbing for children slip past its filters, either by mistake or because bad actors have found ways to fool the YouTube Kids algorithms.

“In recent months, parents like Ms. Burns have complained that their children have been shown videos with well-known characters in violent or lewd situations and other clips with disturbing imagery, sometimes set to nursery rhymes. Many have taken to Facebook to warn others, and share video screenshots showing moments ranging from a Claymation Spider-Man urinating on Elsa of ‘Frozen’ to Nick Jr. characters in a strip club.”

Full piece by SAPNA MAHESHWARI in the Times here.

On Medium, JAMES BRIDLE expands on the topic, and criticizes the structure of YouTube itself for incentivizing these kinds of videos, many of which have millions of views.

“These videos, wherever they are made, however they come to be made, and whatever their conscious intention (i.e. to accumulate ad revenue) are feeding upon a system which was consciously intended to show videos to children for profit. The unconsciously-generated, emergent outcomes of that are all over the place.

“While it is tempting to dismiss the wilder examples as trolling, of which a significant number certainly are, that fails to account for the sheer volume of content weighted in a particularly grotesque direction. It presents many and complexly entangled dangers, including that, just as with the increasing focus on alleged Russian interference in social media, such events will be used as justification for increased control over the internet, increasing censorship, and so on.”

Link to Bridle’s piece here.

⤷ Full Article

November 4th, 2017



Sociologist Zeynep Tufekci engages with Adam Mosseri, who runs the Facebook News Feed

Tufekci: “…Facebook does not ask people what they want, in the moment or any other way. It sets up structures, incentives, metrics & runs with it.”

Mosseri: “We actually ask 10s of thousands of people a day how much they want to see specific stories in the News Feed, in addition to other things.”

Tufekci: “That’s not asking your users, that’s research on your product. Imagine a Facebook whose customers are users—you’d do so much differently. I mean asking all people, in deliberate fashion, with sensible defaults—there are always defaults—even giving them choices they can change…Think of the targeting offered to advertisers—with support to make them more effective—and flip the possibilities, with users as customers. The users are offered very little in comparison. The metrics are mostly momentary and implicit. That’s a recipe to play to impulse.”

The tweets are originally from Zeynep Tufekci in response to Benedict Evans (link), but the conversation is much easier to read in Hamza Shaban’s screenshots here.

See the end of this newsletter for an extended comment from Jay.

  • On looping effects (paywall): “This chapter argues that today's understanding of causal processes in human affairs relies crucially on concepts of ‘human kinds’ which are a product of the modern social sciences, with their concern for classification, quantification, and intervention. Child abuse, homosexuality, teenage pregnancy, and multiple personality are examples of such recently established human kinds. What distinguishes human kinds from ‘natural kinds’, is that they have specific ‘looping effects’. By coming into existence through social scientists' classifications, human kinds change the people thus classified.” Link. ht Jay


Mechanisms and causes between micro and macro

Daniel Little, the philosopher of social science behind Understanding Society, haswritten numerous posts on the topic. Begin with this one from 2014:

“It is fairly well accepted that there are social mechanisms underlying various patterns of the social world — free-rider problems, communications networks, etc. But the examples that come readily to mind are generally specified at the level of individuals. The new institutionalists, for example, describe numerous social mechanisms that explain social outcomes; but these mechanisms typically have to do with the actions that purposive individuals take within a given set of rules and incentives.

“The question here is whether we can also make sense of the notion of a mechanism that takes place at the social level. Are there meso-level social mechanisms? (As always, it is acknowledged that social stuff depends on the actions of the actors.)”

In the post, Little defines a causal mechanism and a meso-level mechanism, then offers example research.

“…It is possible to identify a raft of social explanations in sociology that represent causal assertions of social mechanisms linking one meso-level condition to another. Here are a few examples:

  • Al Young: decreasing social isolation causes rising inter-group hostility (link)
  • Michael Mann: the presence of paramilitary organizations makes fascist mobilization more likely (link)
  • Robert Sampson: features of neighborhoods influence crime rates (link)
  • Chuck Tilly: the availability of trust networks makes political mobilization more likely (link)
  • Robert Brenner: the divided sovereignty system of French feudalism impeded agricultural modernization (link)
  • Charles Perrow: legislative control of regulatory agencies causes poor enforcement performance (link)

More of Little’s posts on the topic are here. ht Steve Randy Waldman

⤷ Full Article

June 24th, 2020

Running Horse


In her 2007 book, Against the Law: Labor Protests in China's Rustbelt and Sunbelt, sociologist CHING KWAN LEE paints an intricate portrait of the two segments of the Chinese working class that have most acutely experienced the country's changing political economy: laid-off and retired workers in China’s industrial rustbelt, and young migrant factory workers in the export-oriented sunbelt.

From the preface:

"Although unemployment and exploitation can be found in many places and at different times, peculiarities of China’s postsocialist conditions have engendered features of labor politics that defy conventional categorization. First, the law, fledgling legal institutions, and the rhetoric of legal rights are central to labor protests throughout China, even though very few workers actually believe in the effectiveness of the regime’s ideology of law-based government. Second, leading to the formation of neither a national labor movement nor representative organizations, the several thousand worker protests that erupted every year throughout the 1990s took the prevailing form of localized, workplace-based cellular activism. With workers blocking traffic in the streets, lying on railroads, or staging sit ins in front of government buildings, these demonstrations presented a palpable threat to social stability, at least in the eyes of the national leadership. What must be emphasized, however, is that workers’ cellular activism has thus far rarely escalated into large-scale, coordinated, cross-regional unrest.

What, then, is the nature of working-class agitation in this period of marketization and globalization? Above all, I have found that the communist regime’s strategy of accumulation, in the form of what I term 'decentralized legal authoritarianism,' both generates the impetus for and places limits on working-class protests in this period of market reform. This larger political economic context of reform shapes not only collective mobilization by workers but also popular rebellion in general, and therefore is a key to understanding the institutional foundations of China’s economic dynamism and sociopolitical tensions."

Link to the publisher's page.

  • "Labour strikes in China are always launched by unorganized workers rather than by trade unions." Feng Chen on China's quadripartite wage setting system. Link.
  • "This chapter investigates the role of social networks during China's most dynamic period of urban protest (1919–1927) in Shanghai." A 2007 book chapter by Elizabeth Perry. Link. See also: Perry's groundbreaking 1993 book on Chinese labor politics in the early 20th century, and her 1980 analysis of peasant rebellions in Huaipei from 1845–1945. Link and link. ht Julian G.
  • Meg Rithmire reviews regional approaches to Chinese political economy, asking: "How have local governments differently interpreted and implemented national reform policies? What explains different decision-making regarding investments and growth strategies? And how have different local growth strategies beget different socioeconomic consequences?" Link.
⤷ Full Article

June 15th, 2020

Green Stripe


As debate and discussion continues over reforms to US policing, attention has been drawn to the share of municipal and state budgets dedicated to police departments. While a useful proxy of governmental priorities, these budgets only tell part of the complex story of the role and function of police in society.

In a their 2008 book chapter titled "The Enforcement-Equality Trade Off," ARJUN JAYADEV and SAMUEL elaborate the role of what they term "guard labor"—the labor units "devoted to the maintenance of order."

From the chapter:

"In order to maintain order, all societies allocate resources to defence, policing, surveillance, contractual monitoring and other activities that sustain the property rights and other claims that characterise status quo institutions. Data from the United States indicate a significant increase in its extent in the USA over the period 1890 to the present. Cross-national comparisons show a significant statistical association between income inequality and the fraction of the labour force that is constituted by guard labour, as well as with measures of political legitimacy (inversely) and political conflict.

Continental European welfare states devote considerably less resources to the maintenance of order than do the English-speaking economies. A possible explanation is that these economies divert fewer resources from directly productive uses to guard labour by undertaking larger transfers of claims on resources in the form of social expenditures and higher wages."

Link to the report.

  • For more on guard labor, link to a 2019 newsletter, in we shared Jayadev's classic 2006 paper with Samuel Bowles on guard labor. Also shared in that letter, a 2014 Times op-ed by Bowles and Jayadev on the subject, with an unbeatable infographic comparing the US' share of guard labor to other rich nations.
  • See Jayadev's paper "Estimating Guard Labor" for more on the employment statistics behind their analysis. Link. And link to a 2018 blog post on police and prison spending in the US and Europe. Link.
  • For more on the relationship between the labor market and policing and prisons, see this recent paper by Seth Prins and Adam Reich. Link. See also a 2002 paper by Eric Gould et al on crime rates and labor market opportunity from 1979-1997. Link.
  • "Inequality and Guard Labor, or Prohibition and Guard Labor?" by Vincent Geloso and Vadim Kufenko. Link.
⤷ Full Article

June 8th, 2020

The Burning


As commentators and policymakers have scrambled to find explanations for and responses to the unprecedented uprisings against police brutality across the United States, interest in the role of police unions in local politics has soared. Recent research into the question joins a decades-long debate in the labor movement over the distinctive character of police associations—not only as regards their power relative to the public, but also their political strength relative to the rest of the public sector.

A 2017 research paper by CATHERINE FISK and L. SONG RICHARDSON examines the evolution of US police unions, analyzes their impact on policymaking, and evaluates the efforts of cities to reform police departments over the past fifty years.

From the piece:

"Police officers formed local unions in various cities in the 1940s, and some police unions affiliated with national labor federations. However, well into the 1960s, police departments routinely fired officers who attempted to unionize, and courts upheld the power of cities to ban officers from joining unions. In the absence of legal rights to unionize or bargain collectively, government employee unions became adept at securing their members’ interests through political activity and negotiating informal agreements with public officials. Unions succeeded in gaining a lasting foothold in American police departments in the late 1960s. Not surprisingly, they negotiated for contractual protections against discipline and lobbied legislators to incorporate these protections in legislation. They opposed constitutional criminal procedure restrictions on police conduct and sought to block civilian oversight of police discipline. The legacy of the 1960s is collective bargaining agreements which make it difficult to investigate and punish officers to this day."

Link to the report.

  • "Cities which have low levels of police protections are also less likely to experience police abuse. Local-level politics does not have a salient effect on the level of police protections, but state labour laws have a significant impact on the level of protections which officers receive." Findings from a novel police protection index drawing on data from the US's 100 largest cities. Link. And a 2008 paper by Samuel Walker looks at, among other things, the relationship between the civil rights movement and the growth of police unions. Link.
  • Analyzing the consequences of a 2003 Florida Supreme Court decision which increased unionization among sheriffs' deputies, Dhammika Dharmapala, Richard McAdams, and John Rappaport find that "collective bargaining rights led to a substantial increase in violent incidents." Link.
  • A recent paper by Michael Zoorob looks at the electoral impact of the Fraternal Order of Police. Link.
  • "Until 1919, the AFL refused to charter police unions. The 1897 AFL convention rejected an application from a police group in Cleveland, explaining that 'it is not within the province of the trade union movement to organize policemen, no more than to organize militiamen, as both are often controlled by forces inimical to the labor movement.'" Joseph Slater's 2004 book recounts the tensions between police and the early American labor movement. Link.
⤷ Full Article

June 2nd, 2020

Clouds, Sun, Sea


This week has seen policymakers, scholars, and the public debate the meaning of collective violence. While political and media discourse often fails to examine the long-term effects of social unrest, a vast literature grapples with the mechanisms that link protests and uprisings with institutional change.

A 1978 book by JAMES W. BUTTON integrates a vast amount of interviews, archival sources, and statistical data to analyze the public response to the US urban uprisings of the 1960s. Focusing the analysis on three federal agencies—the (now-dismantled) Office of Economic Opportunity, HUD, and the DOJ—the book suggests that the 1960s riots were understood by policymakers as political demands.

From the introduction:

"Although domestic collective violence has played a prominent role in American history, few other episodes of urban violence in this country's history have been as dramatic as the black riots of the 1960s. As a result, the causes, precipitating events, and participants of the outbursts have been thoroughly studied over the past several years. Yet what is remarkable about this extensive analysis is the almost complete neglect of the political effects or consequences of these pervasive disorders. By concentrating instead on the factors that may have caused the riots, most investigators have implicitly reflected a normative bias concerning the disutility of domestic violence for affecting social and political change.

The fundamental purpose of this study is to evaluate some of the political consequences of the urban black riots of the 1960s and early 1970s, focusing on responses of the executive branch of the federal government. In fulfilling such a task, it asks: did the riots affect executive officials' decisions and ultimately federal public policy? Did the federal executive branch respond differently to the initial, less intense riots (1963-1966) than it did to the later, more severe disorders (1967-1968) and, if so, why? how have national executive responses to urban rioting been affected by the local political and environmental context and by local reactions to such violence? And how do public officials tend to view the role of violence in American society?"

Link to the book page.

  • A new article by Omar Wasow examines the relationship between violent and nonviolent protests, media, public opinion, and policy alignment from the Civil Rights Era, and in particular on Nixon's election in 1968. Link. And a 2018 paper by Shom Mazumder looks at the persistent effects of Civil Rights protects on political attitudes. Link.
  • A 1978 paper by sociologist Charles Tilly on collective violence: "Historically, collective violence has flowed regularly out of the central political processes of western countries. People seeking to seize, hold, or realign the levers of power have continually engaged in collective violence as part of their struggles." Link.
  • In a 2007 article, historian Michael Kazin asks: "Many of the conditions thought to have precipitated the eruption of civil violence in the 1960s either persist or have grown worse. What accounts for the absence of civil violence on American streets?" Link. And a new book by Mike Davis and Jon Wiener looks at the 1960s in Los Angeles. Link.
⤷ Full Article

May 26th, 2020



Analyses of variation in state-level responses to the coronavirus tend to focus on party determination: On the whole, states led by Democrats have been found to undertake more rapid and extensive responses to the crisis. The focus on immediate political factors, however, masks the broader history of America's uneven and disaggregated bureaucratic capacity.

A 1982 book by STEPHEN SKOWRONEK presents one of the most comprehensive accounts of the origins of the US administrative state. Focusing on reforms in civil administration, the army, and national railroad regulation from 1870-1920, the book demonstrates how regional differences contributed to the particular character of American state development.

"Unravelling the state-building problem in modern American political development places the apparent statelessness of early America in a new light. The governmental forms and procedures necessary for securing order in industrial America emerged through a labored exercise in creative destruction. Modernization of national administrative controls did not entail making the established state more efficient; it entailed building a qualitatively different kind of state.

The Civil War brought national military conscription, a national welfare agency for former slaves, a national income tax, national monetary controls, and citizenship. Yet, this was a state grounded in only half the nation. As the South returned, national electoral politics changed, and these institutional achievements began to be undone. Here, then, was a state only in the sense of the word imputed to it by the interests and strategies of the mass electoral organizations controlling its offices. No institution stood beyond the reach of party concerns. The fate of the wartime governmental apparatus suggests that if new institutional forms are to constitute a new state, they must alter the procedural bonds that tie governmental institutions together and define their relationship to society."

Link to the publisher's page.

  • Theda Skocpol and Kenneth Finegold expand Skowronek's research into the New Deal era. Link.
  • "In societies where social status is a cleavage, elites can use the threat of desegregation to unite wealthy and poor members of high-status groups against taxation and the bureaucratic capacity required to collect taxes." Pavithra Suryanarayan and Steven White on "Slavery, Reconstruction, and Bureaucratic Capacity in the American South." Link. In another article, Roberto Stefan Foa and Anna Nemirovskaya analyze the development of state capacity on the frontier. Link.
  • Daniel Berliner, Anne Greenleaf, Milli Lake, and Jennifer Noveck present "systematic study of relationship between state capacity and labor rights." Link.
⤷ Full Article