Sunday, June 24, 2018

Visiting Hangzhou

I'm on my way to Hangzhou, via Shanghai, for a meeting called by Alibaba.  I'll only be there for two busy days.
Read all about it:

问大家!和诺奖获得者面对面聊人类未来你来不来?


Here are some others who are coming
1.Patrick Bolton (帕特里克·博尔顿):哥伦比亚大学教授,美国艺术与科学院院士、英国科学院院士、美国金融学会院士,当代最著名的金融学家之一。
2、Markus Brunnermeier(马科斯·布鲁纳米尔):普林斯顿大学Bendheim金融中心主任,美国国家经济研究中心和经济政策研究中心成员,撰写有大量关于金融危机和货币政策的著作。
3、Bengt Holmstrom (本特·霍姆斯特罗姆):麻省理工学院教授,美国人文与科学院院士,契约理论的奠基人之一,2016年获得诺贝尔经济学奖。
4、Josh Lerner (杰斯·勒那):哈佛商学院教授,100位对私募行业最有影响力的学者之一,研究主要集中在风险投资机构的结构和作用。
5、Preston McAfee (普斯顿·麦法宜):前谷歌战略技术主管、微软首席经济学家,重视根据微观经济学基础,提出商业战略和竞争解决办法。
6、Christopher A. Pissarides(克里斯托弗·皮萨里德斯):伦敦经济学院教授,英国社会科学院院士,2010年因在市场搜寻理论和宏观经济方面的突出贡献获得诺贝尔经济学奖。
7、Yingyi Qian(钱颖一):清华大学经济与管理研究院院长,制度经济学的代表人物,2012年当选为世界计量经济学会会士、2018年当选为清华大学首批文科资深教授。
8、Alvin E. Roth(阿尔文·罗思):哈佛商学院教授,美国文理科学院院士,美国经济学会主席,在博弈论、市场设计与实验经济学取得卓越贡献,2012年获得诺贝尔经济学奖。
9、Thomas Sargent(托马斯·萨金特):纽约大学教授,理性预期学派的领袖人物,擅长总体经济学、货币经济学、时间序列等领域,2011年获诺贝尔经济学奖。
10、Michael Spence (迈克尔·斯宾塞):美国社会科学院院士,提出的信息不对称理论为广泛的市场理论奠定了基础,形成了现代信息经济学的核心,2001年获得诺贝尔经济学奖。
11、Steve Tadelis (史蒂文·泰迪里斯):加州大学伯克利分校哈斯商学院教授,研究致力于理解企业和市场的契约机制如何影响价值创造,著有《博弈论导论》等。
12、Neng Wang(王能):美国哥伦比亚大学商学院教授、上海财经大学金融学院院长,研究领域包括公司金融学,宏观经济学,资产定价理论,私募及对冲基金等。
13、Shangjin Wei(魏尚进):美国哥伦比亚商学院金融学经济学教授,美国国民经济研究局中国经济研究组主任。主要研究领域为国际金融、国际贸易、政府治理和改革、中国经济以及宏观经济学。
14.、Wei Xiong (熊伟):普林斯顿大学教授,香港中文大学(深圳)经管学院院长,国际顶级期刊《Journal of Finance》联合主编,研究领域包括投机泡沫、金融危机、行为金融学等。
15、Chengang Xu (许成钢):长江商学院教授、香港大学经济学教授,中国政府治理研究的基石,2013年获孙冶方经济科学奖。

Saturday, June 23, 2018

Who accepts which deceased donor kidneys? Agarwal, Ashlagi, Somaini and Waldinger in the AEA P&P


Dynamic Incentives in Wait List Mechanisms






Download Full Text PDF 






Friday, June 22, 2018

Tuneups for kidney exchange by Agarwal, Aslahi, Azevedo, Featherstone and Karaduman

From the May 2018, first issue of the AEA (not R) Papers and Proceedings:
What Matters for the Productivity of Kidney Exchange?
Nikhil Agarwal
Itai Ashlagi
Eduardo Azevedo
Clayton Featherstone
Ömer Karaduman


Download Full Text PDF 


Thursday, June 21, 2018

Repugnance watch: referendum proposes to end (some) legal prostitution in Nevada

The November ballot will have a referendum on prostitution in two Nevada counties where it is currently legal. The WSJ has the story:

Is the Party Over for Nevada’s Legal Brothels? Possibility of a Ban Looms
Referendum measures, if passed, would outlaw nine bordellos in two Nevada counties as women’s advocates take on working conditions for prostitutes

"A coalition of women’s advocacy groups in Nevada has undertaken referendum initiatives against legal prostitution in Lyon and Nye counties, opposite corners of the state that are home to about half of the state’s brothels.
...
"Nevada began legalizing prostitution on a county basis in 1971, the only state to have done so. Seven of the state’s 16 counties have licensed bordellos; prostitution remains illegal in populous Clark County including Las Vegas, and Washoe County including Reno.

"Women’s activists said they began seeking the bans last year after hearing about prostitutes being assaulted at the brothels. “These are all areas of sexual harassment and assault within the workplace, which is why this is a part of the #MeToo movement,” said Melissa Holland, executive director of Awaken, a Reno women’s nonprofit that teamed with other groups to push for referendums.
...
"Frederick Fabian, manager at the Desert Club brothel in Battle Mountain, Nev., which hasn’t been targeted for a ban, said illegal prostitution is a dangerous profession while legalized prostitution isn’t. “The cathouses and brothels are regulated by the county sheriff and the health facilities,” he said."

Wednesday, June 20, 2018

Statement on Harmful Consequences of Separating Families at the U.S. Border, by the U.S. National Academies of Sciences, Engineering and Medicine

The National Academies of Sciences, Engineering and Medicine issued the following statement today: 
Statement on Harmful Consequences of Separating Families at the U.S. Border
We urge the U.S. Department of Homeland Security to immediately stop separating migrant children from their families, based on the body of scientific evidence that underscores the potential for lifelong, harmful consequences for these children and based on human rights considerations.

Reports from the National Academies of Sciences, Engineering, and Medicine contain an extensive body of evidence on the factors that affect the welfare of children – evidence that points to the danger of current immigration enforcement actions that separate children from their parents. Research indicates that these family separations jeopardize the short- and long-term health and well-being of the children involved. In addition, the Committee on Human Rights of the National Academies, which has a long history of addressing issues at the intersection of human rights, science, and health, stresses that the practice of separating parents from their children at the border is inconsistent with U.S. obligations under the International Covenant on Civil and Political Rights.

Parents’ impact on their children’s well-being may never be greater than during the earliest years of life, when a child’s brain is developing rapidly and when nearly all of her or his experiences are shaped by parents and the family environment (NASEM, 2016, p. 1).  Young children who are separated from their primary caregivers may potentially suffer mental health disorders and other adverse outcomes over the course of their lives (NASEM, 2016, p. 21-22). Child development involves complex interactions among genetic, biological, psychological, and social processes (NRC and IOM, 2009, p. 74), and a disruption in any of these – such as family disruption – hinders healthy development and increases the risk for future disorders (NRC and IOM, 2009, p.102-104).  Young children are capable of deep and lasting sadness, grief, and disorganization in response to trauma and loss (NRC and IOM, 2000, p. 387).  Indeed, most mental, emotional, and behavioral disorders have their roots in childhood and adolescence (NRC and IOM, 2009, p. 1), and childhood trauma has emerged as a strong risk factor for later suicidal behavior (IOM, 2002, p. 3).  

Decades of research have demonstrated that the parent-child relationship and the family environment are at the foundation of children’s well-being and healthy development. We call upon the Department of Homeland Security to stop family separations immediately based on this evidence.

Marcia McNutt
President, National Academy of Sciences

C. D. Mote, Jr.
President, National Academy of Engineering

Victor J. Dzau

President, National Academy of Medicine

Canada legalizes recreational marijuana

First, a thoughtful tweet from the Prime Minister, about markets and black markets:
Here's the story from the CBC:

Senate passes pot bill, paving way for legal cannabis in 8 to 12 weeks
Federal government's bill legalizing recreational cannabis passes 52-29

"Now that the bill has passed, it's up to Prime Minister Justin Trudeau and his cabinet to choose the actual date when the legalization of recreational marijuana becomes law of the land. Bill C-45 comes with a provisional buffer period of eight to 12 weeks to give provinces time to prepare for sales of recreational marijuana."


And here's the NY Times:
Vote in Canada Paves the Way for Legalization of Marijuana

"Canada’s Senate passed legislation on Tuesday that will make Canada the first country with a major economy to legalize recreational marijuana use.

"The bill, which was approved by the House of Commons on Monday, goes next to the governor-general, the representative of Queen Elizabeth, as a formality. Once it is formally approved, the legislation is expected to create a multibillion dollar industry, with Canada joining Uruguay as the only countries to allow their citizens on a national level to use marijuana without fear of arrest."

Best paper award to Mohammad Akbarpour and Shengwu Li for "Credible Mechanisms"

Good news from ec18 now going on at Cornell.

The Best Paper Award goes to "Credible Mechanisms" by Mohammad Akbarpour and Shengwu Li 

Here is the paper: Credible Mechanisms

Abstract: Consider an extensive-form mechanism, run by an auctioneer who communicates sequentially and privately with agents.  Suppose the auctioneer can deviate from the rules provided that no single agent detects the deviation.  A mechanism is credible if it is incentive-compatible for the auctioneer to follow the rules. We study the optimal auctions in which only winners pay, under symmetric independent private values.The  first-price  auction  is  the  unique  credible  static  mechanism.   The  ascending auction is the unique credible strategy-proof mechanism.

And here are their websites
Mohammad Akbarpour
Shengwu Li

Tuesday, June 19, 2018

Why high incentives might require muscular informed consent (Ambuehl, Ockenfels and Stewart)

Here's a new paper (or at least just recently on the web) showing that subjects who are enticed by high payments might be disproportionately those who have difficulty gathering information about the risks...i.e.. these potential participants respond more to high payoffs than those who might have been able to gather information easily (and might have participated for a lower payment or been deterred even despite a high payment).

Attention and Selection Effects

CESifo Working Paper No. 7091 (Mai 2018)
Primary CESifo Category: [13] Behavioural Economics 
Abstract:
Who participates in transactions when information about the consequences must be learned? We show theoretically that decision makers for whom acquiring and processing information is more costly respond more strongly to changes in incentive payments for participating and decide to participate based on worse information. With higher payments, the pool of participants thus consists of a larger proportion of individuals who have a worse understanding of the consequences of their decision. We conduct a behavioral experiment that confirms these predictions, both for experimental variation in the costs of information acquisition and for various measures of information costs, including school grades and cognitive ability. These findings are relevant for any transaction combining a payment for participation with uncertain yet learnable consequences.

Monday, June 18, 2018

Alex Teytelboym on Love Island

Alex T. writes to let me know that he has been explaining British reality TV: a show where beautiful people couple up and couple on camera, apparently.

 The economics of Love Island: how game theory explains 'coupling up'
by  

His concluding paragraphs:

"So how does economic theory help contestants win Love Island? It’s very tempting to dump your partner and couple up with someone else, especially when there are eight weeks to go. But leaving your partner for someone else once tells others you are bad egg who cannot be trusted. As the series nears its end, the rewards from commitment are very high – not just within the Love Island house, but
also with the viewers.

"The public love couples who fight but appear committed. A “folk theorem” in game theory tells us that when people are sufficiently patient they can remain committed even when incentives to cheat are high. Sometimes second best is the best you can get, so you shouldn’t be greedy. Ultimately, while being liked by the public clearly helps, Love Islanders would do well to heed game theorists’
advice."

Sunday, June 17, 2018

Title IX reporting incentives and mis-incentives

The Atlantic has a story on a strange Title IX case (and a resulting lawsuit), with the suggestion that it might have to do with perverse legal and procedural incentives:

Mutually Nonconsensual Sex

Here's the critical paragraph:
"The event in précis, as summarized by Robby Soave of Reason magazine:
“Male and female student have a drunken hookup. He wakes up, terrified she's going to file a sexual misconduct complaint, so he goes to the Title IX office and beats her to the punch. She is found guilty and suspended.”

HT: MR

Saturday, June 16, 2018

The market for fake phone numbers

I am not of an age at which I get unwelcome requests for my phone number that I can't simply decline, but if you are, Lifehacker has some suggestions of numbers you could give to creepy strangers:

Give Out These Fake Prank Numbers to Creeps

These two make it clear what is likely the demand side of this market.

  • (646) 926-6614: The Mary Sue Rejection Hotline, which will say, “Oh hello there. If you’re hearing this message, you’ve made a woman feel unsafe and/or disrespected. Please learn to take no for an answer and respect women’s emotional and physical autonomy.  

  • (605) 475-6968: The Rejection Hotline, which politely explains to the individual that whoever gave them this number is not into them.

Friday, June 15, 2018

FRONTIERS OF MARKET DESIGN, June 22 at Cornell

Cornell is a hotbed of market design these days: here's the second conference there next week.
FRONTIERS OF MARKET DESIGN

"The first Frontiers of Market Design workshop will be held in Cornell University, Ithaca, NY on June 22, 2018 in conjunction with the 19th ACM Conference on Economics and Computation (EC).
Market design is a field of applied and theoretical research that sits comfortably on the intersection of economics and computer science. In recent decades, the theory and applications of market design have blossomed. In this workshop, we will focus on a set of promising, new applications of market design. In particular, we are interested in applications of market design which involve complex allocation constraints, vast datasets, and dynamic pricing issues. We also want to explore problems which, despite receiving ample theoretical attention, have not been implemented in practice. We welcome theoretical and empirical papers that deal with new domains of market design as well as papers that discuss practical aspects of market design."


Thursday, June 14, 2018

Workshop on Mathematical Optimization in Market Design, June 18-19

Two market design conferences to be held at Cornell next week: Here's the program for the first one.

INFORMS Workshop Mathematical Optimization 

and Market Design 2018


June 18, 2018

8:00-8:30 am Registration and Breakfast


8:30-8:45 am 
Workshop Opening 
Martin Bichler, Bob Day

8:45-9:30 am  Talk (Chair: Martin Bichler)Solving Large Incomplete-Information Extensive-Form Games, and Beating the Top Human Professionals at Heads-Up No-Limit Texas Hold'em
Tuomas Sandholm, Carnegie Mellon University

9:30-9:45 am Coffee Break 

9:45-11:15 am Session (Chair: Bob Day)
Optimization and Pricing in Non-Convex Markets

Allocation Under Stochastic Demand: A Primal-Dual Approach
Sasa Pekec, Duke University

Competitive Equilibria in Combinatorial Exchanges with Financially Constrained Buyers
Stefan Waldherr, Technical University of Munich (with Martin Bichler)

Linear Prices in Combinatorial Auctions
Bob Day, University of Connecticut

11:15-11:30 am Coffee Break 

11:30-1:00 pm Session (Chair: Ben Lubin)
Iterative Combinatorial AuctionsAn Efficient Ascending Auction for Private Valuations
Oleg Baranov, University of Colorado

Adaptive-Price Combinatorial Auctions
Sebastien Lahaie, Google (with Ben Lubin, Boston University)

Machine Learning-Based Combinatorial Auctions
Ben Lubin, Boston University (with Sven Seuken and Gianluca Brero, Zurich University)
1:00-2:00 pm Lunch Break 

2:00-3:00 pm Session (Chair: Itai Ashlagi)
Matching with Constraints and Complex Preferences

Hidden Substitutes
Scott Kominers, Harvard Business School

Clearing Matching Markets Efficiently: Informative Signals and Match Recommendations
Itai Ashlagi, Stanford University

3:00-3:10 pm Break 

3:10-4:30 pm Panel DiscussionBeyond Strategyproofness
Panel Chair: Martin Bichler, Technical Univ. of Munich 
Panelists: Eric Budish, Univ. of Chicago; Peter Cramton, Cologne Univ.; Michal Feldman, Tel-Aviv Univ.

4:30-4:40 pm Coffee Break 

4:40-5:30 pm Talk (Chair: Ben Lubin)
Markets for Road Use: Eliminating Congestion through Scheduling, Routing, and Real-time Road Pricing,
Peter Cramton, Cologne University (with R. Richard Geddes and Axel Ockenfels)

6 pm Joint Reception with ACM EC at the Gates Hall

June 19, 2018

9:00-10:30 am ACM EC Opening Session and Keynote
  
10:30-12:00 noon Session (Chair: Thayer Morrill)
Market Models and Applications

Quantity Contingent Auctions and Allocation of Airport Slots 
Michael Ball, University of Maryland

Prophet Inequalities Made Easy: Stochastic Optimization by Pricing Non-Stochastic Inputs
Michal Feldman, Tel-Aviv University

Family Ties:  Incorporating Siblings into School Choice
Thayer Morrill, NCSU

12:00-12:10 pm Break

12:10-12:55 pm Talk (Chair: Thayer Morrill)
Market Design and the FCC Incentive Auction
Larry Ausubel, University of Maryland  (with Christina Aperjis and Oleg Baranov)

Tuesday, June 12, 2018

Battle over college admission application platforms

CollegeNET, a software vendor to colleges, is suing the Common App., and also providing software support for it's newer competitor, the  Coalition for Access, Affordability, and Success.

 The Chronicle of Higher Ed has the (gated) story:

How Admissions Competition Brought New Rivalries, Strange Bedfellows, and ‘An Us-Versus-Us Lawsuit’  By Eric Hoover JUNE 03, 2018

"CollegeNET’s complaint claims that the Common Application used unfair tactics to stomp competitors and monopolize the market. It also claims that participating colleges (though not named as defendants) colluded to limit spending on application-processing services, harming other companies as well as applicants. How? By homogenizing the application process and causing "application churn," in which students apply to more and more colleges."


See earlier post:

Tuesday, September 29, 2015

Monday, June 11, 2018

The market for blood plasma: different tissues, same issues

I'll be speaking at The Plasma Protein Forum, June 12-13 in Washington D.C., on "REPUGNANT TRANSACTIONS AND FORBIDDEN MARKETS: DIFFERENT TISSUES, SAME ISSUES"

Here's the whole program:

DAY ONE:    JUNE 12, 2018

7:00 am–5:30 pmRegistration and Exhibit Hall Open
7:00–8:30 amBreakfast—Available in Exhibit Halls 1 & 2
8:30–8:45 amWELCOME & “How Is Your Day?Jan M. Bult, President & CEO, PPTA
8:45–9:00 amCHAIRMAN'S MESSAGE
David Bell, Chair, PPTA Global Board of Directors; Executive Vice President & General Counsel, Grifols
9:00–10:30 am  PERSPECTIVES: ACCESS TO CARE
Moderator:  Larisa Cervenakova, M.D., Ph.D., Medical Director, PPTA
Speakers: 
  • Tony Castaldo, President, US Hereditary Angioedema AssociationHAE perspective
  • Joanna Chorostowska-Wynimko M.D., Ph.D., D.Sc., Professor, Scientific Director National Institute of Tuberculosis and Lung Diseases, Warsaw, Poland
    Alpha-1 perspective

  • Professor Paolo Caraceni, Associate Professor, Department of Medical and Surgical Sciences, University of Bologna, Bologna, ItalyANSWER: Long term albumin administration improves survival in patients with decompensated cirrhosis
10:3011:00 amBREAK
11:00 am–12:30 pm  CURRENT INDUSTRY AND REGULATORY INITIATIVES IN THE AREAS OF INSPECTIONS AND DONOR HEALTH
Moderator:  John Delacourt, Vice President Legal Affairs & Global Operations, PPTA
Speakers:
  • Ginette Y. Michaud, M.D., Director, Office of Biological Products Operations, U.S. Food & Drug AdministrationFDA’s Office of Biological Products Operations – Form Follows Function
  • Toby L. Simon, M.D., Senior Medical Director, Plasma & Plasma Safety, CSL PlasmaDonor health perspectives: Insights from industry plasma vigilance data and future safety initiatives
  • George B. Schreiber, Sc.D., DirectorEpidemiology, PPTA
    Iron depletion in Source plasma donors; A non sequitur.
12:30–2:00 pm
LUNCHBuffet Available in Exhibit Halls 1 & 2    Sponsored by:  DIAMOND Roche Logo 01
2:00–4:00 pm  INTERNATIONAL ACCESS TO CARE LANDSCAPE
Moderator:  Jan M. Bult, President & CEO, PPTA
Speakers:
  • P. Martin van Hagen, M.D., Ph.D., Professor, Head, Clinical Immunology Unit Department of Internal Medicine, Erasmus Medical Center, Rotterdam, The Netherlands
    Is personalized medicine a national or cross border issue?
  • Dr. Ranjeet S. Ajmani, CEO, PlasmaGen BioSciences Pvt Ltd.
    Initiatives to improve access to care in India
  • Antonio Condino-Neto, M.D., Ph.D., President, Latin American Society for Immunodeficiencies (LASID)What needs to be done to improve access to Immunoglobulin therapy in Brazil? Diagnosis, Access, Supply
4:00 pm BREAK – Ice cream social
4:30 pmOtto Schwarz Award presentation
5:00 pmREPUGNANT TRANSACTIONS AND FORBIDDEN MARKETS: DIFFERENT TISSUES, SAME ISSUES
  • Alvin E. Roth, Craig and Susan McCaw Professor of Economics, Stanford University; George Gund Professor of Economics and Business Administration, Emeritus, Harvard University; and 2012 recipient of the Nobel Memorial Prize in Economic Sciences
5:45 pmDAY 1 CLOSING ANNOUNCEMENTS
5:45–7:30 pmReception        PPTA How is your day logo v2

DAY TWO:    JUNE 13, 2018

7:30 am–1:00 pmRegistration and Exhibit Hall open
7:30–8:30 amBreakfast—Available in Exhibit Halls 1 & 2
8:30–8:35 amWELCOME
Joshua Penrod, J.D., Ph.D., Vice President, Source & International Affairs, PPTA

8:35–8:50 amSOURCE DIVISION OUTLOOK
Roger Brinser, ‎Chair, PPTA Source Board of Directors; Director, Regulatory, BioLife Plasma Services/Shire

8:50–11:00 am CURRENT CHALLENGES 
Moderator: Joshua Penrod, J.D., Ph.D., Vice President, Source & International Affairs, PPTA
Speakers:
  • Nicola Lacetera, Ph.D., Associate Professor at the University of Toronto
    Crowding Out
  • Prof Dr. Liu ZhongVice President, Institute of Blood Transfusion, Chinese Academy of Medical Sciences (CAMS)Which is safer source plasma for manufacturing in China: apheresis plasma or recovered plasma?
  • Chen BinDeputy Director, Medical Safety and Transfusion Division, Department of Medical Regulatory and Management, National Health CommissionThe current situation and challenge of the Chinese plasma management

11:00–11:30 amBREAK
11:30 am–1:00 pm CAN COUNTRIES DELIVER ON THEIR OWN?
Moderator:  Julia Fabens, Senior Manager International Affairs, PPTA

Speakers:
  • Bill Bees, Vice President, Plasma Technologies, Prometic Plasma ResourcesCanada–Debunking the Krever Commission Report
  • Peter Jaworski, Ph.D., Professor, Georgetown University, Washington, D.C.
    Ethics of Compensation
  • Joshua Penrod, J.D., Ph.D., Vice President, Source & International Affairs, PPTAGlobal sufficiency: Obstacles and opportunities

***********
Here are my blog posts on plasma, sorted by date (most recent first), going back to one from 2009.

Sunday, June 10, 2018

Computing fairness

One of the areas in which computer science and economics touch each other more than a bit is in understanding what constitutes a fair way of allocating scarce resources.  This was the subject of a recent Northwestern CS workshop:

Quarterly Theory Workshop: Algorithmic Fairness

"Synopsis: As algorithmic systems have increasingly been deployed to make consequential decisions, it is becoming urgent that we grapple with the problem of (un)fairness and discrimination. These are delicate issues — to think rigorously about them, we first need to figure out how to formally define what we want, and then reason about how we might go about achieving our goals algorithmically — and what tradeoffs we will have to manage. This workshop focuses on recent advances in the theory of algorithmic fairness: both on foundational definitional questions, and on algorithmic techniques. The speakers are Nicole Immorlica (MSR), Jon Kleinberg (Cornell), Omer Reingold (Stanford), and Aaron Roth (U. of Pennsylvania).
The technical program of this workshop is organized by Aaron Roth and Jason Hartline."

These were the talks:
Speaker: Jon Kleinberg
Title: Algorithmic Fairness Criteria for Evaluating Individuals and Sets
Abstract: Recent discussion in the public sphere about classification by algorithms has involved tension between competing notions of what it means for such a classification to be fair to different groups. We consider several of the key fairness conditions that lie at the heart of these debates, and discuss recent research establishing inherent trade-offs between these conditions. We also consider a variety of methods for promoting fairness and related notions for classification and selection problems that involve sets rather than just individuals.
This talk is based on joint work with Sendhil Mullainathan, Manish Raghavan, and Maithra Raghu.
Speaker: Aaron Roth
Title: Preventing Fairness Gerrymandering in Machine Learning
Abstract: The most prevalent notions of fairness in machine learning are statistical notions: they fix a small collection of high-level, pre-defined groups (such as race or gender), and then ask for approximate parity of some statistic of the classifier (like positive classification rate or false positive rate) across these groups. Constraints of this form are susceptible to (intentional or inadvertent) fairness gerrymandering, in which a classifier appears to be fair on each individual group, but badly violates the fairness constraint on one or more structured subgroups defined over the protected attributes (such as certain combinations of protected attribute values). Individual notions of fairness avoid this problem, but at the cost of needing to make often unrealistic assumptions about the setting. We propose instead to demand statistical notions of fairness across exponentially (or infinitely) many subgroups, defined by a structured class of functions over the protected attributes. We will show how to achieve this in polynomial time, under the assumption that we are given oracles for solving the unconstrained learning problems (for both the set of classifiers to be learned, and the set of subgroups to be protected). Our algorithm casts the problem as solving for an equilibrium in a zero sum game, and observes that learning oracles are enough to efficiently play no-regret learning dynamics. We then demonstrate experimentally that our proposed algorithms are practical, by investigating their performance on several real datasets when instantiated with learning heuristics in place of oracles.
This talk is based on joint work with Michael Kearns, Seth Neel, and Steven Wu.
Speaker: Omer Reingold
Title: On Algorithmic Fairness Between Groups and Individuals 
Abstract: As algorithms increasingly inform and influence decisions made about individuals, it is increasingly important to address concerns that these algorithms might be discriminatory. Historically, definitions of fairness fell into one of two extremes: (1) broad-strokes statistical guarantees; (2) individual-level protections. Statistical guarantees tend to be much easier to satisfy (information and complexity theoretically), but tend to be much weaker in the protections they provide. We will discuss two recent works towards bridging the gap between statistical and individual protections. These works provide efficient learning algorithms that also ensure every subpopulation within some rich class is protected (according to some fairness notion).
One of the notions we will discuss is that of multicalibration — a new measure of algorithmic fairness that aims to mitigate concerns about discrimination that is introduced in the process of learning a predictor from data. The second notion studies fair classification within the versatile framework of Dwork et al. [ITCS 2012], which assumes the existence of a metric that measures similarity between pairs of individuals. Unlike previous works on metric-based fairness, we study the setting where a learning algorithm does not know the entire metric, but can query similarities between particular pairs of individuals a bounded number of times. We note that we do not make any assumption on the metric and are still able to obtain meaningful protection from systemic discrimination that we refer to as “metric multifairness.”
The talk will be focused on the various ways in which algorithmic fairness can be defined but will also elude to some of the ways in which it can be obtained. It is based on joint works with Úrsula Hébert-Johnson, Michael P. Kim and Guy Rothblum.
Speaker: Nicole Immorlica
Title: Fair Classification and Learning
Abstract:  Classification, learning, and optimization algorithms are increasingly used to make decisions of social consequence. An automated resume reader classifies applicants, deciding which are worth interviewing. A learning algorithm estimates the impact of a drug on health outcomes. A facility location algorithm chooses where to locate new warehouses. For these algorithms to be useful, practitioners strive to maximize their accuracy and optimality. However, this can result in differential treatment across population subgroups, resulting in a sense of inequity. Fairness metrics are used to measure the degree of inequity. In this talk, we explore how to jointly optimize fairness and accuracy using, as a black box, existing algorithms that optimize accuracy or optimality. Our first solution is tailored to classification tasks and proposes decoupling the classification task, using different classifiers for each subgroup. Our second solution can optimize any smooth and continuous function of accuracy and fairness in classification, learning, and optimization settings. It further has the advantage that subgroup membership, while used to train, is not used at run time. This additional power comes with the cost of added computational complexity in training.