DS457/DS657/JD673: Law and Algorithms

Course Overview

This cross-cutting and interdisciplinary course, taught jointly between the School of Law, the faculty of Computing and Data Sciences, and Computer Science investigates the role that algorithms and automated decision-making systems play in law and society. The course connects legal and computational concepts of transparency, fairness, bias, trust, and privacy, though a series of case studies that present recent applications of technology to legal and regulatory situations and explore the challenges in regulating algorithms.

Legal concepts explored will include evidence and expert witnesses, anti-discrimination law concepts of disparate impact and disparate treatment, sectoral information privacy regimes, and public access and transparency laws. Computational concepts explored will include artificial intelligence and machine learning, secure multi-party computation, differential privacy, and zero-knowledge proofs.

Syllabus

This course will meet in LAW 204 in the Law Tower. The syllabus for this course can be found as a PDF here.

Readings

Class 1: Intro to Law and Algorithms (Jan. 20)

Required

Intro to Computational Thinking:
Intro to Law and Legal Thinking
Law, Algorithms, and Power:
Influences and Collisions Across Domains:

Optional

The social effect of algorithms:
The regulation of algorithms:
More introductory law & algorithm readings:
Algorithms and power:

Class 2 – The Development and Legal Protection of Software (Jan. 27)

Required

The Problem:
Introduction to Software Development:
Intellectual Property and Software:
  • If you are new to field of intellectual property, review these introductory concepts from the Digital Media Law Project: Intellectual Property, Copyright, and Trade Secrets.
  • Sonia Katyal, The Paradox of Source Code Secrecy, 104 Cornell L. Rev. 1183 (2019) – read Parts I(C) & I(D) (“The Copyrightability of Software”and “The Continuing Overlap Between Copyright and Trade Secrecy”); the introduction to Part IV (the beginning of “Due Process in an Age of Delegation”) and Part IV(B) (“The Constitutional Cost of Secrecy”).
  • Rebecca Wexler, Life, Liberty, and Trade Secrets, 70 Stan. L. Rev. 1343 (2018) – read Part I (“Trade Secrets in the Criminal Justice System”)
The Role of CS Experts in Law Enforcement Procurement:
The Role of CS Experts in Criminal Trial:

Optional

No Readings Posted Yet

Class 3 – Putting the TrueAllele Algorithm on Trial (Feb. 3)

Required

TrueAllele on Trial
Transparency as a Solution and its Critics
Broader Solutions and Impediments:

Optional

Briefing Courts on Forensic Technology:
Regulatory responses to Forensic Technology
Transparency

Class 4 – The COMPAS Algorithm and the Optimization Paradox (Feb. 10)

Required

Defining Fairness
History of Risk Prediction
The COMPAS Algorithm
The Optimization Paradox

Optional

Formal models of fairness and optimization:
A Return to Transparency:

Class 5 – Is There a “Right” Way to Use Algorithms in Criminal Sentencing? (Feb. 17)

Required

Human/Algorithm Dissonance:
The Data Source Problem:
What is the right model of “fairness”?
The bigger picture

Optional

Class 6 – Artificial Intelligence and Anti-Discrimination Laws (Feb. 24)

Required

Intro to AI/ML and its Limits:
Algorithms and Bias
  • Timnit Gebru, Race and Gender, in The Oxford Handbook of Ethics of AI (2020) – read following sections, circulated separately:
    • "Using Past Data to Determine Future Outcomes Results in Runaway Feedback Loops”
    • “AI-Based Tools are Perpetuating Gender Stereotypes”
    • “Power Imbalance and the Exclusion of Marginalized Voices in AI”
    • “Education in Science and Engineering Needs to Move away from ‘the View from Nowhere’”
  • Ruha Benjamin, Race After Technology (2019) – read “Raising Robots,” circulated separately
  • Solon Barocas & Andrew Selbst, Big Data’s Disparate Impact, 104 Cal. L. Rev. 671 (2016) – read part I only (pp. 677–92)
Algorithms and Antidiscrimiantion Laws

Optional

Class 7 – Can Algorithms Mitigate Bias? (March 3)

Required

Adjustments to models:
Adjustments to model evaluations:
New Laws:
Working within existing anti-discrimination law:

Optional

Class 8 – Vote by Paper, Vote by Mail, Vote by Smartphone (March 17)

Required

Algorithms and Trust:
  • Ethan Zuckerman, “The Economics of Mistrust,” in The End of Trust (2018) – read all, circulated separately
  • Brian Stanton & Theodore Jensen, Trust and Artificial Intelligence, NIST (March 2021) – read sections 2 and 3 only (pp. 1–6)
Case Study: Voting

Optional

Class 9 – How Do We Trust the Vote? (March 24)

Required

Risk Limiting Audits:
Zero-Knowledge Proofs:
Cryptographic Voting Systems:
  • Josh Benaloh, Ronald Rivest, Peter Y. A. Ryan, Philip Stark, Vanessa Teague, Poorvi Vora, End-to-End Verifiability (Feb. 2014) – read all
  • Matthew Bernhard, Josh Benaloh, J. Alex Halderman, Ronald L. Rivest, Peter Y. A. Ryan, Philip B. Stark, Vanessa Teague, Poorvi L. Vora, and Dan S. Wallach, Public Evidence from Secret Ballots (Aug. 2017) – read Parts 3 and 4 only
Limits of Cryptography:

Optional

Class 10 – “Privacy,” “Security,” and “Encryption” (March 31)

Required

Legal Concepts of Privacy:
  • Daniel Solove, Understanding Privacy (2008) – read chapter entitled “Privacy: A Concept in Disarray”
  • Stephen P. Mulligan, Wilson C. Freeman, & Chris D. Linebaugh, Data Protection Law: An Overview, Congressional Research Service (March 25, 2019) – read “Origins of American Privacy Protections (pp. 3–7), the intro to “Federal Data Protection Law” (pp. 7–8), and “Electronic Communications Privacy Act” (pp. 25–29)
Intro to Encryption:
Encryption Policy:

Optional

Class 11 – Differential Privacy and The Census (April 7)

Required

Data, Anonymity, and Re-Identification:
Intro to Differential Privacy:
Utility and Privacy in the Census:
Use of Differential Privacy in the Census:
Legal Challenges to Differential Privacy:

Optional

Class 12 – Conducting Analysis Over Secret Data (April 14)

Required

Intro to Secure Multiparty Computation
  • Jessica Colarossi & Carlos Soler, What Is Secure Multiparty Computation?, Boston University (Feb. 27, 2019) – watch video
  • Yehuda Lindell, A Primer in Secure Multiparty Computation, Unbound (2020) – read “Introduction to Multiparty Computation,” “Mathematical Guarantees of Security,” and “Security Definitions” (pp. 2–6) (For the purposes of this class, you can safely think of '⊕' as meaning '+')
Case Studies:
Boston Women’s Workforce Council
Smartphone-based Contact Tracing
Google and Mastercard
When is MPC regulated by information privacy law?
Under the Drivers Privacy Protection Act
Under the CCPA
Under the GDPR

Class 13 – Law and Algorithms (April 21)

Required

Changes in Governance:
  • Andrew Tutt, An FDA for Algorithms, 69 Admin. L. Rev. 83 (2017) – read Section II only (pp. 104–11)
Change in Remedies:
Changes in Design and Power:
Changes to Professional Duties:

Optional