Weapons of Math Destruction

INTRODUCTION

  • Childhood Fascination with Numbers

    • Used to gaze at traffic and study license plates, breaking down numbers into prime factors.

    • Example: 45 = 3 x 3 x 5 (definition of factoring).

    • Interest in prime numbers as a budding math enthusiast.

  • Developing a Love for Mathematics

    • Attended math camp at age fourteen, developed a passion for mathematics.

    • Math served as a refuge; described it as orderly and ever-expanding.

    • Majored in mathematics and received a Ph.D., focusing on algebraic number theory.

TRANSITION TO FINANCE

  • Career Shift from Academia to Finance

    • Worked as a tenure-track professor at Barnard College in collaboration with Columbia University.

    • Transitioned to a quant role at D. E. Shaw, a hedge fund, marking a shift from pure theory to practical applications of mathematics.

  • The 2008 Financial Crisis

    • Initially excited by working within the global economy, the 2008 crash revealed the dark side of mathematics.

    • Mathematics perceived as a tool that exacerbated financial crises, including:

    • Housing crisis

    • Collapse of financial institutions

    • Increased unemployment

BIG DATA ECONOMY

  • Growth of Mathematical Techniques Post-Crisis

    • Post-crisis, mathematical models became increasingly prominent across various domains.

    • Focus shifted from financial markets to human behavior, studying:

    • Consumer desires

    • Spending habits

    • Predictive measures of trustworthiness and potential in various roles.

  • The Appeal of Algorithms in Decision-Making

    • The ability to process vast amounts of data quickly (e.g., résumés and loan applications) was marketed as efficient and fair, free from human bias.

    • By around 2010, the mathematical influence on human affairs was greater than ever, largely accepted by the public.

  • Dangers of Opaque Mathematical Models

    • The models often reflected human biases and prejudices.

    • Resulted in punishments skewed against the poor and marginalized, worsening inequality.

    • Named these harmful models "Weapons of Math Destruction" (WMDs).

CASE STUDY: WMD IN EDUCATION

  • Background on Education Reform in Washington D.C.

    • In 2007, Mayor Adrian Fenty aimed to reform D.C.'s underperforming schools hiring Michelle Rhee as chancellor.

    • Emphasis was placed on evaluating teachers to optimize school performance and student outcomes.

  • Implementation of Teacher Assessment Tool: IMPACT

    • Developed to identify and eliminate the lowest-performing teachers.

    • In 2009-10, teachers in the bottom 2% were fired; in the following year, an additional 206 teachers were removed from their positions.

EXAMPLE OF SARAH WYSOCKI

  • Sarah Wysocki's Experience

    • A fifth-grade teacher at MacFarland Middle School, received positive feedback from parents and principals.

    • Despite this, she was rated poorly under the IMPACT evaluation due to the value-added modeling used to assess her teaching effectiveness, which weighed test scores heavily.

  • Nature and Impact of Value-Added Modeling

    • The model measured educational progress but failed to account for numerous confounding variables (e.g., student background, personal issues affecting performance).

    • Wysocki felt devastated by the rating system she did not understand, prompted by a complex scoring algorithm developed by Mathematica Policy Research.

COMPLEXITY OF EVALUATING TEACHERS

  • Challenges in Measuring Teaching Effectiveness

    • Attempts to isolate a teacher's impact on student performance are convoluted and statistically dubious.

    • Contrasted to corporations like Google that run robust statistical tests using massive datasets to evaluate performance.

  • Limitations of Feedback Systems in WMDs

    • Models without sufficient feedback are self-perpetuating, continuously generating unreliable outcomes without learning from errors.

    • Example: WMD systems firing 206 teachers as reflecting efficiency but not truth.

EXAMPLES IN WMD SYSTEMS

  • The Role of Credit Scores in Employment

    • Employers perceive low credit scores as an indicator of poor job performance, leading to higher unemployment rates among those affected.

    • These scoring systems operate in a feedback loop of poverty and joblessness, compounding existing disadvantages for low-income individuals.

  • The Opaque Nature of Algorithmic Decisions

    • The complexity and obscurity of WMDs discourage accountability. Individuals impacted cannot understand or contest the ratings affecting their careers or lives.

    • Example: Complaints from teachers about performance ratings and demands for transparency rarely yield results.

CONCLUSION AND CALL TO ACTION

  • Reflections on the Rise of WMDs

    • Following the housing crash, the author recognized the prevalence of harmful algorithms across multiple industries, emphasizing an awareness of data structures.

    • Initiated a blog, MathBabe, to raise awareness regarding misuse of statistics and WMDs.

  • Involvement in Economic Justice Movements

    • Joined groups advocating for economic reform, recognizing the importance of public understanding in battling WMDs.

    • Noted that ill-conceived mathematical models significantly impact critical life decisions.

  • Emphasis on the Dark Side of Big Data

    • Acknowledged the prevalence of financial incentives driving the continued use of flawed WMDs.

    • A call to provide agency and ethical consideration towards those who suffer the consequences of algorithmic decisions.