Weapons of Math Destruction Major Figures

Cathy O’Neil

Although her book is primarily based on examples from the public sphere, O’Neil presents her own varied career experiences as a “journey of disillusionment” in Chapter 2. She tells of how, after an early career in pure mathematics, she joined a hedge fund as an analyst who used mathematical models to guide investment decisions. Her realization that these practices were contributing to the global financial crisis spurred her to leave that job and to research the use of such models in other areas. After blogging for years alongside her work as a data scientist, O’Neil wrote Weapons of Math Destruction. Although little autobiographical material is found after this chapter, O’Neil does supply anecdotal examples from her daily life (e.g., planning family dinners) to show that not all modeling is necessarily formal in structure or sinister in its outcomes.

Barack Obama

Barack Obama (born 1961), president of the United States from 2009 to 2017, receives mention in Weapons of Math Destruction as an example of a political candidate whose campaigns benefited from the power of mathematical modeling. O’Neil describes Obama’s 2012 reelection campaign in Chapter 10 (“The Targeted Citizen”), where she notes that his team used sophisticated modeling to find persuadable swing-state voters and potential donors. In Chapter 3 (“Arms Race”), O’Neil describes the Obama administration’s attempts to address some of the negative effects of college ranking systems, a form of mathematical modeling with an outsized impact on college recruitment and spending decisions.

Mitt Romney

American politician Mitt Romney (born 1947) was the governor of Massachusetts when he made his bid for the presidency in 2012. After winning the Republican primary, Romney lost to incumbent Barack Obama and temporarily retired from electoral politics. O’Neil’s mention of Romney (Chapter 10) focuses on a well-publicized incident from this campaign, in which he claimed 47% of Americans were dependent on government entitlements and considered themselves “victims.” This remark, recorded and leaked from a private gathering, underscores O’Neil’s point about the increasing tendency of politicians to target specific subgroups of voters with different, even incompatible messages. In the years after Weapons of Math Destruction was published, Romney sought and won election to the U.S. Senate, taking office in 2019 as the junior senator from Utah.

Hillary Clinton

Hillary Clinton (born 1947) served as a U.S. senator from 2001 to 2009, then as Secretary of State from 2009 to 2013. She is perhaps best known as the Democratic candidate for the U.S. presidency in the election of 2016, where she ultimately lost to Republican nominee Donald Trump (born 1946). Clinton’s presidential campaign appears in Chapter 10 and the afterword as an example of the use of statistical modeling in politics.

From a data-science perspective, O’Neil describes Clinton’s role in the 2016 election as noteworthy in two main ways. First, Clinton’s team further developed the voter “micro-targeting” used during the 2012 election, applying tools from the private-sector field of customer relationship management. Second, as mentioned in the afterword, Clinton’s eventual loss caught many political analysts off guard despite their sophisticated models and extensive polling. This, for O’Neil, shows the difficulty of modeling rare events like elections that have many changing variables.

Sarah Wysocki

Sarah Wysocki is one of the teachers who was caught up in the Washington, D.C., school district’s push to score its educators and fire the lowest ranked. Her story appears in the introduction as O’Neil’s first major illustration of how poorly designed and irresponsibly applied mathematical models can harm individuals. Well-liked by administrators and parents and considered an effective teacher by colleagues, Wysocki received an inexplicably low score in the model applied to weed out underperforming D.C. teachers. Her quest for clarification as to why she scored so low proved futile since the scoring model is not available for public inspection. O’Neil suggests that Wysocki’s students may have had their standardized exam scores inflated in the previous grade, masking the value later added by Wysocki’s teaching. Ultimately, Wysocki found work at a wealthier school district that did not rely on an algorithm to make its human-resources decisions.