Introduction
You likely assume that software, algorithms, and automated decisions are neutral because math cannot hold personal prejudices. However, digital tools frequently conceal, accelerate, and deepen systemic racial inequities under the guise of scientific progress.
Ruha Benjamin argues that technology operates as the New Jim Code, embedding historical discrimination into modern hardware and software design. You will see how automated systems punish marginalized communities while masking bias behind the veneer of progress.
The book traces this dynamic across five major dimensions.
- How do technical systems reproduce structural discrimination?
- Why do predictive algorithms transform historical injustices into objective conditions?
- When does digital visibility become a trap of hyper-surveillance?
- How do corporate fixes repackage racial bias as efficiency?
- What tools can you use to counter carceral technologies?
1. Engineered Inequity
Technical systems and automated decision tools reproduce structural discrimination by encoding human prejudices into software and hardware design.
You might assume that machines are neutral because they lack personal malice. When software outputs discriminatory results, defenders often claim that algorithms cannot be racist without a brain. Yet systemic discrimination operates without conscious malice through the uncritical adoption of historical data and default settings.
Technical systems encode prejudice into software.
On August 2, 2016, creators of a contest called Beauty AI expressed dismay at biased results. Deep learning software was judged against embedded preferences using pre-labeled images. Out of 44 winners across 6,000 submissions from approximately 100 countries, all except six were white, and only one finalist had visibly dark skin.
This software judges worth by mirroring societal hierarchies. Darker-skinned people are implicitly coded as unhealthy and unfit, mirroring historical eugenic ideologies. You can trace the etymology of the word stereotype back to printing trade practices. A stereo was a solid plate used to make fixed copies, and the term eventually designated images perpetuated without change.
Robots and Bias
Silicon Valley maintains a stark internal hierarchy. You see high-paid creative elites, primarily white men and some white women, contrasted with low-paid manual laborers and outsourced workers. The word robot originates from the Slav term for servitude and hardship. A 1957 article in Mechanix Illustrated predicted the return of personal slavery using robot slaves by 1965.
Public agencies and private companies use automated decision systems to outsource responsibility and evade democratic oversight. Virginia Eubanks documented how US social welfare agencies incorporate predictive analytics to determine benefit eligibility in 2018. These tools allow institutions to depersonalize welfare distribution under the banner of objectivity, creating a postdemocratic approach to governing.
Automating Anti-Blackness
Mundane automated technologies apply racialized logic by treating specific skin tones as normative. Near infrared technology in automated soap dispensers requires light to bounce back from the user. Skin with more melanin absorbs more light and fails to trigger the sensor, rendering certain populations invisible.
Automated social credit systems evaluate citizens based on conformity to dominant behavioral norms. China announced plans for a citizen scoring system in 2014, with mandatory enrollment set to begin in 2020. These programs track financial, educational, criminal, and leisure activities to score trustworthiness. In the Black Mirror episode "Nosedive," protagonist Lacie navigates a social credit system where star ratings dictate access to housing and transport.
2. Default Discrimination
Database designs and predictive algorithms transform historical injustices into objective weather conditions that disproportionately police and penalize marginalized communities.
Ruha Benjamin calls this phenomenon default discrimination. You encounter it when digital tools present human prejudice as impartial mathematics. Instead of viewing glitches as fleeting interruptions, you should recognize them as signals revealing how a corrupted system operates. In the film The Matrix, a black cat crossing a path twice serves as a déjà vu glitch that alerts freedom fighters to an imminent confrontation.
Princeton media specialist Allison Bland drove through Brooklyn and heard the Google Maps narrator direct her to turn right on "Malcolm Ten Boulevard," interpreting the letter X as a Roman numeral rather than recognizing Malcolm X.
The Glitch Systemic
Society often focuses on individual bad apples whose offensive speech is exposed in the media, treating them as temporary anomalies. This public shaming functions as a distraction from the broader social structures, housing policies, and institutional environments giving those discriminatory practices their license.
Donald Sterling faced intense public condemnation and was forced to sell the Los Angeles Clippers after a private racist phone call leaked in 2014. Meanwhile, his documented pattern of housing discrimination against Black, Latinx, and non-Korean tenants had drawn little sustained scrutiny. The Los Angeles NAACP planned to honor Donald Sterling with a lifetime achievement award in 2009.
Automated risk assessment and predictive policing algorithms use historical crime data to generate self-fulfilling prophecies that disproportionately target marginalized communities. Because predictive policing tools draw from historical arrest records and police surveillance priorities concentrated in minority neighborhoods, they treat existing police bias as objective weather conditions. These algorithms convert systemic racial bias into automated scores that determine risk profiles.
ProPublica investigators found that Broward County recidivism risk scores falsely flagged Black defendants as future criminals at nearly twice the rate of white defendants, while mislabeling white defendants as low risk more often.
Architecture and Algorithms
Physical and digital architecture use design elements to reinforce social hierarchies, restrict public access, and control marginalized populations. Physical structures like defensive campus architecture, divided benches, and low-hanging parkway bridges parallel digital algorithms by actively limiting public gatherings and mobility. These designed artifacts incorporate political choices and reinforce socio-economic boundaries.
Robert Moses designed low-hanging overpasses along the Long Island parkway system that prevented public buses from passing, restricting lower-income and non-white individuals who relied on public transit from reaching suburbs and beaches. The historical span of the Jim Crow era lasted from 1876 to 1965.
Online search engines and language-processing algorithms replicate human cultural biases and historical stereotypes through the automated processing of digital data. Because algorithms learn patterns from massive piles of web data and human writing, they ingest and reproduce societal prejudices related to race, gender, and class. Google uses over 200 unique signals or clues in its search algorithms.
Searching for three Black teenagers on Google yielded criminal mug shots, searching for three white teenagers produced photos of smiling youths, and searching for three Asian teenagers presented scantily clad women.
3. Coded Exposure
Visibility through digital technologies acts as a trap that subjects racialized populations to hyper-surveillance, scopic vulnerability, and data-driven state control.
Frantz Fanon described the scopic vulnerability of being looked at but not truly seen by a white child on the streets of Paris, exclaiming about his Blackness.
Exposing Whiteness
Kodak utilized standardized reference images called Shirley Cards featuring a white woman from the 1950s to the 1990s to calibrate color exposure. Because white skin was set as the default norm, darker skin tones were routinely underexposed and rendered blurry.
Artists Adam Broomberg and Oliver Chanarin used decades-old film and a Polaroid ID-2 camera in South Africa to demonstrate how the narrow light range rendered a black person invisible next to a white person.
Polaroid sold photo-identification units and cameras with flash boost buttons to South Africa, which were used to capture images of Black citizens for passbooks that heavily restricted their movement under apartheid. African American employees Ken Williams and Caroline Hunter founded the Polaroid Workers' Movement, protesting that Polaroid was effectively imprisoning Black people in 60 seconds and leading a multi-year boycott until Polaroid pulled out of South Africa.
Law enforcement facial recognition databases are disproportionately populated with African Americans, while the underlying algorithms perform poorly at recognizing Black faces. Georgetown Law School researchers analyzed over 10,000 pages of information from more than 100 police departments and found that facial recognition databases are disproportionately African American while software is especially bad at recognizing Black faces.
Exposing Citizenship
Governments increasingly deploy biometric databases, isotope testing, and digital ID systems at borders and domestically to police populations and exclude vulnerable groups. These projects use scientifically flawed methods under the guise of objective efficiency to criminalize populations and create digital outcasts.
India implemented the Aadhaar 12-digit ID system tied to biometric markers, which became mandatory for welfare programs, leaving citizens who opt out or fail authentication unable to access food or school lunches. Kuwait enacted a mandatory DNA database law targeting 1.3 million citizens and 2.9 million expatriates with an estimated cost of $400 million before the courts struck it down.
4. Technological Benevolence
Corporate fixes, AI recruitment tools, and medical diagnostics repackage racial bias under the progressive veneer of efficiency, personalization, and diversity.
Proponents frame electronic monitoring as cost-effective and conducive to family and work integration. The technology actually increases surveillance, relies on biased risk assessment algorithms, and shifts the financial burden of containment onto individuals.
Grace, a single mother from El Salvador, was imprisoned for over a year before being released with an ankle monitor she had to pay for herself, living under constant threat of re-detention.
Platforms like HireVue analyze thousands of verbal and non-verbal cues from recorded interviews against top-performing employees. Because these tools rely on historical data reflecting workplace demographics, they replicate and streamline hiring discrimination.
Amazon had to scrap an AI recruitment tool because it discriminated against women by downgrading resumes that listed women's colleges or women's chess clubs.
Job seekers deploy clever tactics to bypass these systems:
- trading answers to employers' tests and creating fake applications as informal audits;
- slipping the words Oxford or Cambridge into a CV in invisible white text to pass automated screening.
Technological Redlining
Companies circumvent civil rights legislation prohibiting direct collection of ethno-racial data by using predictive software that infers ethnicity from names and zip codes. Firms like Diversity Inc. use name data combined with geographic profiles for industries such as finance, housing, and healthcare, turning historical segregation into a profitable commodity.
To assign a racial identity to Sarah Johnson, whose last name does not predict ethnicity, the system checks her zip code to see if it is a majority Black neighborhood. Diversity Inc. has delineated and built over 150 distinct ethnicities and claims a 96 percent accuracy rate in building knowledge around client names.
Institutional celebrations of diversity often serve as superficial gestures that monetize cultural differences while leaving structural inequality untouched. Institutions adopt happy talk and flat conceptions of diversity to celebrate cultural differences as competitive advantages, effectively masking ongoing systemic inequities.
Bankers meeting in a town casually moved from discussing stock market trends to mocking Indigenous Peoples' Day and sharing their favorite places for ethnic food.
Targeted advertising and data-tracking systems rely on generalized assumptions about social groups rather than unique individual traits, undermining privacy and civic values under the guise of openness.
STS scholar Janet Vertesi tried to keep her pregnancy private from online data trackers by avoiding baby searches and paying with cash, only to be flagged for suspicious activity and excessive gift card purchases.
Fixing Health
Medical technologies embed racial assumptions directly into diagnostic instruments, using race-corrected metrics to justify disparities in treatment and compensation. Devices like the spirometer incorporate race correction buttons that assume innate biological differences in lung capacity between racial groups, which historically minimized payouts and disability claims for Black workers.
An insulation manufacturer in 1999 used race correction on spirometers to defend against a class-action lawsuit from 15,000 asbestos workers, requiring Black workers to show worse lung symptoms to qualify for compensation.
Healthcare hotspotting uses geospatial and insurance data to reallocate resources to high-cost patients, but in doing so it stigmatizes vulnerable populations and reinforces systemic disparities. By applying policing and spatial profiling techniques to medicine, hotspotting categorizes marginalized patients as socially disintegrated and responsible for healthcare financial burdens.
In Camden, New Jersey, healthcare providers began using insurance claims and GIS mapping to identify super-utilizers and medically vulnerable populations. This environment reflects a landscape where 5 percent of Americans account for nearly 50 percent of healthcare costs, 1 percent comprises 20 percent of costs, and 13 percent of the population accounts for 80 percent of healthcare costs in Camden.
5. Retooling Solidarity, Reimagining Justice
Abolitionist tools and community-led data democratization counter carceral technologies and prioritize collective equity over market efficiency.
Dr. Kortney Ziegler and Tiffany Mikell partnered with the National Bail Out movement to launch Appolition, an app that converts daily spare change into bail money. Appolition gathered 8,000 enrollments after its November launch, landing in the top ten most innovative companies in 2018, raising $230,000, and freeing over 65 people. This initiative contrasted with a modest kickoff enrollment goal planned by Dr. Kortney Ziegler in September 2017 of 600 people. Around 700,000 people are the average daily number condemned to local jails simply because they cannot afford bail. A county can spend a maximum amount of $200 a day holding someone in jail.
Technological reforms marketed as decarceration start-ups often expand the prison-industrial complex by substituting physical cages with digital surveillance cells. The decarceration start-up Promise charges $17 for GPS monitoring and received $3 million in venture capital. Backed by rapper Jay-Z, the Promise app collaborates with law enforcement and charges significantly less than holding someone in jail.
Virtual reality marketed as an empathy machine commodifies human anguish and turns suffering into a source of corporate revenue and cultural tourism. In the first quarter of 2016 alone, venture capitalists invested $1.2 billion in VR technologies, marking a 50 percent increase in VR investments compared to the previous quarter. The UN's VR film Clouds over Sidra raised $3.8 billion. Facebook founder Mark Zuckerberg used a VR app to tour the wreckage of hurricane-ravaged Puerto Rico alongside another executive, generating widespread criticism.
Design thinking and its five-step human-centered methodology often function as a colonizing project that erases non-designer agency and sanitizes structural inequalities. At the 2018 Design Research Society meeting, panelists awkwardly ignored audience questions posed via Slido regarding why design is so white and how it relates to oppressive systems.
Beyond Code-Switching
Rethinking cultural codes through equity audits and data democratization offers a viable alternative to mere code-switching and uncritical tech innovation. Relying on individual data rights under frameworks like the General Data Protection Regulation, which became applicable across European member states on May 25, 2018, is insufficient. Organizers with the Detroit Digital Justice Coalition developed the DiscoTech model and the Our Data Bodies project to demystify technology and build community-led digital defense strategies.
You must evaluate technological interventions by answering specific questions:
- Ask what the unintended consequences of designing systems at scale on the basis of existing patterns in society are.
- Determine when and how AI systems should prioritize individuals over society and vice versa.
- Assess when introducing an AI system is the right answer and when it is not.
Speculative art and subversive design flip the script on predictive technologies by exposing the hidden crimes of the wealthy and empowering marginalized communities. Creators Brian Clifton, Sam Lavigne, and Francis Tseng developed the White-Collar Early Warning System, a parody heat map using executive LinkedIn profile photos to predict corporate financial crimes. The training set for the White-Collar Early Warning System algorithm used 7,000 corporate executives and their profile photos downloaded from LinkedIn.
10 Key Ideas
- Technical systems and automated decision tools reproduce structural discrimination by encoding human prejudices into software and hardware design.
- Database designs and predictive algorithms transform historical injustices into objective weather conditions that disproportionately police and penalize marginalized communities.
- Visibility through digital technologies acts as a trap that subjects racialized populations to hyper-surveillance, scopic vulnerability, and data-driven state control.
- Corporate fixes, AI recruitment tools, and medical diagnostics repackage racial bias under the progressive veneer of efficiency, personalization, and diversity.
- Abolitionist tools and community-led data democratization counter carceral technologies and prioritize collective equity over market efficiency.
- Public agencies and private companies use automated decision systems to outsource responsibility and evade democratic oversight.
- Automated risk assessment and predictive policing algorithms use historical crime data to generate self-fulfilling prophecies that disproportionately target marginalized communities.
- Law enforcement facial recognition databases are disproportionately populated with African Americans, while the underlying algorithms perform poorly at recognizing Black faces.
- Companies circumvent civil rights legislation prohibiting direct collection of ethno-racial data by using predictive software that infers ethnicity from names and zip codes.
- Technological reforms marketed as decarceration start-ups often expand the prison-industrial complex by substituting physical cages with digital surveillance cells.