Module 1, Sep 9, Seminar and lab

The history of technology and social work

Background

American poor relief ran through buildings before it ran through records. Boston opened its first workhouse in 1660, and by the nineteenth century the poorhouse was the main form of public aid, its minimal food, forced labor, and crowding kept deliberately punitive to discourage anyone with another option. An 1883 investigation exposed conditions at the almshouse in Tewksbury, Massachusetts, where Anne Sullivan, later Helen Keller’s teacher, spent part of her childhood; she called her years there “a crime against childhood.”

In the late nineteenth century, “friendly visitors” from Charity Organization Societies investigated families experiencing poverty and filed what they found into centralized case registries, cross-checked across agencies to catch “duplicate” applicants. The movement’s founding plan called for a complete registry of every person in the city receiving public or private assistance, and its views dominated private charity philosophy until the 1930s. It was an anti-fraud database built by the profession you are entering, and it is a direct ancestor of the risk-scoring systems counties run on families today. Eubanks traces the full line from poorhouse to scientific charity to welfare computerization to eligibility algorithms, and her point about “rational discrimination” is that once rules are automated, no animus is required for the system to punish people in poverty at speed and scale.

In the same era, Hull House residents mapped wages and nationalities block by block, and W. E. B. Du Bois hung hand-drawn charts of Black life in Georgia at the 1900 Paris Exposition. Du Bois and a team of Atlanta University students drew 63 charts for the exhibit and paired them with more than 500 photographs. In 1895, the year the Hull House maps were published, Ida B. Wells-Barnett’s The Red Record tabulated lynching statistics (241 people lynched across 26 states and territories in 1892 alone), drawn deliberately from white-owned newspapers so the numbers could not be dismissed as biased reporting. It is the first documented statistical report on lynching in the United States.

The mapping tradition rode its evidence into law: child labor regulation, tenement housing codes, the first juvenile court in the country. Both lineages counted people experiencing poverty; what separates them is who designed the system, whose problem definition won, and who paid when it was wrong.

Key ideas

Scientific charity and the casework registry

The poorhouse “helped” by deterring, with conditions inside made grim so only the desperate would apply. The institution did not simply fade when relief reformed. Katz shows that poorhouses were reorganized into de facto old-age homes as younger and able-bodied residents found other options, an institutional line that runs into modern public nursing homes. The Northampton County Almshouse in Virginia, established in 1803 as a combined workhouse and poorhouse, confined enslaved people in its basement, a history reconstructed only recently from archives and oral testimony. The paperwork of the era survives in bulk; Boston’s Overseers of the Poor left 19 boxes of admission and discharge registers, warning-out books, and correspondence covering 1733 to 1925.

Scientific charity refined the logic with paperwork: investigate the applicant, verify the story, individualize the fault, and record everything. A “friendly visitor” was assigned to advise a family and supervise its progress, and the societies emphasized objective investigation and professional training. The registry made poverty legible to institutions, and that legibility made applicants easier to deny.

Mary Richmond’s intake “face sheet” in Social Diagnosis (1917) gave casework its first standardized data-collection instrument, a direct ancestor of today’s case files and risk-assessment intake forms. Richmond’s larger method, investigating a client’s circumstances, relationships, and history before deciding on intervention, established structured fact-gathering as formal casework technique.

The debate over what this lineage became is live. Dorothy Roberts, reviewing Eubanks, argues that big data, predictive analytics, and automated decision-making now organize a “digitized carceral state” reaching beyond prisons into policing, public assistance, health care, and child welfare, managing the very inequalities it helps produce. She closes by arguing for an abolitionist response to these systems.

The mapping tradition

The Hull House maps and Du Bois’s data portraits used detailed data about communities experiencing poverty to argue that conditions cause poverty and that conditions can be changed. Hull-House Maps and Papers anchored its hand-colored wage and nationality maps with essays by Florence Kelley, Jane Addams, and others on sweatshop labor, child labor, and the neighborhood’s ethnic enclaves. The Library of Congress holds Du Bois’s surviving set: roughly 60 charts drawn in pen, ink, and wash on poster boards about 71 by 56 centimeters, plus several hundred photographs. The exhibit’s numbers documented change, Black-owned property in Georgia valued at 13.4 million dollars by 1899 and literacy up from 1 percent in 1860 to roughly 50 percent by 1900. Klein’s chapter argues that Du Bois paired charts with photographs because, after the 1899 lynching of Sam Hose, he judged that statistics alone could not convey the full truth of Black life.

They are early American evidence that careful data in reformers’ hands moves policy, and they are the ancestors of every community dashboard and equity audit. Both projects have present-day descendants. Mapping Inequality has digitized the 1930s federal redlining maps into more than 150 interactive maps used by about 2.5 million people since 2016, and the Eviction Lab built the first national database of evictions from more than 80 million court records.

Technology as public infrastructure

A second lineage rarely gets filed under “tech for good”: public libraries, rural electrification, sanitation, and the postal service. Private utilities charged rural customers rates about four times city rates, and the United States lagged Europe in rural electrification because power companies refused to serve farms. The Rural Electrification Administration, created in May 1935, lent money to local cooperatives that ran their own lines; by 1939, 288,000 households drew their electricity from rural electric co-ops.

The settlement houses also campaigned on sanitation. Jane Addams recounts that during one August and September the Hull-House Woman’s Club filed 1,037 substantiated reports of garbage-law violations with Chicago’s health department, and that after she bid for the ward’s garbage-removal contract, the episode “induced the mayor to appoint me the garbage inspector of the ward” at a salary of a thousand dollars a year. She got up at six to see the crews start work, followed the loaded wagons to the dump, and reported that the ward’s death rate fell from third to seventh among the city’s wards. In New York, Nathan Straus opened a pasteurized milk laboratory in 1892 and set up milk depots that distributed the laboratory’s milk and infant formulas in poor areas of the city, a model he later carried to other American cities and abroad.

Frances Perkins, a settlement-trained social worker who spent what time she could at Hull House and rushed to the Triangle Shirtwaist factory fire as it burned, drove the Social Security Act of 1935 into existence, which required building one of the largest administrative information systems in the world.

That system ran on punch cards. The Social Security Board could not find a Washington building able to carry the weight of its 400 accounting machines, so it moved its records operation into the Candler Building, a twelve-story structure Coca-Cola had put up in Baltimore in 1912; 2,500 employees were at work there by January 1, 1937, most of the machine operators women, and the agency stayed until 1960. IBM won the contract on September 16, 1936, was soon producing five to ten million punch cards a day, and supplied 1,200 keypunch machines along with new machines developed for the job, including the Type 77 collator that merged punched cards into a single ordered file.

The profession also built the human infrastructure on the telephone network: crisis lines, the information-and-referral services that became 211, and the lineage running to 988. The FCC set 211 aside as the national social-services number in 2000, and the network now answers more than 13 million requests for help a year through more than 200 local agencies, most about housing, food, transportation, and health care; in 2025 it made more than 19 million connections to help, resources, and services. On the crisis side, the country’s first suicide prevention center opened in Los Angeles in 1958, founded by Norman Farberow and Edwin Shneidman out of research begun at the county coroner’s office, with a round-the-clock telephone line answered by trained volunteers and staff. Draper and McKeon trace the line from mid-century volunteer hotlines through the 2001 federal funding that built a single national suicide-prevention number, Spanish-language service in 2006, the Veterans Crisis Line in 2007, and the 2018 law that created the three-digit 988 number, launched in 2022.

This lineage has its own scholarship, separate from the critical canon. Honig calls such shared objects “public things”, the infrastructure democracy needs to hold citizens in relation to one another; her argument is that neoliberal politics singles out state bureaucracy as wasteful and treats privatization as a self-evident cure. Klinenberg’s work on social infrastructure documents how much community life runs on it. His evidence begins with Chicago’s 1995 heat wave, where neighborhoods with stronger social infrastructure had markedly lower death rates, and includes a Philadelphia study in which cleaning and maintaining vacant lots cut nearby gun violence by nearly 40 percent.

Social work was doing public interest technology a century before universities coined the term. When Slaughter, Walker, Kramer, and Bray argued in 2019 for building the field, the new Public Interest Technology University Network had 21 member institutions, and Schneier compared the effort required to the foundation funding that built public-interest law starting in the 1960s. Still, historians of the phrase point out that each era’s “public interest” served a particular public; their proposed anchor for the field is acknowledged conflicts of interest and the centering of marginalized populations.

Early computerization debates

When computers reached social services agencies in the 1970s, case records were among the first things loaded onto them. British departments started by borrowing time on their councils’ existing machines, an arrangement that, in Bryan Glastonbury’s 1985 practitioner handbook Computers in Social Work, “so blatantly set up the possibility of data sharing, of allowing other departments and agencies to get hold of confidential case files.” Purpose-built systems followed; the best known of the early ones, the SOSCIS client information system, was developed in Gateshead between 1974 and 1977.

Glastonbury’s chapter list includes “The dehumanising computer” and “The Robot Social Worker,” and he reports that workers’ worries about confidentiality and file security grew to parallel, perhaps exceed, their fears of managerial control. His own judgment was that the computer had been “unfairly blamed,” since the reorganization that created large social services departments had already widened authorized access to case files. On dehumanization he argued that peripheral use carried the greater risk, because a worker who only feeds a management information system gets software written for managers, while heavier use pulls programs toward practice needs.

The resistance had a reputation to match. Jan Steyaert’s history opens by naming it, that social work is seen as “averse of technology, of being Luddites,” even though the profession took up the telephone early in the century and the personal computer late in it, carrying both a fear that privacy and warm contact would erode and a nervous excitement about information at the fingertip. Jonathan Singer and Jimmy Young tell that resistance story through Walter LaMendola, who earned one of social work’s first information-systems doctorates at the University of Minnesota in 1976 and went on to direct the first information-technology center at a school of social work, opened at the University of Denver in 1985.

The confidentiality worry still shapes what enters the record. A 2022 qualitative study in Digital Society found social workers deliberately leaving sensitive information out of client information systems or recording it only as broad summary, treating themselves as gatekeepers for clients’ privacy; the same omissions degrade the data that algorithmic decision systems are built and run on.

The profession’s computing turn

The registries and the reform maps were built by hand. Only decades later did social work start building computing infrastructure of its own. The network that did it formed in 1981, when human-service technology specialists met at a CSWE conference in Louisville, Kentucky. That year Dick Schoech, a social work professor at the University of Texas at Arlington, began editing its quarterly newsletter for agency staff and faculty comparing notes on hardware, software, and early applications. The Computer Use in Social Services Network newsletter (CUSSN) reached members worldwide, roughly four issues a year, and ran until 1992. By the end of 1981 the network had more than 350 members, and in 1985 it added CUSSNet, a bulletin-board network built on FidoNet that exchanged members’ messages overnight during off-peak telephone hours, with nodes in American cities, the UK, and the Netherlands. The husITa mailing list still running today started in 1985 as CUSSNet.

The online story ran ahead of the web. Linda Grobman and Susan Mankita, who took part in much of it, date the milestones: a human-services bulletin board built by Gerald Vest and Mike Connealy in New Mexico in 1986; the SOCWORK listserv that Simon Mielniczuk started in 1988 so students in a University of Toronto planning class could talk, which grew to roughly 1,600 members before flame wars and traffic above a hundred messages a day thinned it; the BPD-L listserv that Marshall Smith opened in November 1994 and that had 37 subscribers within a month; and the AOL Social Work Forum that Susan Mankita launched in 1995, whose first live chat one Thursday night in May drew seven people and which NASW took over in the mid-2000s.

By then the field had its own journal. It launched in 1985 as Computers in Human Services, with Schoech as founding editor, and still publishes today as the Journal of Technology in Human Services. Schoech’s own retrospective treats the newsletter as the journal’s direct ancestor and reviews the field’s growth from 1980 forward. Two years after the journal launched, on September 11, 1987, delegates from 17 countries meeting in Birmingham, England formally established husITa, an international body for human-service computing.

The push to formalize the field kept going. In 2006, writing in NASW’s own journal Social Work, Parker-Oliver and Demiris argued for a dedicated specialty they called social work informatics, naming technology competence as its own area of practice. In 2017, four professional bodies, NASW, ASWB, CSWE, and CSWA, issued joint standards for technology in practice: 27 standards in four domains, developed over roughly two years of literature review and public comment. Within the Grand Challenges for Social Work, the Harness Technology for Social Good challenge, run by seven co-leads across universities, aims to use digital technology to “reach more people with greater impact on our most vexing social problems”; its network issued policy recommendations in 2016 and again in 2024, with a five-year impact report in January 2021. As AI reaches practice, the profession has started measuring its own adoption; a 2026 national survey by the Moritz Center and NASW gathered responses from 1,179 social workers on AI use, perceptions, and preparedness. Reamer’s companion analysis centers informed consent, third-party data flows, algorithmic bias, and the displacement of professional judgment, and recommends ethics committees and algorithm peer review as safeguards. Week 8 picks up the industry structure that grew around that need in the decades since.

The sociotechnical questions protocol

Who designed it? Whose problem definition? Whose data and labor? Who bears the risk of failure? What social arrangement does it lock in? The protocol adds three conditions-of-success questions for the wins. The tool operationalizes what fairness researchers call the abstraction traps, the errors that follow from evaluating a technical system apart from its social setting. Selbst and colleagues name five: the framing, portability, formalism, ripple effect, and solutionism traps.

The Allegheny Family Screening Tool is the artifact these questions get run on, and it stands at the current end of the registry lineage. It has scored child-neglect referrals on a 1-to-20 scale since August 2016, predicting whether a child will enter foster care within two years, using birth, Medicaid, substance abuse, mental health, jail, and probation records. An ACLU audit found it scores whole households as risky by association even when individual members score low, draws on criminal-justice and disability-service data that disproportionately implicates Black families, and rests on design choices its own developers describe as “rather arbitrary.” Carnegie Mellon researchers found that between August 2016 and May 2018 it flagged 32.5 percent of Black children reported for neglect for mandatory investigation, against 20.8 percent of white children, and that social workers disagreed with its scores about a third of the time. For more than two years, a technical glitch showed workers the wrong scores.

The one-pager lives on the protocol page.

Before class

In class

Seminar

Vote, then the two lineages: ledger to algorithm, map to dashboard. We crosswalk three social work values (dignity and worth of the person, self-determination, social justice) onto the nineteenth-century fight between scientific charity and the settlement movement, a fight that reads as current once the vocabulary is updated. The seminar ends with the infrastructure inheritance, including Mapping Inequality’s digitized 1930s redlining maps where federal appraisers graded neighborhoods by race in their own notes, and the revote.

Lab: the protocol, run twice

Your group gets a paired artifact set: either Richmond’s 1917 face sheet and the COS registry alongside the Allegheny Family Screening Tool (with the county’s own materials and the critics side by side), or the Hull House wage maps and Du Bois portraits alongside the Eviction Lab’s national eviction map. You run all five protocol questions on both artifacts, add the conditions-of-success questions, then ask an AI assistant to critique one row of your grid and mark, against the primary sources, what it got wrong or thin. You leave with a completed comparison grid. In the last five minutes, the class opens the Problem Inventory, a shared document where you log real service problems you have seen firsthand; describing one well, whose problem it is and who carries the difficulty, is the protocol’s first unsupervised use.