There is a number I keep returning to, the way a tongue keeps finding a missing tooth. In the 2023 financial year, the money American jurisdictions claimed from the federal government to keep children out of foster care came to less than two percent of everything they claimed under Title IV-E, the largest federal child-welfare funding stream in the United States. It is the finding of the US Department of Health and Human Services’ own policy research office, published under a title that reads like a confession: prevention makes up less than two percent of claims, and many states do not claim at all. Less than two cents in the dollar for the front door. The rest for everything that happens once the child is through it. I want to handle that sentence carefully, because it is strong enough to be misused. The other ninety-eight cents is not a fund for taking children. It pays for foster care maintenance, caseworkers, training, administration, and assistance to adoptive and guardianship families. Some of it pays for reunification. A slogan that calls all of it “removal money” would be false, and a false number is a gift to the people who would rather this conversation never happened. So here is the precise version, and it is worse than the slogan. Where the guarantee attaches Title IV-E is an open-ended entitlement. If a state does the paperwork, Washington reimburses a share of the cost, without a cap. For most of its history that guarantee attached to one moment: the moment a child entered care. Before that moment, the family’s rent, the mother’s treatment, the childcare that would let her keep a job, sat outside the ledger or in smaller, capped side channels. In 2018 Congress passed the Family First Prevention Services Act, and for the first time opened that entitlement to services meant to stop children entering care in the first place: mental health treatment, substance use treatment, in-home parenting programs, kinship navigation. Then look at what happened next. By the end of FY2023, sixty percent of states, territories and Tribes had never submitted a single prevention claim. Of the 95 programs the federal Clearinghouse had rated as evidence-backed by late 2025, fewer than a third appeared in any approved state plan. Spending on prevention grew roughly twenty-fold from 2020 to 2023, and still came to under two percent of the whole. About two percent of the children served through Title IV-E were served through prevention. Then came the detail that changed how I read the story. In February this year, two of the most senior officials at the Administration for Children and Families wrote publicly that the budget office had projected prevention would reach only five percent of Title IV-E spending a full decade after the law passed. Read that slowly. The two percent is not the reform failing. The reform was costed, at birth, to stay small. The architecture was always going to keep the money where it already lived. Where the money lives, and how it grows A companion federal brief tracks the other side of the ledger. Between FY2019 and FY2023, administrative costs became the largest driver of Title IV-E foster care claims, and the average monthly administrative cost per child rose more than forty percent. Over the same period, caseloads fell. Fewer children in care. More spent administering each one. The upstream is being taken apart Head Start is not child-welfare funding, and it should not be confused with it. It is early learning, health screening, family support and case management for low-income families with children under five. If the federal government has an upstream, this is it. It is shrinking, and the shrinkage is not an accident of demand. Head Start served 771,121 children and pregnant women in FY2025. In FY2016 it served 1,067,965. That is a fall of twenty-eight percent in under a decade. Between FY2021 and FY2024, requests from local programs to cut their own funded enrolment rose from 39 a year to 746. Programs were asking to serve fewer children because the money no longer stretched to the children they had. Then came 2026. In June, HHS proposed rescinding the 2024 rules that required Head Start to move its teachers toward pay and benefits comparable to public schools, a change opponents in Congress said would have lifted most teachers’ pay by around $10,000 a year. In August, HHS went further. A 200-page proposed rule titled “Reducing Federal Burden for Head Start Programs” would rescind and replace the Head Start Program Performance Standards, the rulebook that defines what a Head Start program has to provide. HHS presents it as a return to the program’s roots: less bureaucracy, more trust in parents and communities, more places and $2.2 billion saved a year. Its Assistant Secretary lays the record fall in places at the feet of the previous administration. Authority over staff ratios, group sizes, staff education requirements, background checks and transport would pass to the states. The cap on what a program may spend developing and administering itself would fall from fifteen percent of total cost to five. Read the rule itself and the changes are specific. By its own account it rescinds more than 1,400 provisions. Centre-level parent committees become optional. All teaching must be in English, with an exemption for tribal programs. The protections that limit suspending or expelling small children for their behaviour are removed. The regulatory minimum hours for Head Start preschool go, leaving only the statutory floor of three hours a day. The regulations on health services, oral health, mental health support, staff qualifications and staff training are removed, surviving only where the Head Start Act itself still requires them. Parent organisers at United Parent Leaders, who walked families through the comment process, warned that the rules keeping children supervised at all times and training staff to prevent abuse and neglect would go with them. Mental health support, health screening, trained staff: that is prevention in its most literal form, the work done before any caseworker is called. The suspension clause matters more than it looks. A three-year-old expelled from preschool does not disappear. The family loses its childcare, often a job with it, and the child’s distress travels on to the next institution that notices it. And note the contradiction at the centre of the government’s own case: a rule that calls parents their children’s primary teachers makes optional the committees through which parents help run their centres. One more number deserves checking. HHS’s press release promises as many as 236,000 places. The rule’s own regulatory impact analysis projects about 162,000 by 2031, and only on the assumption that Congress adds no new money. The headline and the fine print do not agree. Connecticut’s governor and attorney general, Wisconsin’s governor, members of Congress and unions called for withdrawal. Connecticut’s submission named children in foster care among those who would be hurt. Public comment, on federal docket ACF-2026-0595, closed two days ago. Hold those two administrative numbers side by side. Downstream, the cost of administering each child in foster care rose more than forty percent in four years, and that cost is reimbursed. Upstream, the program that keeps families from ever meeting a caseworker is being told to run itself on five cents in the dollar. I am not claiming the upstream contraction causes foster care entries; the data does not yet exist. I am describing a shape: a prevention channel designed to stay thin, between an upstream being narrowed and a downstream growing more expensive per child. The hum underneath the spreadsheet I did not learn this from spreadsheets first. I learned it in my body, decades before I could name it. I was separated from my mother by a system that paid for every part of the separation: the institutions, the officers, the forms, the court, the sealed record, the amended birth certificate. It did not pay for the one thing that might have let her keep me. I carry that missing line item in my nervous system. It sits under everything, a low, steady pressure that knew something was absent long before any document admitted it. People sometimes ask why a forced-adoption survivor spends his nights reading American reimbursement data. This is why. Every budget is a statement about which children a state is prepared to pay for, and at what point in their lives. I know which side of that statement I was born on. The loop the money draws Thoughtless Delineation’s working framework for this pattern is what I call the Analog Fractal Feedback Recursion of Adoption: harm that transmits not as a clean copy but as a signal, degraded enough at each step that every actor can deny it is the same pattern, while the shape survives intact across family, institution and state. The step that matters here is feedback, where the wound becomes the alibi. A system that funds the consequence of separation will read the volume of consequences as proof of need. More children in care becomes an argument for more care capacity. Rising cost per child becomes an argument for more administrators. The output of the machine is fed back in as its justification, and nobody in the chain needs to intend anything for it to keep running. The federal government’s own research shows the loop at its most literal. ASPE has a dedicated line of work on custody relinquishment: children who enter foster care primarily to receive behavioural health or disability services, not because anyone harmed them, because those services were not available any other way. In some places, the price of getting a child care is handing over the child. This is also what the publication calls a Maintenance Economy: not a conspiracy of villains, but a reimbursement structure that reliably pays for the aftermath and only reluctantly pays for the alternativ
