logan koepke
LightNews LightNews
banner
jlkoepke.bsky.social
logan koepke
@jlkoepke.bsky.social
940 followers 1.6K following 330 posts
senior project director @upturn.org. i work on AI/ML + civil rights research. personal views.

www.jlkoepke.com
Posts Replies Media Videos
jlkoepke.bsky.social
(separately, something that has confused me: assuming for argument sake that the proposals really *were* duplicative [they are not], what is the harm? presumably they would just … be duplicative? why worry?)
November 22, 2025 at 1:02 PM
jlkoepke.bsky.social
one can critique state algorithmic discrimination proposals for many things — including if they will materially advance anti-discrimination!

but suggesting they are duplicative of existing civil rights law shows an unfamiliarity with what civil rights law does and doesn’t do.
November 22, 2025 at 1:02 PM
jlkoepke.bsky.social
1) such a view imagines civil rights law as fixed. but of course, the Trump administration is dismantling civil rights law.

2) existing civil rights law does not cover all relevant actors! Title VII and ECOA do not cover *all* entities that are relevant to anti-discrimination.
November 22, 2025 at 1:02 PM
jlkoepke.bsky.social
read the report here: www.upturn.org/reports/2025...

www.upturn.org/work/calcula...
Calculated Need: Algorithms and the Fight for Medicaid Home Care
Upturn advances equity and justice in the design, governance, and use of technology.
www.upturn.org
October 28, 2025 at 5:15 PM
jlkoepke.bsky.social
thanks in part due to public records on how these systems are designed + implemented, the report documents how automated eligibility determinations are political tools that draw a *subjective line* between who is deserving and undeserving of getting care in their home
October 28, 2025 at 5:15 PM
jlkoepke.bsky.social
the result: someone eligible for home and community-based services in Nebraska and Mississippi but ineligible in New Jersey, Missouri, and DC.
Report text begins with Table 1.1. Comparison of Eligibility for Example HCBS Participant in Missouri. Source: Our LOC algorithm comparison is based on algorithms we obtained through public records requests and documents released by state agencies. See Appendix A. To illustrate how different the scoring process can be across states, we examine and compare the use of the interRAI HC assessment in LOC algorithms in five states: Mississippi, Missouri, Nebraska, New Jersey, and Washington, DC.60 To understand how different choices state agencies and vendors make when designing an LOC algorithm can result in different eligibility determinations in different states, let’s look at an example scenario. This example uses anonymized assessment data from someone who was receiving home and community-based services in Missouri and was then found ineligible when Missouri changed its LOC algorithm. According to their interRAI HC assessment, this person needs limited assistance with personal hygiene and dressing their lower body and needs set-up help for bathing, dressing their upper body, walking, and going to the bathroom. They need significant help with housework and shopping. They are incontinent, have consistent balance issues, and have severe pain and fatigue. They have been diagnosed with chronic obstructive pulmonary disease, congestive heart failure, coronary heart disease, diabetes, anxiety, and depression, and have had a stroke. This person had been receiving services in Missouri for years and expressed a desire to continue to receive services so that they could stay in their home. Based on the same exact responses on the interRAI HC assessment, this person would be eligible for home and community-based services in Nebraska and Mississippi but ineligible in New Jersey, Missouri, and DC, as shown in Table 1.1.
October 28, 2025 at 5:15 PM
jlkoepke.bsky.social
but states use these assessments *very differently.*

when designing their level of care algorithms, state agencies and vendors may choose different subsets of assessment questions to include in eligibility determinations and assign different weights or scores to those questions.
Report text: LOC algorithms: Scoring the assessment to make an eligibility determination The assessment on its own does not provide a determination about whether someone is eligible for home and community-based services under their state’s program. The way states use the assessment as input to determine HCBS eligibility varies significantly. When designing their LOC algorithms, state agencies and vendors may choose different subsets of assessment questions to include in eligibility determinations and assign different weights or scores to those questions and responses. As a result, even when states use the same assessment, their LOC algorithms can arrive at completely different conclusions. Simply knowing which assessment a state uses does not provide sufficient information to know how the institutional level of care requirement is operationalized. Despite the algorithm being crucial to understanding how the eligibility determination is made, states and vendors obscure their use of LOC algorithms and make them difficult to obtain.
October 28, 2025 at 5:15 PM
jlkoepke.bsky.social
standardized assessment is the first part of the eligibility determination process. the interRAI Home Care assessment is used in more than 25 different states. Since 1990, >100 million interRAI nursing home assessments have been completed in the US.
Report text: The most widely used assessment for HCBS programs for elderly people and people with physical disabilities is the Home Care (HC) assessment created by interRAI, a vendor of assessments and algorithms that has been involved in automated HCBS determinations since the 1990s.50 As of 2024, the interRAI HC assessment is used in more than 25 different states.51 Over 100 million interRAI nursing home assessments have been completed in the US since 1990.52 Some states modify the interRAI HC assessment or other out-of-the-box assessments, and some design their own custom assessments.53 In this report, we focus on the interRAI HC assessment because it is widely used, including by the five states in our LOC algorithm comparison. Much of this analysis can be extended to other assessments and the interRAI HC is not unique in the issues it poses.
October 28, 2025 at 5:15 PM
jlkoepke.bsky.social
while this is a federal program, each state is responsible for designing its own rules and systems to determine which of its residents are eligible for HCBS benefits and how much home care they should receive.

every state uses automated tools to help determine HCBS eligibility
Report text: Though it is federally mandated, the institutional level of care is not defined at the federal level. Every state defines its own level of care requirement through legislation, rulemaking, and the development of a LOC algorithm, typically built by private vendors. Our own and other research show wide variation in how level of care is defined and operationalized across states.25 This means, for example, that someone living in New Jersey might be eligible for home and community-based services in that state, but if they were to move to Washington, DC, they might not be. This system severely limits people’s ability to live where they want to live and still get the care they need. State Medicaid agencies frame institutional LOC algorithms as tools that ensure the “right” people get approved for home care, based on a standardized measure.26 In this process, measuring whether someone meets the state’s definition of the level of care is equated with whether that person needs home care. In reality, however, level of care is a misnomer. LOC algorithms do not measure how much care people need but instead enforce the state’s definition of who deserves to receive care in their home. Using standardized assessments and LOC algorithms does not make deciding who should get home care objective or fair. It merely obscures what is actually a subjective and political exercise behind claims of technical expertise and the opacity of proprietary interests.27 In the process, the use of standardization and automated tools depoliticizes the question of who deserves care and hinders advocacy.
October 28, 2025 at 5:15 PM
jlkoepke.bsky.social
millions of elderly people + people with disabilities rely on Medicaid Home and Community-Based Services to meet their daily needs (from help bathing, getting dressed, and eating meals) as well as to work + participate in their communities.
Report text: Disability benefits programs in the United States, including Medicaid Home and Community-Based Services, provide essential — though too often inadequate — support for people with disabilities. These programs often treat people with disabilities as untrustworthy and undeserving, as evidenced by restrictive and burdensome eligibility processes, severely limited benefits, and an emphasis on identifying individual fraud. These and other benefits programs are designed around the assumption that people are going to try “to take advantage” of the system — a racist and ableist idea that has deep social and historical roots in the United States.17 Benefits programs are also a product of class struggle.18 They were concessions won by workers and organizers for the benefit of all of society, and their systematic dismantling is an attempt to break the power of the working class and oppressed.19 In July 2025, Congress passed sweeping Medicaid cuts that will strip away essential health care benefits from millions of people to fund the Trump administration’s tax cuts for the rich.20 While we still don’t know exactly how these cuts will affect home and community-based services Report text: history has shown that they are a likely target.21 This significant attack on Medicaid coverage is part of a decades-long process — outlined, but by no means begun, in Project 2025 — to keep poor and working people barely at the threshold of survival, while the ruling class pockets the wealth of society.22 Welfare cuts are just one part of the story. Further limiting who is eligible for benefits programs is one of the many ways these programs have been scaled back over time. State and private actors have built an extensive infrastructure of tools and technology to determine access to disability benefits. This infrastructure obscures the political decision to deny the right of people with disabilities to participate in society and replaces it with technocratic questions about measuring who should get benefits and how much they should get. This report focuses on one example of this infrastructure: eligibility determinations for Medicaid Home and Community-Based Services for elderly people and people with physical disabilities. HCBS programs are the primary provider of paid home care for elderly people and people with physical disabilities in the United States. Access to Medicaid Home and Community-Based Services enables people to live in their communities instead of in a nursing facility or other institution. Starting in the 1980s, states began expanding access to home care, driven by changes in federal policy.23 The disability rights movement was crucial to creating the necessary pressure on legislators to invest in home care for elderly people and people with disabilities. At the same time, federal policymakers were concerned about the growing costs of long-term care in institutions. Home care offered the promise of lowering those costs for the government, while also responding to grassroots pressure for these changes.
October 28, 2025 at 5:15 PM
Reposted by logan koepke
upturn.org
3/ There’s no escaping the fact that these automated Medicaid eligibility determinations are arbitrary and subjective: based on our analysis, someone who is eligible for home care in Mississippi and Nebraska could be *ineligible* in Missouri, New Jersey, and DC.
Calculated Need: Algorithms and the Fight for Medicaid Home Care
Upturn advances equity and justice in the design, governance, and use of technology.
www.upturn.org
October 21, 2025 at 6:26 PM
Reposted by logan koepke
upturn.org
2/ Our report offers an in-depth analysis of five states’ eligibility algorithms, case studies on how states work with vendors to develop their algorithms, and historical research on the emergence of the federally mandated eligibility standard for Medicaid home care.
Calculated Need: Algorithms and the Fight for Medicaid Home Care
Upturn advances equity and justice in the design, governance, and use of technology.
www.upturn.org
October 21, 2025 at 6:26 PM