Judge rules $400M algorithmic system illegally denied Medicaid benefits
A federal judge has ruled that Tennessee’s $400 million Deloitte-built TennCare eligibility system illegally denied Medicaid coverage to large numbers of people, highlighting how automated decision-making can wrongfully block access to critical healthcare. Commenters question the opaque role of major consulting firms in government IT, call for open-source and auditable public-benefits software, and debate how to balance transparency with medical privacy. Many argue the underlying problem is political—complex rules and perverse incentives to restrict benefits—pointing to similar scandals abroad and suggesting simpler, universal programs as a more humane alternative.
System Failures & Legal Ruling
- The TennCare Connect eligibility system reportedly mishandled data, mis-assigned households, and made incorrect Medicaid eligibility decisions.
- Commenters stress that when people are legally entitled to coverage, making it depend on “luck, perseverance, and zealous lawyering” is unacceptable.
- Many emphasize state responsibility: the system was approved, procured, and deployed by the state, not just the vendor.
Transparency, Appeals & Due Process
- Some note that denial letters must state reasons, but describe them as opaque, intimidating, and hard to challenge without lawyers.
- There is concern that arbitration and bureaucratic friction effectively block many from contesting wrongful denials.
- Suggestions include legal rights to demand the data and logic used in decisions and better audit trails.
Open Source, Auditability & Privacy
- Strong support for making government-funded software open source to increase accountability and avoid vendor lock-in.
- Some propose tamper-proof logs or blockchain-like systems where rules and decisions can be replayed; others call this overkill and impractical.
- Multiple commenters push back hard on publishing claim-level data (even “anonymized”), arguing re-identification risk is real and unacceptable.
- Debate over whether broader data access for research and oversight is worth exacerbating already-bad privacy leakage.
Government Contracting & Incentives
- Widespread skepticism about large consultancies: seen as expert at winning big contracts, not at building good software.
- Observations that governments often lack in-house technical expertise and can’t pay market rates, making them dependent on such firms and unable to evaluate quality.
- Some argue contractors partly serve as political “fall guys” for policies designed to restrict benefits.
Policy Design & Welfare Philosophy
- Many argue the core problem is complex, punitive eligibility rules, not just bad code.
- Recurrent theme: it might be cheaper and more humane to provide universal or baseline benefits (public insurance, UBI) than to spend hundreds of millions screening people out.
- Comparisons made to criminal-justice standards: better to err on the side of helping some ineligible people than denying eligible ones.
Human Impact & Comparisons
- Personal stories from Tennessee describe multi-year battles for coverage, high denial rates, and suggest some denials contributed to suicides and worse long-term health costs.
- Comparisons to Florida’s intentionally unusable unemployment system and Australia’s “robodebt” scandal; both cited as examples of automated systems harming vulnerable people and evading accountability.
Reform Ideas
- Proposals include:
- Building strong in-house or public-interest tech teams (e.g., US digital service models, Code for America style).
- Simplifying laws so eligibility is easy to encode and verify.
- Independent validation and regulation of algorithms used in public programs, similar to model validation in banking.
- Recommended readings like “Automating Inequality” and “Recoding America” for deeper structural analysis.