The cascade of rigidity is a term used by Jennifer Pahlka for the phenomenon in which a well-written top-level law or policy becomes progressively more rigid, literal, and counterproductive as it descends the bureaucratic hierarchy toward operational implementation. Pahlka introduced the term in "AI Meets the Cascade of Rigidity" (Digitalist Papers Vol. 1) and elaborated the underlying dynamic in her 2023 book Recoding America. Pahlka argues that each approval layer strips out discretion and judgment, converting flexible guidance into rigid checklist compliance, and that this dynamic is central to why government AI adoption is failing.
Mechanism
Pahlka describes the cascade as a recurring structure with five stages:
- Law or guidance at the top is written with appropriate flexibility — "agencies may choose controls relevant to the circumstances," "HR should ensure non-discrimination."
- Multiple approval layers below require sign-off, often from non-domain-experts who lack the technical knowledge to assess discretion.
- Reviewers face asymmetric incentives: a penalty if they approve and something goes wrong, but no penalty for blocking. They therefore default to maximal compliance.
- Any single thumbs-down kills the project, while approval requires unanimous sign-off — the vetocracy dynamic.
- The implementation version of the guidance becomes rigid and defensive, in Pahlka's account often worse than no guidance at all, as literal checklists replace flexible judgment.
Case studies
Pahlka draws her two principal illustrations from federal security compliance and federal hiring.
FISMA security controls
The Federal Information Security Management Act (FISMA) provides roughly 300 security controls as a menu for agencies to choose from based on project needs. In Pahlka's account, government tech teams are in practice forced to implement all 300 because compliance officers will not risk approving anything less. The result is that development timelines balloon by months or years, testing is compressed, and actual security degrades because resources are diverted from relevant controls toward checkbox completion (The Digitalist Papers (Stanford, Volumes 1–2)).
Civil service hiring (Pendleton Act, 1883)
The Pendleton Act required merit-based federal hiring. Pahlka describes a current practice in which only HR professionals may assess candidate resumes, even for technical roles, to minimize bias risk, so that subject-matter experts cannot participate until final interviews. She cites the case of Jack Cable, a Defense Digital Service contest winner who had beaten 600 security researchers but was rejected in the first HR screen because his resume listed "IonicJS, Angular, Node.js, MongoDB, npm, Express gulp, Babel" and the HR reviewer saw "a grab bag of gobbledygook." Cable was told to work at Best Buy for "a few years" before reapplying, and was hired only after high-level intervention (The Digitalist Papers (Stanford, Volumes 1–2)).
Pahlka notes that the US Digital Service and the Office of Personnel Management have piloted subject-matter-expert-led candidate assessment, which she reports produces better hires, but that the practice remains uncommon because "many in government believe it to be illegal" when, she argues, it is not (The Digitalist Papers (Stanford, Volumes 1–2)).
Relation to AI policy
Pahlka presents the cascade of rigidity as an explanation for why government AI adoption is failing. The USAi platform and other government AI pilots encounter compliance layers optimized for a pre-AI era, and AI-specific guidance — such as Biden's Executive Order 14110 before its rescission and Trump's AI executive order — layers on top of existing FISMA, FedRAMP, and authority-to-operate (ATO) processes. In Pahlka's analysis, every new AI mandate worsens the cascade even when each mandate is individually reasonable (The Digitalist Papers (Stanford, Volumes 1–2)).
Pahlka's prescription is summarized in the phrase "culture eats policy": flexible top-down guidance is, in her view, insufficient on its own, and government must instead invest in capacity building (technical hiring, enabling infrastructure, and technical leaders at every level), competency growth (trusting subject-matter experts with discretion), and default-enabling infrastructure (authority to run experiments, not only authority to block them) (The Digitalist Papers (Stanford, Volumes 1–2)).
Contrast with traditional reform frameworks
Pahlka positions the cascade of rigidity as orthogonal to the regulation/deregulation axis. Conventional reform debates target either a perceived gap in constraints (regulation, adding rules) or a perceived excess of constraints (deregulation, removing rules). The cascade of rigidity instead names the implementation-layer dynamic that, in Pahlka's account, makes both reforms underperform expectations: both conservative "cut red tape" and progressive "add protections" reforms fail when the bureaucracy reinterprets them maximally at the point of implementation.
| Framework | Reform target |
|---|---|
| Regulation (add rules) | Perceived gap in constraints |
| Deregulation (remove rules) | Perceived excess of constraints |
| Cascade of rigidity | The implementation-layer dynamic that makes both reforms underperform expectations |
Relationships
- supports: The Digitalist Papers (Stanford, Volumes 1–2) — Pahlka's essay.
- supports: Vetocracy — cascade of rigidity is vetocracy within the administrative state.
- supports: Government AI Procurement — procurement is the acutest failure mode.
- related: GSA OneGov Program and USAi Platform (August 2025) — US government AI-adoption effort navigating cascade dynamics.
- related: America's AI Action Plan — Trump EOs partly address rigidity via deregulation, but Pahlka's critique implies they underperform without a capacity-building complement.
- related: DOGE — Department of Government Efficiency (AI Deployer) — DOGE's efficiency-through-reduction approach is the opposite of Pahlka's capacity-building approach.