## Notes from 13 April 2026 [[2026-04-12|← Previous note]] ┃ [[2026-04-14|Next note →]] # On institutional memory and preparing for the future I came across a document I hadn't seen before: **"[Building a More Effective, Responsive Government: Lessons Learned from the Biden-Harris Administration](https://rooseveltinstitute.org/publications/building-a-more-effective-responsive-government/)"**, published by the [[Roosevelt Institute]] in October 2025 by **[[Hannah Garden-Monheit]]** and **[[Tresa Joseph]]** — former senior advisors during the [[Biden Administration|Biden]] years. It’s a structured **after-action review** (a candid “what worked/what didn’t” assessment) based on interviews with over 45 senior officials, synthesized into 161 actionable recommendations. Honestly, the document itself might be more interesting than any single proposal within it. This kind of structured, interview-based institutional memory exercise is rare in any government. Transition memos and government plans (or coalition agreements) tell you what to do; this one tries to capture how the machinery actually failed, and does so in an organized, usable way. We need much more of this. Ideally, something even richer: an oral history project, recorded testimony—something that preserves not just the lessons, but the texture of how people experienced the work. Even having it written down and made public like this is remarkable. Most political groups just let this knowledge walk out the door. In this case, the core argument is that the Biden administration had the policy ambitions but lacked the institutional infrastructure (hiring pipelines, legal risk tolerance, execution sequencing) to deliver them within the speed that political windows actually require. Here are some of the most interesting proposals: --- ### Reclaiming time and focus The **[[Congressional Implementation Office]]** is one of those ideas that's obvious in retrospect: a CBO equivalent that scores legislation not for fiscal cost, but for _implementation time_. How long until constituents actually feel it? It factors in legacy IT, procurement timelines, and agency capacity. The CBO shapes political behavior precisely because it makes cost visible and legible; there's no reason the same logic couldn't apply to time. The **"[[Deletion Docket]]"** is the other side of the same coin: starting a new administration with a formal "to-stop" list before adding new priorities. This includes discontinued reports, low-impact legacy programs, and duplicative interagency processes. The instinct in government is always additive; this deliberately runs against it (much like the [[Department of Government Efficiency (DOGE)|DOGE]] concept I guess?). There's also a proposal for a **[[DARPA-style setups|DARPA-Main Street]]** variant: applying DARPA’s high-risk, high-reward model to everyday economic problems like supply-chain fragility and local manufacturing gaps, and doing so fast enough to catch short-lived opportunities rather than getting stuck in the usual federal layers. The idea of transplanting that logic from defense and frontier research into more “mundane” policy is interesting, but it **runs straight into a familiar paradox**: either you create [[Escape from Public Law|multiple carve-outs]] from procurement/HR/oversight rules (so you can actually move), or you try to transform those underlying systems themselves (so you don’t need carve-outs). --- ### Fixing the "front door" Several proposals tackle what the report calls service delivery—the experience of actually trying to get something from the government. The most elegant is the **data proxy** idea: multiple agencies already hold verified income, tax, and benefits data, so requiring citizens to resubmit this information to access programs they're already eligible for [[2026-03-18|is a design choice]], not a necessity. Automatic or near-automatic enrollment based on administrative data already exists in several OECD countries; it's not a radical idea. The **Federal Navigators** proposal — a permanent corps of generalist caseworkers who can guide people across agency jurisdictions — is a compelling idea. It’s pitched as a way to reduce the “bureaucratic tax” people pay when their needs cut across programs: not a single-issue navigator (healthcare, taxes, veterans), but a durable public-facing capacity that can _triage_, coordinate, and stick with a case as it moves between agencies and levels of government. > [!note] French analogue It reminds me of projects in France built around physical spaces like the **[[Maisons France Services]]**: catch-all hubs where citizens can access services in person (and online), with staff who help people navigate digital systems and figure out which office, benefit, or process applies. The model is an operational state network of over **2,700 physical hubs** staffed by trained generalists known as **“animateurs”**. > > These agents act as intermediaries between citizens and the bureaucracy, providing in-person assistance for digital procedures across nine core government agencies, including tax, health, and employment services. While official data shows high satisfaction rates and success in maintaining a state presence in rural or marginalized areas, the model's primary function is **digital mediation** (_lutte contre l’illectronisme_) rather than full legal representation. These staff members are trained to navigate portals and scan documents, but they do not replace specialized caseworkers for complex legal disputes or highly technical administrative decisions within the parent agencies. Finally, another interesting proposal is to record and show **user-testing experiences to enrich congressional debates**. The idea is almost mischievous: record real citizens failing to navigate federal benefit applications, then show the videos to the legislators who designed the systems. It's hard to argue with footage. --- ### On giving government more leverage to act The **"yes energy" GCs** proposal I find both right and slightly off. The diagnosis is accurate: agency lawyers being the first voice in the room rather than a later stress-test on a decision already made is a structural problem, and it systematically biases agencies toward delay. However, a GC whose primary mandate is getting to "yes" may produce approvals that collapse in court—which is worse than a cautious delay. A more precise framing: "yes energy" belongs to political leadership, and the GC's job is to make that "yes" legally bulletproof. Different roles, different defaults. The problem isn't that lawyers say no; it's that they're often consulted before anyone has decided what they're actually trying to do. **Marketing the win** is the most basic of the proposals and probably the most underrated. Government does good things and tells no one. Direct-to-constituent communication (a letter, an SMS, a notification) can clearly attribute a specific benefit to the action responsible. You recovered a pension; here's who did it and how. Much more money should go into "marketing for public services" to turn it into an applied science. However, differentiating this from glorifying the government of the day is a challenge the report doesn't fully engage with. There is also a proposal to “redesign the funnel” of judicial control over administrative policy: instead of allowing disputes over agency action (regulations, administrative decisions, etc.) to move up the chain and ultimately reach the Supreme Court, the idea would be to **create a new specialized federal court to review administrative action**. This would provide a technically prepared channel for reviewing administrative action. The backdrop is the assessment that judicial review has become a structural bottleneck for governments trying to implement economic agendas, as rules can be struck down after years of work and the Supreme Court is seen as having adopted doctrines that make the survival of administrative action more uncertain. > The comparison with France and Germany helps “normalize” the institutional intuition: in those countries, there is a **separate, specialized administrative judiciary** for disputes between individuals and the state, with its own court structure and a top administrative court at the apex (in France, the **Conseil d’État**; in Germany, a three-tier system culminating in the **Bundesverwaltungsgericht**, the Federal Administrative Court). --- ### The most important insight The report’s deepest structural argument is that implementation lag is often self-inflicted: the Biden administration treated organizational build-out as a post-election chore, when it should be part of the transition’s core work. What it calls for is a pre-built deployment capacity: - Small, specialized teams that can rapidly staff priority posts. - Pre-arranged recruiting pipelines that start producing hires immediately. - Legally available fast-track routes into government (including Schedule A–style excepted-service hiring) set up and ready to activate. In other words, the administration shouldn’t be assembling the toolkit while governing; it should arrive with it already packed (not unlike what [[Project 2025]] attempted to do for a second Trump administration). --- The question the document leaves open is whether any of these fixes are achievable within the existing architecture, or whether they require the kind of structural legislative reform that the report treats as a constraint rather than a variable. The 1970s operating system ([[Office of Management and Budget (OMB)|OMB]], [[US Office of Personnel Management (OPM)|OPM]], [[US Senior Executive Service (SES)|SES]], etc.) wasn't designed for the implementation speed that modern political cycles demand. Patching it is hard. Replacing it is harder. But at least someone is trying to write down what broke and why—which is more than most political coalitions manage to do once they are out of power.