## Notes from 11 February 2026 [[2026-02-10|← Previous note]] ┃ [[2026-02-12|Next note →]] ## Workarounds as _tacit admissions_ I've just read an excellent post by [[Martha Dacombe]] regarding the UK ("_[Why Building New Institutions Cannot Save Whitehall Alone](https://dacombe.substack.com/p/why-building-new-institutions-cannot)_"). It starts with a simple observation: institutions like [[DARPA-style setups|ARIA]], the AI Safety Institute, and the Sovereign AI Unit are **workarounds** — concessions that the British state, as currently constituted, cannot deliver what is needed. They are faster, cleaner, and more politically rewarding than traditional departments. However, every new institution built to bypass existing regulations is a tacit admission that we have given up on reforming the core. This diagnosis has a direct parallel in what Brazilians call the **"fuga do Direito Público"** ([[Escape from Public Law]]): the state creates special agencies, social organizations, and foundations — flexibility mechanisms that bypass the rigidity of the administrative core rather than reforming it. These are survival mechanisms (or a "shiny" new app) in a system whose "operating system" is too heavy to update directly. Dacombe is describing, without knowing it, a British version of the same phenomenon. ## Institutional accumulation This bypass strategy leads to a crowded landscape. [[Quasi-Autonomous Non-Governmental Organization (Quango)|Arm's Length Bodies (ALBs) and Quangos]] of all kinds in the UK [grew from 474 in 2015 to 603 in 2025](https://moderncivilservice.blog.gov.uk/2024/12/18/understanding-arms-length-bodies-a-fresh-look-at-britains-public-sector/), completely reversing the 2010 "bonfire of the quangos" promised by the conservative-liberal coalition. Dacombe uses the British Business Bank as an illustrative case: created in 2013 with a focused mission, it eventually became an "everything-institution" with a diluted mandate. The risk is that these new institutions eventually become exactly what they were designed to escape. ## The talent deficit The talent dimension is where Dacombe is sharpest. The Civil Service Fast Stream received 72,691 applications for just 754 places in the 2024-25 cycle — a 1% acceptance rate. With that volume, the system should be selecting the very best. Instead, it relies on situational judgment tests that current Fast Streamers describe as arbitrary. There is also a tension between representativeness and meritocracy. Dacombe argues that, once the investment has been made to diversify the upstream pipeline (access to Oxford, Cambridge, etc.), selection at the point of entry should be "unapologetically for capability." I believe this is a misjudgment of what diversity efforts actually entail; she implies these efforts are equivalent to "lowering the bar" rather than fostering cognitive diversity. Ultimately, I agree with her conclusion: while creating new institutions exempt from standard civil service hiring processes is helpful, it cannot blind us to the fact that the hiring system across the rest of Whitehall remains fundamentally broken. ## The "_Guerrilla Way_" out (Waissbluth) The problem with Dacombe's prescription lies in her call for a broad reform of the "old buildings," which underestimates the political economy of resistance. [[Mario Waissbluth]], in an [excellent interview](https://www.youtube.com/watch?v=xOorrmpBymE) for [[Pivotes]] in November 2025, classifies this kind of ambition as a **"_World War_ strategy"** — a macro-structural attempt to update the entire public administration at once. For Waissbluth, this is a **"guaranteed dead end"** because it triggers a unified front of resistance from unions, entrenched bureaucracies, and political actors who benefit from the status quo. The alternative is **"_Guerrilla Warfare_"**: professionalizing the state from the inside out through tactical moves: - **Strengthen the center:** Professionalize the transversal agencies that control the "plumbing" first — budgets, civil service rules, procurement, and technology. Reform the controllers before the controlled. - **Localized battles:** Instead of a sweeping national reform, select specific strategic agencies to modernize one at a time, supported by a "reform front" of transversal agencies. This prevents a unified national opposition from forming. - **Management-led professionalization:** Use a Senior Civil Service system to appoint highly competent, professional directors to targeted agencies. These leaders can then negotiate modernization directly with their staff, using managerial autonomy as a wedge against systemic inertia. The divergence here is one of strategy. Where Dacombe seeks to fix Whitehall broadly, Waissbluth proposes **concentrating** effort in key agencies to drive top-down reform within individual "battlefields." The core insight remains: the path forward is likely not a grand architectural overhaul, but a **series of tactical victories**. In this context, **[[2026-01-29|focus should be a source of speed, not constraint]]**.