Public-Interest Tech Collective (LM-Dawn class)

Class card for the LM-Dawn cluster of civic-tech collectives, government- embedded digital-services teams, and federated civic-tech cooperatives that bridge engineering-craft and democratic governance. The class form is defined by three structural properties: (1) code-as-public-good — outputs are open-source, licensed under copyleft or permissive licenses, and committed to government repositories (code.gov, GitHub public forks, OpenStreetMap contributions) rather than proprietary SaaS stacks; (2) government-as-customer not VC-as-extractor — the funding relationship is grant, government contract, or cooperative membership dues, not venture capital with exit expectations; (3) engineering-craft inside civic accountability — the practitioners hold dual identity as technologists AND civic actors, applying software engineering competence to public-interest problems (recidivism prediction, FOIA automation, benefits eligibility, government digital services) with democratic accountability mechanisms. NAMED INSTANCES [EXTRAP]: Code for America (2009-2012 founding, ~1500 fellows + brigade network; US state DOC recidivism cohort prediction systems; benefits enrollment automation; 2012 presidential summit with Obama; ~$50M annual budget 2024); 18F (US GSA digital-services group, ~150 staff, code.gov 2014+; cloud.gov; login.gov; federal digital-services contracting reform; ~$100M annual revenue from interagency agreements 2023); GDS UK (Government Digital Service, ~800 staff since 2011; GOV.UK platform; digital identity stack; Notify/Pay/Forms common components; ~£250M annual budget); Mozilla Foundation Public Interest Technology Fellowship (2019+; ~50 fellows/yr; cross between technical fellowship and civic advocacy training); Tech for Good / Code for All federation (~30 country-chapter federation: Code for Germany, Code for Pakistan, Code for Australia, Open Lab Mexico, mySociety UK, etc.; federated civic-tech cooperative operating shared-code commons per jurisdictional adaptation); New America Public Interest Technology Initiative (PIT-UN network: 40+ universities embedding public-interest tech tracks in CS programs 2019+); USDS (US Digital Service, 2014+, ~200 staff, White House embedded; health.gov, Medicare appeals automation, VA benefits digitization); DTA Australia (Digital Transformation Agency, 2015+); Estonia e-Governance Academy (international civic-tech export, 2002+); Nava PBC (public-benefit corporation doing government digital services); Skylight Digital; Ad Hoc LLC (government digital services primes operating as public-benefit entities). ADJACENCY-LIFT MECHANISM (Hidalgo §3): Three DM/MM ancestors combine to produce a configuration no single ancestor achieves. AWS cloud (2006) provides elastic compute substrate: civic-tech teams can deploy production government services without owning data centers — login.gov, cloud.gov, NHS Digital are AWS tenants (or GCP/Azure equivalents). GitHub code collaboration (2008) provides the distributed code-craft substrate: 18F's open-source-by-default policy, Code for America's brigade network, and Code for All's shared-code commons all operate on GitHub-style git workflows with public pull-request review. The US New Deal admin state (1933) provides the public-administration archetype: civic-tech collectives inherit the New Deal's "technocratic competence in service of democratic mandate" logic — the Tennessee Valley Authority and WPA as organizational templates for state-embedded technical expertise. The LM lift is the combination: cloud-substrate + code-craft + civic-mandate = a new institutional form operating shared-code commons under democratic accountability, structurally resistant to capture by holding outputs as public goods. CAPTURE RESISTANCE MECHANISM [EXTRAP]: The class's primary capture-resistance mechanism is structural: open-source licensing + government-as-customer eliminates the VC-exit incentive that drives DM platform capture. When 18F builds login.gov on open source, a private vendor cannot proprietary-lock the authentication system. When Code for America's benefits enrollment tool is open-sourced, any state can fork and deploy without vendor dependency. However, PRACTICAL capture resistance is partial (capture_resistance_index: 0.58 [EXTRAP]): (a) government procurement rules favor large defense primes who can absorb compliance overhead — civic-tech collectives compete at a structural disadvantage for large contracts; (b) the "government digital service" model is vulnerable to political-cycle capture (GDS UK was deprioritized under 2016-2019 Conservative government; 18F budget repeatedly threatened under Republican administrations); (c) the Beltway Bandit dynamic: skilled civic-tech practitioners are poached by defense primes and management consultancies at 2-3x government salaries. PROLETARIANIZATION RISK is MEDIUM-HIGH (0.62 [EXTRAP]): The civic-tech collective class requires rare competence: engineers who hold both technical excellence AND civic-accountability norms, AND government-procurement navigation skills. This triple-competence profile is not reproduced by standard CS education. The institutional memory of "how to build good government digital services" is thin: GDS UK lost 40%+ of its founding team between 2014-2020; 18F experienced ~30% annual turnover 2015-2022. If the fellowship/brigade pipeline thins, the class persists as a category (code.gov exists, PIT-UN exists) while effective civic-tech delivery capability degrades — the canonical Stiegler proletarianization scenario. emergence_subtype [v0.2-gap — field missing, recorded here]: civic_craft_cooperative. The class instantiates a cooperative craft-guild form applied to government digital services — not emergent from market dynamics but from deliberate institutional design by practitioners who refused the VC-extractor path. Substrate gap [STUB-substrate-enum-gap]: Substrate.institutional enum value missing from v0.1 schema. The class requires an institutional substrate (government charters, cooperative articles of incorporation, fellowship program governance) that is neither purely social nor purely semiotic. Workaround: [social, semiotic, cognitive] per embedded constraints; noting gap here. All quantitative values are [EXTRAP] unless otherwise flagged. Framing sources: Code for America founding (2009-2012) [CANON]; GDS UK GOV.UK launch (2012) [CANON]; 18F founding (2014) [CANON]; USDS founding (2014) [CANON]; Wave-6 adjacency-lift logic [EXTRAP]; Rao world-machines era framing [EXTRAP].