{"path":"research/governance-and-financing.md","content":"# Governance and Financing for Deliberus\n\n*Research document — March 28, 2026*\n*Last material update: April 16, 2026 (SFF / LTFF / fiscal-sponsorship addendum)*\n*Status: Intake/research phase. Nothing decided.*\n\n> **2026 addendum (April 16):** This doc predates awareness of the Survival and Flourishing Fund's 2026 application window and the EA-Sweden fiscal-sponsorship pattern. Those materials now live in [sff-2026-and-fiscal-sponsorship-path.md](sff-2026-and-fiscal-sponsorship-path.md), which should be read alongside this document and which corrects this doc's recommended primary financing sequence. The short version: SFF Main Track has an April 22, 2026 deadline, an HSEE round follows on July 8, and the practical unlock is fiscal sponsorship via EA Sweden — not Swedish incorporation. See the companion doc for full reasoning and the public precedent study.\n\n---\n\n## 1. How Similar Projects Are Structured\n\nThe epistemic infrastructure space has developed several distinct organizational archetypes. Understanding them in detail is the prerequisite for choosing a path.\n\n### Polis / Computational Democracy Project\n\n**Structure**: 501(c)(3) nonprofit, USA (The Computational Democracy Project)\n**Org story**: Polis started as a startup experiment. Colin Megill and co-founders open-sourced the technology at the 2016 g0v summit in Taiwan. Liz Barry then helped establish the Computational Democracy Project to steward the open-source code as a nonprofit, completed roughly 2018–2020. The transition from commercial to nonprofit was a deliberate strategic choice to serve public interest without VC pressure.\n**Funding**: Grant-dependent. In 2021, Colin Megill was advising OpenAI on deliberation at scale and helped award $1 million in grants to teams working on deliberation problems. No substantial revenue-generating product — the org survives on grants, consulting services for governments and journalists, and occasional foundation support.\n**Key lesson**: Polis deliberately rejected commercialization. It remained small and mission-driven but fragile — dependent on external validation through high-profile deployments (Taiwan, EU) to attract the next grant. It never solved self-sustaining revenue.\n\n### Wikipedia / Wikimedia Foundation\n\n**Structure**: 501(c)(3) nonprofit, USA\n**Org story**: Founded 2001, incorporated as nonprofit 2003. For the first several years it survived on tiny fundraisers — literally passing a hat. The banner donation drives started around 2008–2010 and proved transformative. Revenue grew 5x over the subsequent decade.\n**Current scale**: 8 million donors in FY2023-2024, average donation $10.58. Wikimedia Endowment reached $144 million in FY2023-2024. Total annual revenue ~$180 million. Wikimedia Enterprise (paid API tier for search engines) generated $3.4 million in FY2023-2024.\n**Path to sustainability**: (1) Build something genuinely irreplaceable, (2) let millions of free-riders accumulate, (3) run emotionally compelling \"help us survive\" donation drives, (4) diversify into enterprise revenue once large enough. This took ~8 years of near-subsistence operation before it was safe.\n**Key lesson**: Wikipedia's trajectory required reaching global cultural ubiquity before donations became reliable. The lesson for Deliberus is not to replicate this path exactly — Wikipedia had the advantage of encyclopedic content that aggregated instantly — but to understand that nonprofit donation sustainability requires genuine mass adoption first.\n\n### Metaculus\n\n**Structure**: Public Benefit Corporation (PBC), USA (restructured 2022)\n**Funding**: Grant-heavy. EA Infrastructure Fund: $300K (2021). Open Philanthropy: $5.5 million (2022). NSF grant: $175K (2024). No advertising. Some tournament prize money.\n**Business model**: Free platform + grants + potential future B2B institutional forecasting services. The PBC structure gives flexibility to pursue both mission and revenue without the restrictions of a pure nonprofit.\n**Key lesson**: Metaculus raised over $6 million in grants by demonstrating credibility in the EA/effective-altruism and institutional decision-making communities. Grant-dependency still poses risks — a single large funder (Open Philanthropy) represents most of their operating budget. The PBC structure was chosen precisely to allow future revenue diversification without compromising mission.\n\n### Signal Foundation\n\n**Structure**: 501(c)(3) nonprofit, USA\n**Org story**: Brian Acton (WhatsApp co-founder) gave a $50 million loan in 2017 to bootstrap Signal development, later converted to a foundation donation. This is a critical data point: Signal was essentially philanthropist-founder-funded for its first decade.\n**Current financials**: Revenue grew from $8.1M (2021) to $11.1M (2023) to $25.8M (2024). But expenses outpace revenue — in 2023 Signal declared expenses exceeded funds at $35.8M. By 2025 they need roughly $50M/year to sustain operations at current scale. 177 employees as of 2024.\n**Key lesson**: Signal is a cautionary example of scale trap — open-source infrastructure that's free to end users hits a wall where success (user growth) increases costs but not revenue. The Acton loan/donation bought them 8+ years, but long-term sustainability remains unresolved. For Deliberus: infrastructure projects of this type require either a philanthropist patron, a revenue model, or a much smaller team.\n\n### Mozilla Foundation + Mozilla Corporation\n\n**Structure**: Nonprofit foundation (Mozilla Foundation, 501(c)(3)) wholly owning a for-profit subsidiary (Mozilla Corporation). The subsidiary can generate unrestricted revenue while the foundation sets mission.\n**Revenue**: ~86% of Mozilla Corporation revenue comes from Google paying to be the default search engine in Firefox. This is a single-point-of-failure revenue model that Mozilla has been trying to diversify for years. Total revenue was several hundred million USD annually at peak, but declining in 2024 due to search deal changes.\n**Key lesson**: The Foundation + Subsidiary model is elegant but requires finding a revenue source. Mozilla found theirs accidentally (Google paying for default status). The structural lesson is useful: foundation for long-term ownership and mission protection, operating entity for commercial activities. This is a legitimate model for Deliberus Phase 2+, but requires knowing what the revenue-generating service is before building the structure around it.\n\n### Bluesky\n\n**Structure**: Public Benefit Corporation (PBLLC → PBC), USA\n**Funding**: $8M seed (2023), $15M Series A (Oct 2024), $97M Series B (Jan 2025). Total raised: ~$120M.\n**Revenue model**: No ads as of late 2024. Jay Graber committed to avoiding \"enshittification.\" Future: optional subscriptions, user-to-user payments.\n**Key lesson**: Bluesky is VC-funded infrastructure with no proven revenue model — a bet that scale achieved through PBC mission framing will eventually unlock sustainable revenue. This works if you can raise VC, which requires: (a) a potential billion-dollar outcome, (b) connections to the right firms, (c) accepting that VCs will eventually push for returns. The PBC structure is a legal commitment to not maximize shareholder value, but it does not eliminate investor pressure. For Deliberus: VC is the wrong path unless the vision is commercialized at scale in ways that conflict with the societal mission.\n\n### Mastodon\n\n**Structure**: 501(c)(3) nonprofit (USA, after losing German gGmbH nonprofit status). Moving toward a community foundation with governance in 2025.\n**Funding**: Patreon + OpenCollective crowdfunding, Samsung grant, European Commission grant.\n**Patreon numbers**: ~$7,000/month in early 2022 → $31,000/month during the Twitter exodus surge (Dec 2022) → ~$33,465/month peak with 9,411 patrons (Jan 2023) → ~$25,000-30,000/month stabilized with ~8,500 backers (Feb 2024).\n**Key lesson**: One developer (Eugen Rochko) sustained a major open-source social infrastructure project on ~$25-33K/month in Patreon income — enough for a small team in Germany. The surge around Twitter events shows the vulnerability of donation-based models: income spikes with news cycles and then drops. Mastodon's success is partly because Rochko was willing to work at below-market rates for years for the mission. The European Commission grant and Samsung grant demonstrate that legitimately-structured nonprofits can attract institutional donors without being grant-proposal factories.\n\n### Internet Archive\n\n**Structure**: 501(c)(3) nonprofit, USA\n**Funding**: Mix of service revenue (Archive-It subscription: $37M+ annual revenue by 2019), Kahle-Austin Foundation ($13.5M over 20 years), Mellon Foundation, federal grants (NEH), and public donations. Brewster Kahle's personal wealth and mission commitment were critical in the early years.\n**Key lesson**: The Internet Archive demonstrates that a mission-driven nonprofit can achieve financial stability through service diversification — their paid web archiving service for institutions (Archive-It) provides predictable revenue alongside donations. This is the model that actually works: identify who benefits institutionally and price services for them while keeping the public offering free.\n\n### Kialo\n\n**Structure**: Nonprofit, funded by a private philanthropic foundation\n**Funding**: Errikos Pitsos (founder/CEO) personally funded early operations. Now described as \"entirely funded by a private philanthropic foundation.\" No advertising, no user fees.\n**Scale**: Roughly 50 full-time employees. Has not achieved mainstream adoption despite being the closest existing product to structured argumentation.\n**Key lesson**: Private philanthropic foundation funding is the most extreme version of patron dependency. It buys total mission purity — no revenue pressure, no VC misalignment — but creates existential risk if the patron's priorities shift. Kialo has avoided the commercialization trap at the cost of existential dependency. Not a replicable model without a known patron.\n\n### Hypothes.is (→ Anno)\n\n**Structure**: Started as nonprofit, then created PBC subsidiary \"Annotation Unlimited, PBC\" (Anno) in August 2022\n**Funding path**: Mellon Foundation (multi-year grant), Omidyar Network ($500K), Sloan Foundation ($394K to investigate sustainable models). After grant funding became unreliable: formed Anno with $14M seed round (ITHAKA, At.inc, Triage Ventures, Esther Dyson, Mark Pincus).\n**Key lesson**: Hypothes.is explicitly documented that being a nonprofit \"limited the company to grants and donations and with several key funding sources no longer available\" — then formed a PBC. This is the canonical case study for why pure-nonprofit infrastructure projects face sustainability crises. The hybrid structure (nonprofit advocacy + PBC operating company) preserved mission while unlocking investment. Their path matters for Deliberus: don't assume nonprofit is permanent — design the structure to evolve.\n\n---\n\n## 2. Organizational Models Compared\n\n| Model | Pros | Cons | Best For |\n|-------|------|------|----------|\n| **Solo developer** | Zero overhead, complete freedom, no stakeholder management, fast iteration | Income ceiling, burnout risk, no redundancy, hard to attract contributors | MVP phase, proof-of-concept, research projects where one visionary carries the work |\n| **Small AI-native team (2–4)** | 8–15x productivity multiplier over traditional teams, diversity of perspectives, shared context, can tackle more simultaneously | Equity/IP coordination, trust requirements, communication overhead, legal setup needed | Projects past MVP with a roadmap requiring multiple parallel workstreams |\n| **VC-backed startup** | Fast capital, network access, talent pipeline, credibility | Mission alignment risk, exit pressure, VC interest ≠ societal good, must demonstrate path to billions | Commercial products where scaling requires capital and speed is a competitive moat |\n| **Foundation/nonprofit (501(c)(3) or Swedish ideell förening/stiftelse)** | Donation eligibility, grant access, mission protection, tax advantages, community trust | Revenue restrictions, bureaucratic overhead, slower decision-making, cannot give founders equity | Long-term infrastructure that needs public trust and institutional grant access |\n| **PBC (Public Benefit Corporation)** | Combines investment eligibility with mission commitment, flexible revenue options, can accept donations AND investment | Mission commitment is a reputational constraint, not a legal lock — investors can still push hard | Infrastructure projects that need both grant-funding legitimacy and future investment optionality |\n| **Open-source community** | Distributed maintenance, contributor network, trust from developers | Coordination costs, uneven contribution quality, hard to sustain without a core paid team | Code libraries and protocols, not platforms requiring product discipline |\n| **University/research lab** | Grant eligibility, credibility, infrastructure, collaborator access | Slow, publication-pressure, IP complications, cannot ship product quickly | Foundational research, datasets, standards work — complements but doesn't replace a product org |\n\n---\n\n## 3. Financing Options for Pre-Revenue Epistemic Infrastructure\n\n### Patreon / Ko-fi / Open Collective\n\nRealistic numbers based on comparable projects:\n- **Early stage** (first 6–12 months, small committed community): $500–2,000/month\n- **Established open-source project with media coverage**: $3,000–10,000/month\n- **Mastodon-level (major infrastructure, periodic news surges)**: $25,000–33,000/month peak\n- **Highly popular creator educator adjacent**: $50,000–100,000/month (not relevant here)\n\nThe honest picture: Patreon income for civic/epistemic tech is highly variable and surge-dependent. It rarely reaches $10,000/month without either a large existing audience or a catalyzing event (Twitter meltdown equivalent). It works as *supplemental* income, not a primary operating budget for a platform. Best used in combination with grants during Phase 1.\n\n### GitHub Sponsors\n\nFor open-source projects: easier to convert developers who use the code. Average GitHub sponsor contribution is ~$5–15/month. Works better for developer tools than for platforms. Likely produces $200–2,000/month for a well-maintained open-source argumentation library.\n\n### Grants — International\n\n**Open Philanthropy** ($5.5M to Metaculus; funds AI safety, biosecurity, global health): Strong interest in epistemics and decision-making infrastructure. Requires demonstrated community/credibility. Grants range from $50K to $10M+. Very selective; track record matters. Contact: openphilanthropy.org/grants\n\n**Knight Foundation** (Information & Society program): Focuses on journalism, civic tech, informed communities. Individual grants typically $50K–$500K. Has funded civic tech through specific programs — not an open call but through invited proposals and partner relationships. Recent $4M+ investments in civic tech data. Contact: knightfoundation.org/apply\n\n**Mozilla Foundation**: Funds open internet infrastructure and privacy. Mozilla Technology Fund: individual awards of $50K–$200K. Mission alignment is excellent (open, epistemic, non-commercial). Applications evaluated quarterly.\n\n**Omidyar Network**: Has funded Hypothes.is ($500K) and epistemic/technology/accountability work. Grants in the $100K–$2M range for organizations with demonstrated traction. Also invested $6.5M into reimagining capitalism academic work alongside Hewlett.\n\n**Hewlett Foundation**: Funds open education, technology, and democracy. \"Cyber Initiative\" and \"Democracy\" programs most relevant. Individual grants $100K–$5M, multi-year relationships. Very selective, favors established organizations with track record.\n\n**Open Society Foundations**: Funds democracy, transparency, and human rights technology. Relevant programs: Information Program (media, disinformation) and Scholarship. Grants typically $50K–$500K for early-stage work.\n\n**Sloan Foundation**: Explicitly funded Hypothes.is to \"investigate and establish sustainable business models\" for annotation infrastructure. $394K grant. Precedent directly relevant to Deliberus. Focus: digital infrastructure, open science, technology.\n\n**Andrew W. Mellon Foundation**: Humanities and cultural institutions. Funded Hypothes.is for publishing infrastructure. Relevant if Deliberus is framed around academic discourse and digital humanities. Grants typically $100K–$1M.\n\n**NSF (National Science Foundation, USA)**: Funded Metaculus ($175K). Programs: CISE (computing), SBE (social/behavioral/economic sciences). Relevant calls include human-centered computing, collective intelligence, AI for social good. Amounts: $50K–$2M depending on program. Requires US institutional affiliation unless applying as industry.\n\n**DARPA / IARPA**: Long-shot but worth monitoring. IARPA has funded forecasting/prediction infrastructure (IARPA ACE program → spawned Metaculus and Good Judgment Project). Very competitive, usually requires US-based org.\n\n### Grants — Sweden and EU\n\n**Vinnova** (Sweden's Innovation Agency):\n- **Innovativa Startups 2025**: Up to SEK 500,000 (~€45,000). Must be Swedish AB, not listed, turnover <SEK 10M. Project max 9 months. Good for early MVP funding.\n- **Verification for growth** and other programs: SEK 1–10M depending on stage. Requires business potential framing.\n- **Key constraint**: Vinnova funds *organizations*, not individuals. You need at minimum an enskild firma or aktiebolag before applying. Processing takes 2–6 months.\n\n**ALMI** (Sweden's business development bank):\n- Innovation Loan: SEK 50,000–500,000, typical ~SEK 300,000 (~€27,000). Interest ~10–12%. For early-stage innovative ideas.\n- Business Loan: SEK 250,000 to several million for established companies.\n- ALMI Invest (venture capital): SEK 2–10M initial investment. Requires equity stake. Less relevant for mission-driven infrastructure.\n- **Key strength**: ALMI loans don't require existing revenue. They exist specifically for early-stage innovative companies. Lower bar than Vinnova grants.\n\n**EU Horizon Europe**:\n- Horizon Europe 2021–2027: €95.5 billion total. AI and digital programs: €14 billion in recent work programme.\n- **AI4Deliberation project** (ARG-tech led, Dundee): This is an active EU-funded consortium around AI and deliberation tools. Contact Chris Reed's group at the University of Dundee — the path to EU funding may be through partnership with this consortium rather than independent application. EU projects typically require 3+ partners from different member states.\n- **Digital Europe Programme**: Separate from Horizon. Funds deployment of digital technologies. More accessible for smaller organizations in consortium projects.\n- **ERC (European Research Council)** Starting Grants: €1.5M for researchers within 7 years of PhD. Requires academic affiliation.\n- **Realistic path**: Apply to Horizon calls as a project partner with ARG-tech (Dundee) or a Swedish university. Horizon grants for collaborative research projects typically run €2–8M total split across partners.\n\n### Website-Based Funding (\"Support Deliberus\")\n\nWikipedia's banner approach works because of massive reach. For smaller projects the equivalent is:\n- A \"Founding Members\" tier: €5–50/month for early believers who get recognition, early access, input on direction\n- One-time \"invest in epistemic infrastructure\" donation page with progress indicators\n- Realistic for Phase 1: may generate €500–5,000/month if community is engaged. Better than nothing, not a business model.\n\n### Consulting / Services (Epistemic Facilitation)\n\nUsing the Deliberus methodology as a paid facilitation service while building the platform:\n- Governments, think tanks, NGOs, companies pay for structured deliberation facilitation\n- Rates: €3,000–15,000 per facilitated deliberation engagement\n- Could realistically generate €50,000–200,000/year at full capacity for one person\n- **Critical advantage**: Funds development *and* generates real-world use cases to validate the platform\n- **Critical risk**: Consulting becomes the business, platform development stalls\n- This model worked for early Polis — they did facilitation contracts for governments while developing the platform\n\n### Freemium\n\nFree for public deliberations (communities, civic topics), paid for private/organizational use:\n- Comparable: Loomio charges organizations €149–499/month for private decision-making workspaces\n- Kialo Edu is free (donor-funded); Kialo main platform free\n- Deliberus private organizational tier: €200–1,000/month per organization is realistic if the product is genuinely useful\n- **When to build this**: Phase 2, once the public platform has demonstrated value. Building a freemium billing system before product-market fit is waste.\n\n### Academic Partnerships\n\nCo-development with established argumentation research groups:\n- **ARG-tech, University of Dundee** (Chris Reed): The largest academic argumentation infrastructure group in the world. They built the Argument Web, maintain the largest public argumentation corpus, and lead EU-funded projects. A partnership with ARG-tech could unlock: Horizon Europe funding, researcher contributions, corpus access, academic credibility.\n- **MIT Center for Collective Intelligence**: Has studied deliberation and group decision-making extensively. Connection potential through past research affiliations.\n- **Harvard's Berkman Klein Center**: Internet and society research, open internet advocacy. Grant writing and network introductions.\n- **Benefit**: Academic partnerships can provide 1–3 funded researchers without payroll cost. Risk: academic timelines are slow, publication pressure may conflict with product shipping.\n\n---\n\n## 4. The AI-Native Small Team Model\n\n### What \"AI-Native\" Means in 2026\n\nThe 2025 data from Stripe's Indie Founder Report showed 44% of profitable SaaS products run by a single founder — doubled since 2018. At Y Combinator Winter 2025, 25% of startups reported codebases that were 95%+ AI-generated.\n\nThe documented productivity multiplier for experienced developers with AI tooling (Claude Code, Cursor Agent, Windsurf) is 8–15x against industry baseline (500–1,000 tested LOC/day for senior developers). At peak velocity with solid specs: 15–25x for well-scoped features (see the 222x case in the global CLAUDE.md for the spec-first paradigm — spec is the work, implementation is transcription).\n\nA 2–4 person AI-native team in 2026 can:\n- Maintain a codebase that would previously require 15–25 engineers\n- Ship features in hours that would take traditional teams weeks\n- Keep quality high through AI-assisted review and test generation\n- Move at startup velocity without the engineering overhead\n\n**Notable examples**:\n- **Base44**: Built alone by Maor Shlomo, acquired by Wix for $80M in June 2025, six months post-launch, 250,000 users\n- **BoltAI**: Solo developer (Daniel Nguyen), desktop AI tool, $15K+ MRR within months of launch\n- General pattern: AI-native solo and duo developers routinely reaching $100K–$1M+ ARR in 2024–2025 for focused tools\n\n### Finding and Evaluating Potential Co-Developers\n\nFor a project like Deliberus, the ideal second person is not a generalist developer but someone who brings a specific combination:\n- Deep interest in epistemics, argumentation, or collective intelligence (not just a hired engineer)\n- Comfort with AI-assisted development velocity\n- Alignment on the non-commercial mission\n\nWhere to find them:\n- **LessWrong / EA Forum** communities: Heavy overlap with argumentation, epistemics, prediction markets\n- **ACL / ArgMining workshops**: Researchers at the intersection of NLP and argumentation\n- **Kialo / Pol.is communities**: People who already live in this space\n- **Academic argumentation conferences**: COMMA (Computational Models of Argument) biennial conference\n- **The existing FB group** (452 members, many now 30–45 years old, originally found by a 23-year-old with LessWrong-adjacent interests): Many of those people have spent 15 years caring about this problem\n\n**Evaluation criteria**:\n1. Can they articulate what's wrong with existing platforms without prompting?\n2. Do they have a personal project or writing about collective intelligence or epistemics?\n3. Do they understand why adoption is the hardest problem (not the technology)?\n4. Are they comfortable with the non-commercial mission framing?\n\n### Trust Frameworks for Intense Collaboration\n\nFor remote-first small teams working at high velocity:\n- **Weekly async video recap** (Loom, 10 minutes): Prevents communication drift without synchronous meeting overhead\n- **Shared decision log**: A simple file or Notion page where significant design/architecture decisions are recorded with rationale — prevents the \"why did we do it this way?\" conversation from repeating\n- **Explicit work allocation**: Who owns what, clearly. Shared ownership leads to duplicate work or dropped balls at velocity\n- **Regular architecture reviews**: Monthly, 60–90 minutes. Review what was built, identify drift from vision, adjust course\n- **Documented disagreement process**: When you disagree on direction, write both positions down, decide, and record the decision. At velocity, unresolved disagreements compound fast\n\n### Legal Structures for Small Teams\n\n**For a duo or trio working on Deliberus**:\n\n*Partnership agreement (before formal incorporation)*: Before any code or IP is created jointly, agree in writing on: IP ownership (likely jointly held or assigned to a forthcoming entity), equity split, decision-making process, what happens if someone leaves. This does not require a lawyer but benefits from one. Cost: €500–2,000 for a simple agreement.\n\n*Swedish aktiebolag (AB)*: Standard choice for Swedish companies planning revenue. Requires SEK 25,000 in share capital. Can issue shares with different classes. Taxed at 20.6% corporate rate. Dividend taxation applies when profits distributed. Can apply for Vinnova and ALMI. **Downside**: harder to be grant-eligible as a for-profit if applying to foundations that prefer nonprofits.\n\n*Ideell förening (nonprofit association)*: Swedish nonprofit association. No capital requirement. Democratic governance (member meetings, board). Tax-exempt on activities that serve the public purpose. Can receive donations and apply for cultural/civil society grants. **Downside**: cannot have equity, harder to attract investment if needed later. Best for Phase 0–1.\n\n*Stiftelse (foundation)*: Capital-based, defined by founder's will. Very stable (hard to dissolve), but also inflexible — governance defined at creation. Better for long-term endowment than for an active development project.\n\n*European Cooperative Society (SCE)*: EU-wide structure for cooperative enterprises. Relevant if building a member-governed deliberation platform where users/contributors are also owners.\n\n**Recommended path**:\n- Phase 0: No formal entity. Personal work, no joint IP yet.\n- Phase 1: Ideell förening if co-developing with mission-aligned collaborators. Enables grant applications, keeps non-commercial framing.\n- Phase 2: Consider aktiebolag or PBC equivalent (Sweden doesn't have PBC, but a AB with an explicit societal purpose clause in the articles is possible) if revenue models are being pursued.\n\n---\n\n## 5. Swedish-Specific Considerations\n\n### Tax Structures Compared\n\n| Structure | Capital Req | Tax Rate | Grant Eligible | Investment Eligible | Notes |\n|-----------|-------------|----------|----------------|---------------------|-------|\n| Enskild firma | None | Income tax (personal) | Limited | No | Simplest, for solo freelancing |\n| Aktiebolag (AB) | SEK 25,000 | 20.6% corporate + dividend tax | Yes (Vinnova, ALMI) | Yes | Standard startup choice |\n| Ideell förening | None | Exempt (if public purpose) | Yes (foundations, EU) | No | For nonprofit/mission work |\n| Stiftelse | Varies | Exempt (if charitable) | Yes (some) | No | Long-term endowment, inflexible |\n\n**2026 change**: From 2026, aktiebolag can receive tax reduction when donating to approved nonprofit activities (20.6% reduction on donations up to SEK 800,000/year). This creates a mechanism for corporate sponsors to support Deliberus tax-efficiently once it has ideell förening status.\n\n### Vinnova — Specific Relevant Programs (2025–2026)\n\n1. **Innovativa Startups**: SEK 500,000, 9-month project, requires Swedish AB. Apply as soon as AB is formed. Suitable for MVP development phase.\n2. **Utmaningsdriven innovation (Challenge-Driven Innovation)**: Multi-year collaborative projects, SEK 2–15M. Requires consortium. Framing Deliberus as a solution to democratic legitimacy or epistemic infrastructure challenge could qualify.\n3. **Digital infrastructure and trustworthy AI**: Vinnova funds AI safety and trustworthy AI development. Deliberus's transparent argumentation graph framing aligns.\n\n### EU Funding — Horizon Europe Path\n\nMost realistic EU path:\n1. Contact ARG-tech at University of Dundee (Chris Reed's group) directly — they are leading EU argumentation/deliberation projects\n2. Express interest in joining as an industry/startup partner in their next Horizon call\n3. A Swedish AB or ideell förening can be a project partner in Horizon consortia\n4. Swedish organizations can apply to Horizon Europe calls as Sweden is an associated country\n\n**AI for Democracy calls**: EU has been investing in AI tools for democratic participation. Horizon Europe's Cluster 4 (Digital, Industry, Space) and Cluster 2 (Culture, Creativity, Inclusive Society) both have relevant calls.\n\n### ALMI — Recommended First Step\n\nALMI's Innovation Loan (SEK 50,000–500,000) is the most accessible public funding source in Sweden for early-stage innovative ideas. It requires:\n- Swedish company registration (AB or certain other forms)\n- An innovative idea with growth potential\n- A basic business plan\n- No existing revenue required\n\nAt ~10% interest on SEK 300,000 (~€27,000), the cost is ~€2,700/year — acceptable for early development capital while pursuing grants. ALMI also provides business development coaching included in the loan package.\n\n---\n\n## 6. Staging Strategy\n\n### Phase 0: Solo Developer, Self-Funded (Now → End of 2026)\n\n**Goal**: Design, prototype, validate core thesis. No code commits, then careful experimental code.\n**Funding**: Self-funded via side project income.\n**Legal structure**: None required. Begin as enskild firma if invoicing is needed.\n**Key activities**:\n- Complete the ontology and architecture design work (weeks to months of deep thinking)\n- Build a minimal functional prototype demonstrating the core graph structure\n- Engage 5–15 researchers/practitioners in the deliberation space for feedback\n- Document everything in a form that can attract collaborators\n\n**Decision gate**: Does the prototype demonstrate something that doesn't exist and that people immediately understand the value of?\n\n### Phase 1: Open Source Release + Small Community + First Grants (2026–2027)\n\n**Goal**: Establish credibility, attract first collaborators, fund continued development.\n**Funding**: Vinnova Innovativa Startups (~SEK 500K), ALMI Innovation Loan (~SEK 300K), Mozilla Foundation grant (~$50–100K), Patreon/Open Collective for community members (~€500–3,000/month)\n**Legal structure**: Form ideell förening or aktiebolag (depends on grant requirements — Vinnova requires AB)\n**Key activities**:\n- Open-source the core graph data model and basic interface\n- Publish to LessWrong, EA Forum, ACL mailing lists, ArgMining community\n- Engage ARG-tech (Dundee) about collaboration/partnership\n- Begin grant applications to Knight Foundation, Mozilla Foundation, Sloan Foundation\n- Consider consulting services (deliberation facilitation) as supplemental income\n\n**Decision gate**: Is there an active contributor community? Do institutions express interest in deploying Deliberus for real deliberations?\n\n### Phase 2: Grant Funding for Research Goals (2027–2028)\n\n**Goal**: Fund specific research milestones with institutional grants while maintaining mission integrity.\n**Funding**: Open Philanthropy ($300K–5M), Knight Foundation ($100K–500K), Hewlett Foundation, EU Horizon consortium project ($500K–2M as consortium partner), Omidyar Network ($200K–1M)\n**Legal structure**: Ideell förening + optional aktiebolag subsidiary for earned revenue (analogous to Mozilla model but Swedish)\n**Key activities**:\n- Publish results demonstrating the argument graph model works at scale\n- Establish 2–4 person core team (founders + grant-funded researchers + community contributors)\n- Deploy in a high-visibility context (government deliberation, academic peer review, media fact-checking)\n- Identify the revenue-generating service layer (institutional subscriptions, facilitation, API access)\n\n**Decision gate**: Is there evidence that structured argumentation changes outcomes? Can the platform demonstrate epistemic lift — measurably better reasoning — compared to unstructured discourse?\n\n### Phase 3: Foundation or PBC for Long-Term Sustainability (2028+)\n\n**Goal**: Achieve organizational form appropriate for infrastructure that outlives its founders.\n**Funding**: Mix of earned revenue (institutional subscriptions, facilitation services), endowment, ongoing grants\n**Legal structure**: Either (a) Swedish stiftelse with aktiebolag operating subsidiary, or (b) International nonprofit equivalent, or (c) Community-governed cooperative structure\n**Key activities**:\n- Establish endowment (target: 3–5 years operating budget = €1–5M)\n- Formalize community governance (give users/contributors real voice)\n- Seek Wikipedia-model API revenue from platforms that use the argumentation graph\n- Explore Wikimedia Enterprise analogy — premium access for large organizations\n\n---\n\n## 7. SFF / LTFF / Fiscal Sponsorship — April 2026 Addendum\n\nThis section was added in April 2026 to correct an important gap in the original research, which did not mention the AI-safety / x-risk grant ecosystem despite that ecosystem being the tightest thematic fit for Deliberus's pitch. Full reasoning lives in [sff-2026-and-fiscal-sponsorship-path.md](sff-2026-and-fiscal-sponsorship-path.md); the short version is here so the recommended sequence below can be read without cross-referencing.\n\n**Survival and Flourishing Fund (SFF)** — Jaan Tallinn's grant vehicle, $20–40M per year across six tracks, operated via the \"S-Process\" panel-of-recommenders model. The April 2026 working model treated Main + HSEE as stackable and speculation grants as a near-automatic procedural gate; the July 2026 factual refresh supersedes that. The same project may not apply to multiple 2026 SFF rounds, HSEE recommendations are announced at the end of November 2026 with disbursement by the end of March 2027, and speculation grants should be treated as the live eligibility / expedited-funding channel rather than as a guaranteed approval statistic. HSEE remains the live 2026 Deliberus target because \"augmenting human judgment without replacing it\" is almost a direct paraphrase of the Deliberus mission. Default open-source / CC-BY / MIT+Apache-2 IP obligation — Deliberus already complies.\n\n**Long-Term Future Fund (LTFF)** — rolling-deadline grant vehicle operated by Effective Ventures Foundation for early-stage long-term x-risk work. Grant sizes typically $20K–$100K; decisions in 6–10 weeks. Historically the path-finder grant for x-risk projects that later scale into SFF and Open Philanthropy.\n\n**The fiscal sponsorship unlock.** SFF (and many other x-risk grants) do not require applicants to have their own charity status. They require that the applicant either be a registered non-profit *or* have a fiscal sponsor — an existing non-profit that receives and administers grant funds on behalf of an un-incorporated project in exchange for a small administrative fee (commonly 5–10%). This collapses the \"we need to incorporate before applying\" problem into \"we need a sponsor agreement.\" The recommended-path ladder further down in this doc (enskild firma → ideell förening → aktiebolag) is the correct self-financed bootstrapping sequence, but it is *not* a prerequisite for SFF eligibility. Fiscal sponsorship is.\n\n**EA Sweden as the likely sponsor.** The Swedish effective altruism association is the closest publicly-documented fiscal-sponsor institution for Swedish x-risk work. The most concrete public precedent is Ulrik Horn's Minimum Viable Shelters for Mirror Bacteria project, which per [its own LessWrong write-up](https://www.lesswrong.com/posts/2t2QuD3bPyWZ2aDBr/last-line-of-defense-minimum-viable-shelters-for-mirror) is fiscally sponsored by EA Sweden, received initial LTFF funding, and is currently SFF- and Jaan-Tallinn-funded. Topical overlap with Deliberus is zero (biosecurity vs epistemic infrastructure) but the *procedural* overlap is total: Swedish base, solo-led at origin, grew into LTFF → SFF + Tallinn direct funding through EA Sweden sponsorship. This is documented as a precedent and template to learn from, not as a personal relationship claim.\n\n**Risks inherited from the x-risk funding pattern.** Single-donor concentration (SFF is Tallinn). Grant-cycle latency (SFF Main Track recommendations arrive September, funding before end of 2026 — a 5–8 month gap). The speculation-grant stage as a narrower filter than the aggregate statistics suggest. Mission drift from grant framing — the x-risk framings SFF recommenders respond to do not cover the full Deliberus thesis, especially the attunement pole and the cooperation-synthesis framings, and the application should not flatten those in the name of grant-readability. All four are covered in detail in [sff-2026-and-fiscal-sponsorship-path.md §7](sff-2026-and-fiscal-sponsorship-path.md).\n\n---\n\n## Summary: Recommended Path for Deliberus\n\nGiven the explicit mission (societal truth-seeking infrastructure, not SaaS), the self-funding horizon (through end of 2026), and the Swedish context:\n\n**Organizational form**: For SFF / LTFF eligibility, the practical unlock is **fiscal sponsorship via EA Sweden** rather than immediate incorporation. Swedish incorporation (ideell förening, and later aktiebolag for Vinnova/ALMI) remains the right long-term structure, but fiscal sponsorship collapses the \"need an entity before applying\" blocker for x-risk grants and defers the operating-entity decision to when it is actually knowable. The two-entity structure (nonprofit association + operating company) is the right destination for Phase 2+.\n\n**Avoid**: VC funding entirely (mission misalignment guaranteed at exit), Kialo-style patron dependency (single point of failure), pure-nonprofit without revenue plan (Signal trap).\n\n**Primary financing sequence (updated April 2026)**:\n1. Self-funded MVP (now — already shipping)\n2. **EA Sweden fiscal sponsorship conversation** — the structural unlock for everything below; see [sff-2026-and-fiscal-sponsorship-path.md](sff-2026-and-fiscal-sponsorship-path.md) for full reasoning\n3. **SFF 2026 HSEE / speculation channel** — live 2026 path after the July factual refresh; Main+HSEE stacking is not available for the same project\n4. **LTFF** (rolling deadline) in parallel — faster-turnaround grant as path-finder track record and to cover the SFF evaluation gap (submission April → funding December)\n5. ALMI Innovation Loan (SEK 300K) + Vinnova Innovativa Startups (SEK 500K) = ~€75K Swedish bootstrap, once AB is formed (a Phase-2 move, not a prerequisite for x-risk grants)\n6. Mozilla Foundation + Sloan Foundation grants for open-infrastructure framing ($100–400K)\n7. Open Philanthropy or Hewlett once community traction demonstrated ($1–5M)\n8. Horizon Europe consortium partnership with ARG-tech for research funding\n\n**Revenue model**: Institutional subscriptions for private deliberations + facilitation services. Free public tier forever. This is how to make the mission self-sustaining without depending on perpetual donations.\n\n**Team**: Stay solo through Phase 0. Recruit one mission-aligned collaborator with deep epistemics background in Phase 1 — not a hired developer, but a partner who already cares about this problem and brings a complementary lens (philosophy of argumentation, behavioral science, or HCI).\n\n---\n\n*Sources used in this research:*\n- [Pol.is Wikipedia](https://en.wikipedia.org/wiki/Pol.is)\n- [Computational Democracy Project](https://compdemocracy.org/polis/)\n- [Metaculus Wikipedia](https://en.wikipedia.org/wiki/Metaculus)\n- [Metaculus Crunchbase](https://www.crunchbase.com/organization/metaculus)\n- [Mastodon – Eugen Rochko TechCrunch interview](https://techcrunch.com/2022/12/23/mastodon-creator-eugen-rochko-talks-funding-and-how-to-build-the-anti-twitter/)\n- [Signal Foundation Wikipedia](https://en.wikipedia.org/wiki/Signal_Foundation)\n- [Signal statistics (Business of Apps)](https://www.businessofapps.com/data/signal-statistics/)\n- [Mozilla Corporation Wikipedia](https://en.wikipedia.org/wiki/Mozilla_Corporation)\n- [Mozilla State of 2024](https://www.mozilla.org/en-US/foundation/annualreport/2024/)\n- [Bluesky PBC announcement](https://bsky.social/about/blog/2-7-2022-overview)\n- [Bluesky business plan](https://bsky.social/about/blog/7-05-2023-business-plan)\n- [Kialo pricing / why it's free](https://www.kialo-edu.com/pricing-why-kialo-is-free)\n- [Hypothes.is / Anno (Library Journal)](https://www.libraryjournal.com/story/ithaka-invests-in-open-access-annotation-service-hypothesis)\n- [Wikimedia Foundation 2023-2024 Annual Report](https://wikimediafoundation.org/annualreports/2023-2024-annual-report/)\n- [Internet Archive Wikipedia](https://en.wikipedia.org/wiki/Internet_Archive)\n- [Vinnova – Innovativa Startups 2025](https://www.vinnova.se/en/calls-for-proposals/innovative-startups/innovative-startups-2025/)\n- [ALMI Innovation Loan](https://www.almi.se/en/loan/innovation-loan/)\n- [Knight Foundation Grants](https://knightfoundation.org/grants/)\n- [CE Sweden – Guide to Ideell Förening and Stiftelse](https://www.ce.se/a-guide-to-establishing-a-non-profit-ideell-forening-or-foundation-stiftelse-in-sweden/)\n- [Patreon statistics (Sci-Tech Today)](https://www.sci-tech-today.com/stats/patreon-statistics-updated/)\n- [Solo developer SaaS examples (Medium/CodeOrbit)](https://medium.com/@theabhishek.040/solo-developers-building-100k-1m-revenue-micro-saas-2024-110838470a2a)\n- [AI-native small teams (Freemius)](https://freemius.com/blog/state-of-micro-saas-2025/)\n\n*Sources added in the April 2026 addendum:*\n- [Survival and Flourishing Fund 2026 application](https://survivalandflourishing.fund/2026/application)\n- [Survival and Flourishing Fund — overview](https://survivalandflourishing.fund/)\n- [Jaan Tallinn — Future of Life Institute](https://futureoflife.org/person/jaan-tallinn/)\n- [Long-Term Future Fund (Effective Ventures Foundation)](https://funds.effectivealtruism.org/funds/far-future)\n- [Last Line of Defense: Minimum Viable Shelters for Mirror Bacteria — LessWrong](https://www.lesswrong.com/posts/2t2QuD3bPyWZ2aDBr/last-line-of-defense-minimum-viable-shelters-for-mirror)\n"}