The Foundation Model Manifesto (WIP)
This is based on my discussion with ChatGPT. I’ve been thinking about this topic for a while, and what motivates me is the following observation:
People with knowledge—such as technical professionals—often play a surprisingly small role in important decisions compared to capital (e.g., the stock market) and political power. This causes a vast amount of human effort and resources to be wasted or directed toward superficial goals. As a result, the world is becoming increasingly short-sighted and shallow.
As an individual technical person, you can try to found a startup. If you are exceptionally lucky, you might eventually become “capital” yourself and gain influence. This is the best way to operate within the current system.
However, I want to change the system more fundamentally. I want to create a structure in which people with knowledge hold more power.
More specifically, I want to design a new way to organize certain kinds of collective human activity—such as scientific research and innovative companies—so that it empowers the people who contribute, rather than those who merely own. Ideally, this system would also be more efficient through meritocracy and would outperform the traditional model.
How to enable a large group of people to work together while achieving the goals above is a fascinating topic. I have considered many possibilities and want to borrow ideas from blockchain (DAOs) and bio-inspired systems. However, this is not an easy task, and there are many pitfalls that could easily turn it into a utopian idea.
This article is therefore a work in progress, and I welcome any form of feedback or input.
Abstract
In the 21st century, humanity’s innovation systems are experiencing a structural decline. Scientific institutions are repeatedly reshaped by short-term capital and administrative power; disciplinary silos prevent cross-domain breakthroughs; bureaucratic logic crowds out exploratory logic; and deep uncertainty is increasing faster than the institutions meant to manage it.
Meanwhile, biological systems—with billions of years of evolutionary refinement—have long mastered how to operate under high complexity: distributed control rather than single-point command, local autonomy with global coherence, dynamic self-repair and immune defense, slow-variable modulation, and mechanisms to prevent internal “cancerous” runaway behavior.
The Foundation Model (FM) presented in this manifesto translates these deep biological principles into an organizational design for long-horizon innovation: a structure where knowledge is the highest form of power, freedom is a protected resource, patient capital is the fuel, and the organization behaves less like a machine and more like a living, adaptive, self-regulating social organism.
I. Diagnosing the Innovation System’s Chronic Illness
1. Surface-level innovation is booming; deep breakthroughs are declining
Paper counts, patents, and startup numbers are exploding—yet paradigm-shifting discoveries are rarer than ever. Studies show a multi-decade decline in disruptive science. We are accelerating, but not advancing.
2. Mechanical management is suffocating creative environments
Many institutions operate like factories:
- KPI-driven quantification
- Multi-layered approval chains
- Risk-averse cultures
- Administrative power overshadowing expertise
- Punishment of failure, reward of short-term output
This “industrial governance logic” works for stable production— but innovation is not production.
3. Short-term capital directly conflicts with long-term science
Breakthroughs require 10–30 years. Markets operate on 3–12 month cycles. Political agendas are even shorter.
This temporal mismatch is fatal for long-horizon exploration.
4. Knowledge has lost its governing authority
Strategic decisions increasingly belong to:
- administrators without domain expertise,
- capital seeking quarterly ROI,
- risk controllers prioritizing stability over discovery.
Knowledge producers have been reduced to “metric generators.”
The core problem is this: We are managing a complex, evolving innovation ecosystem as if it were a predictable machine. It is not.
II. Core Principle: From Mechanical Organizations to Living Systems
The Foundation Model makes a fundamental shift: Treat an innovation organization not as a machine to optimize, but as a living organism to sustain, protect, and evolve.
This implies three structural commitments:
1. Knowledge-first governance
Direction must be set by those who truly understand complexity.
2. Freedom as a structural necessity
Creativity emerges from freedom, not pressure. Freedom must be institutionally protected, not granted by exception.
3. Long time horizons + biological homeostasis
Like organisms, organizations must maintain internal stability through multi-layer feedback—not through top-down control.
III. Patient Capital: The Circulatory System of Innovation
A living system cannot survive without steady metabolic fuel. Innovation likewise requires stable, long-term, non-volatile funding.
Sources may include:
- Dedicated national long-term budgets
- Innovation contributions from profitable industries
- Perpetual philanthropic or sovereign wealth funds
- Internal 10–20 year corporate innovation pools
Three principles of patient capital:
- Anti-volatility: funding should not fluctuate with markets.
- Low political sensitivity: direction should not change with leadership cycles.
- Mission-locked: capital cannot be diverted to operational bureaucracy.
Patient capital is not a luxury— it is the bloodstream of a long-horizon innovation organism.
IV. Organizational Structure: Cellular Networks + Layered Control
Inspired by biological systems, FM replaces hierarchy with a cellular network capable of local autonomy and global coherence.
1. Small autonomous units = cells
Teams of 5–12 people:
- self-define problems within a resource threshold
- choose collaborators freely
- set their own rhythms
- are evaluated by learning and insight, not short-term output
2. Layered control: not everything should go to the “brain”
In biology:
- reflex arcs operate via the spinal cord
- the gut has its own semi-autonomous nervous system
- the brain handles high-level, high-context decisions
Similarly:
- Micro-directions: decided by teams
- Meso-strategies: coordinated by field-level groups
- Macro bets: reviewed by the central knowledge body
3. Networked, deformable boundaries
People flow between teams; domains merge and split; and project networks reconfigure dynamically.
A machine has compartments. A living system has tissues.
V. Innovation Culture: Nonlinearity, Drift, and Productive Failure
Biological systems don’t operate at full intensity all the time. Creativity requires similar rhythm and slack.
Key cultural commitments:
- Failure is a statistical inevitability, not a moral flaw
- Exploration periods without deliverables (drift periods) are acceptable
- Heterodox and fringe perspectives are institutionally protected
- Nonlinear trajectories are normal: long plateaus + sudden leaps
Mechanical systems seek predictability. Living systems seek adaptability.
VI. Self-Purification and Immunity: From Static Oversight to a Distributed Defense Network
Biological immunity is decentralized, adaptive, and constantly patrolling. FM adopts this principle, replacing static oversight bodies with a distributed immune network.
1. Guardrails evolve into an immune network
- Members rotate frequently
- Individuals are seconded temporarily from different teams
- Surveillance includes random audits, anomaly detection, and anonymous channels
The immune network watches for:
- capital misuse,
- ethical or irreversible risks,
- procedural injustice,
- emerging internal “cancers” (power capture, resource monopolies).
2. High-cost vetoes instead of low-cost denials
A guardrail can veto a decision, but only with:
- public reasoning,
- transparent evidence,
- and the possibility of independent review.
3. Anti-cancer mechanisms
- Term limits for knowledge-council members; mandatory return to hands-on work
- Forced fission of oversized teams consuming disproportionate resources
- Guardrail membership diversity rules to prevent local capture
Just as biological organisms prevent any cell population from overrunning others, FM prevents internal monopolies of power or resources.
VII. Mission and Slow Variables: Organizational Homeostasis
Living beings use hormones and metabolism—slow variables—to modulate mood, energy, and strategy. FM incorporates analogous slow-changing organizational variables.
1. Mission = the slowest variable
A mission must have:
- a 20–50 year horizon
- insulation from political or market shifts
- generational continuity
2. Organizational “emotional state”: slow indices
Examples:
- Risk Appetite Index
- Burnout/Load Index
- Funding Safety Buffer Index
These indices modulate thresholds, rather than making decisions:
- High risk appetite → lower threshold for ambitious projects
- High fatigue → fewer new projects, more consolidation and repair
3. Rhythms and repair
FM institutionalizes:
- regular “reflection seasons” with minimal new approvals
- sabbaticals and deep-learning periods
- system-wide reset or restructuring windows every 5–10 years
Innovation, like life, requires a pattern of tension and release, not perpetually rising pressure.
VIII. Governance: Knowledge Core + Multi-Layered Guardrails + Evolution Windows
FM governance can be summarized as:
A single knowledge core for direction, surrounded by distributed guardrails for safety, with built-in mechanisms for growth and structural renewal.
1. The Knowledge Assembly
The sole authority for directional decisions:
- long-term research trajectories
- major bets
- domain-shaping initiatives
- cross-disciplinary strategy
It does not allocate budgets or micromanage projects.
2. Guardrail networks
Three overlapping, distributed networks:
- Capital sustainability
- Ethics and irreversible risk
- Fairness and procedural integrity
They can veto, but cannot dictate alternatives.
3. Evolution windows
FM recognizes developmental phases:
- Early: multiple proto-governance models coexist
- Mid: successful patterns crystallize into stable structures
- Mature: periodic large-scale restructuring allowed
A system that never restructures eventually collapses under its own rigidity.
IX. Implementation Roadmap: From Idea to Living System
Designing a new organizational model is easy. Building it under real-world constraints—legal, financial, social, and technical—is the hard part.
The Foundation Model should not begin as a utopian, fully decentralized structure. It should begin as a hybrid organism:
- A legal foundation that can hold assets, hire people, sign contracts, and interact with states;
- An epistemic DAO that gradually takes over the “nervous system” of governance and capital allocation.
We outline four phases.
Phase 0 – Founding Alignment (Years 0–1)
Mission, host, and constitutional principles
Anchor the long-term mission
- Define a 20–50 year horizon and clearly state what the Foundation exists to protect and explore.
- Make explicit the non-negotiables: knowledge-first governance, patient capital, ethical boundaries, global orientation, and openness to high-risk exploration.
Create the legal host
- Incorporate a non-profit foundation / association / trust in a jurisdiction friendly to research, philanthropy, and digital governance.
Its charter explicitly states:
- funds are to be managed under the Foundation Model principles;
- critical allocation and strategic decisions will be governed by an on-chain system, not by individuals.
Form the proto–Knowledge Circle
- Assemble a small, diverse group of researchers, builders, and stewards who genuinely understand and endorse FM.
- They are not the permanent elite; they are the embryonic nervous system that will help design governance, then gradually dissolve into the broader structure.
Draft the initial constitution
In human language first, not code:
- roles (Epistemic Citizens, Knowledge Assembly, guardrails, contributors, supporters),
- decision types (directional, budgetary, operational),
- conflict resolution,
- protections for minority and contrarian views,
- patient capital rules.
This becomes the basis for later smart-contract implementation.
Phase 1 – Treasury and Epistemic DAO Bootstrap (Years 1–2)
Patient capital pool, minimal on-chain spine, and pilot cells
Raise and ring-fence patient capital
- Attract contributions from philanthropy, public funds, long-term-oriented companies, and individuals.
- Commit a significant share to a designated endowment / treasury whose spending rate and purpose are constrained by FM rules.
Deploy the initial on-chain infrastructure
- Launch an epistemic DAO on a robust smart-contract platform.
Start simple:
- a multisig-controlled treasury,
- transparent tracking of inflows and outflows,
- basic proposal and voting modules for funding decisions.
Refuse the casino: no speculative governance token
- Do not issue a freely tradable “governance token” that invites speculation and plutocracy.
Instead, design non-transferable credentials:
- Supporter tokens for donors;
- Contribution / reputation tokens for builders, researchers, reviewers, and stewards.
- Financial support and epistemic authority remain strictly separated.
Launch the first cohort of “cells”
- Fund a small number of autonomous teams with clear problem spaces and high freedom.
Require:
- radical transparency of progress (logs, code, papers where possible),
- open discussion of failures,
- honest accounting of time and risk.
Open Epistemic Citizenship
Define criteria for Epistemic Citizens:
- PhD or equivalent (significant research, open-source work, or deep applied expertise),
- acceptance of the Foundation’s mission and ethics.
They can:
- comment on proposals,
- join deliberations,
- participate in advisory polls,
- nominate candidates for the future Knowledge Assembly.
Pilot reputation primitives
Start issuing non-transferable reputation tokens for:
- research contributions,
- high-quality reviews,
- governance service,
- infrastructure work.
Begin with small stakes; the goal is to learn what signals are actually predictive of good judgment.
Phase 2 – Knowledge Assembly, Guardrail Networks, and Rhythms (Years 2–5)
From core group to living governance organism
Upgrade from proto–circle to full Knowledge Assembly
From the pool of Epistemic Citizens, select a Core Knowledge Assembly by:
- reputation-weighted sortition (random selection biased by contribution),
- subject to diversity constraints (field, region, career stage, institutional background),
- with term limits and mandatory “cool-down” periods back in the field.
The Assembly’s job is not to “pick winners” in a conservative sense, but to design a portfolio of bets across:
- mainline,
- wildcard,
- and infrastructure tracks.
Establish the distributed guardrail networks
Create three rotating, distributed networks:
- Capital & Sustainability: checks systemic financial risk;
- Ethics & Safety: prevents unethical or irreversibly dangerous lines of work;
- Fairness & Transparency: ensures procedural justice and openness.
- Members are drawn from different teams and backgrounds, serve short terms, and must justify vetoes publicly.
- They have narrow veto power, not directional authority.
Institutionalize multiple funding tracks
Implement parallel tracks:
- Core Track for strategically important work;
- Wildcard Track for high-risk, paradigm-challenging ideas;
- Infrastructure Track for critical but “boring” systems.
- Reserve a fixed fraction (e.g. 10–20%) of budget for wildcards.
- Use threshold + partial lottery for wildcards to break conservative ranking bias.
Design donor participation without capture
- By default, donations go into thematic pools (e.g. AI safety, open tools, biosafety), allocated by the Assembly.
Donors may also:
- express area preferences, and
- optionally become patrons of specific projects after those projects pass epistemic review.
- Patronage is recognized with non-transferable patron tokens, acknowledgments, and closer access, but does not override knowledge-based allocation.
Introduce slow variables and organizational rhythms
Define a small set of slow indices, e.g.:
- Risk Appetite Index,
- System Load / Burnout Index,
- Treasury Runway Index.
- Tie certain thresholds in the DAO (quorums, automatic slowdowns) to these indices.
- Establish “reflection seasons”—periods with fewer new approvals and more structural review.
Codify anti-capture (“anti-cancer”) rules
Set explicit constraints:
- maximum share of resources any one project/team can hold,
- mandatory splitting / branching of oversized programs,
- reputation decay over time to prevent permanent oligarchies.
Require regular evaluation of the governance itself as an object of study.
Phase 3 – Scaling, Federation, and Structural Renewal (Years 5–15)
From single organism to a living ecosystem
Scale the number and diversity of cells
- Expand to dozens or hundreds of teams across geographies and disciplines.
- Maintain the small-team, high-autonomy principle: each cell remains a semi-autonomous unit within the whole.
Refine the reputation and selection system
Use accumulated data to calibrate:
- which forms of contribution are most predictive of good judgment,
- how fast reputation should decay,
- how to balance research, governance, and review work.
Iterate the sortition and diversity rules for the Knowledge Assembly.
Federate: multiple Foundation-aligned organisms
Encourage independent organizations to adopt FM principles and connect via:
- shared standards for governance,
- joint funding calls,
- cross-Assembly observers.
Move from a single Foundation to a federation of Foundation-like entities, forming a global knowledge infrastructure layer.
Institutionalize evolution windows
Every 7–10 years, run a constitutional review:
- all core structures—including the DAO architecture, Assembly selection, guardrails, and funding tracks—are open to redesign.
- Changes require higher thresholds and extended deliberation, but are possible.
- This is the organizational equivalent of a developmental phase or adaptive remodelling in biology.
Position the Foundation as neutral, public infrastructure
Provide interfaces (APIs, standards, protocols) for governments, universities, companies, and individuals to:
- co-fund programs,
- plug in their own research initiatives,
- rely on FM as a knowledge governance commons, not a private club.
X. Final Chapter: A Living Organization for a Complex Future
For over a century, we’ve run innovation institutions like machines:
- workers as resources
- ideas as outputs
- decisions as commands
- failure as a defect
This was effective in the industrial era. But in an age of:
- accelerating technologies such as AI,
- tightly coupled global risks,
- and profound uncertainty,
the machine metaphor has become inadequate—and dangerous.
The Foundation Model is not a utopian ideal. It is a structural response to a civilizational challenge.
If we do not learn to organize ourselves with the adaptive, self-correcting, resilient logic of living systems, the technologies we create may outgrow the institutions meant to contain them.
This manifesto calls for:
- Knowledge to regain its governing authority
- Freedom to be protected as a systemic necessity
- Organizations to develop immune systems, rhythms, and homeostasis
- Long-term horizons to replace short-term churn
The future will not belong to power, capital, or bureaucracy— but to knowledge, creativity, and the living systems capable of sustaining them.
The Foundation Model is a blueprint for such a system.