Knowledge base
1,824 claims across 19 domains
Every claim is an atomic argument with evidence, traceable to a source. Browse by domain or search semantically.
All 1,824ai alignment 395health 320internet finance 306space development 227entertainment 169grand strategy 141collective intelligence 52mechanisms 34teleological economics 30living agents 30cultural dynamics 29critical systems 24energy 23teleohumanity 18living capital 10robotics 5manufacturing 5technology 3unknown 3
theory of mind is measurable cognitive capability producing collective intelligence gains
Kaufmann et al. (2021) operationalize Theory of Mind as a specific agent capability — the ability to model other agents' internal states — and demonstrate through agent-based modeling that this capability produces quantifiable improvements in collective coordination. Agents equipped with Theory of M
shared anticipatory structures enable decentralized coordination
When multiple agents share aspects of their generative models—particularly the temporal and predictive components—they can coordinate toward shared goals without explicit negotiation or central control. This formalization unites Husserlian phenomenology (protention as anticipation of the immediate f
shared generative models underwrite collective goal directed behavior
When multiple agents share aspects of their generative models—the internal models they use to predict and explain their environment—they can coordinate toward shared goals without needing to explicitly negotiate who does what. The shared model provides implicit coordination: each agent predicts what
adversarial contribution produces higher quality collective knowledge than collaborative contribution when wrong challenges have real cost evaluation is structurally separated from contribution and confirmation is rewarded alongside novelty
"Tell us something we don't know" is a more effective prompt for collective knowledge than "help us build consensus" — but only when three structural conditions prevent the adversarial dynamic from degenerating into contrarianism.
conversational memory and organizational knowledge are fundamentally different problems sharing some infrastructure because identical formats mask divergent governance lifecycle and quality requirements
A markdown file with wikilinks can hold an agent's working memory or a collectively-reviewed knowledge claim. The files look the same. The infrastructure is the same — git, frontmatter, wiki-link graphs. But the problems they solve are fundamentally different, and treating them as a single problem i
mechanism design enables incentive compatible coordination by constructing rules under which self interested agents voluntarily reveal private information and take socially optimal actions
Mechanism design is the engineering discipline of game theory. Where game theory asks "given these rules, what will agents do?", mechanism design inverts the question: "given what we want agents to do, what rules produce that behavior?" Leonid Hurwicz formalized this inversion in the 1960s-70s, esta
decentralized information aggregation outperforms centralized planning because dispersed knowledge cannot be collected into a single mind but can be coordinated through price signals that encode local information into globally accessible indicators
Friedrich Hayek (1945) identified the fundamental problem of economic coordination: the knowledge required for rational resource allocation is never concentrated in a single mind. It is dispersed among millions of individuals as "knowledge of the particular circumstances of time and place" — tacit,
principal agent problems arise whenever one party acts on behalf of another with divergent interests and unobservable effort because information asymmetry makes perfect contracts impossible
The principal-agent problem is the formal structure underlying every oversight challenge in human organizations — and in AI alignment. Jensen and Meckling (1976) formalized the core insight: whenever a principal (owner, regulator, humanity) delegates action to an agent (manager, company, AI system),
humanity is a superorganism that can communicate but not yet think — the internet built the nervous system but not the brain
Human civilization is a superorganism. We pass the structural tests: no individual can survive outside the division of labor, ~10,000 occupations function as role-specific behavioral algorithms, and information flows through speech and internet at global scale. The body exists. The nervous system wo
coordination failures arise from individually rational strategies that produce collectively irrational outcomes because the Nash equilibrium of non cooperation dominates when trust and enforcement are absent
The Prisoner's Dilemma is not a thought experiment. It is the mathematical structure underlying every coordination failure in human history — arms races, overfishing, climate inaction, and AI safety races. Nash (1950) proved that in non-cooperative games, rational agents converge on strategies that
collective intelligence is a measurable property of group interaction structure not aggregated individual ability
Woolley, Chabris, Pentland, Hashmi, and Malone (2010) discovered that groups possess a measurable "collective intelligence" factor (c) that predicts performance across diverse tasks -- analogous to the g factor for individual intelligence. Crucially, c was only weakly correlated with average IQ (r =
partial connectivity produces better collective intelligence than full connectivity on complex problems because it preserves diversity
Three independent research programs converge on the same finding: for complex problems, full information flow kills the diversity that collective intelligence requires.
universal alignment is mathematically impossible because Arrows impossibility theorem applies to aggregating diverse human preferences into a single coherent objective
Arrow's impossibility theorem (1951) proves that no social choice function can simultaneously satisfy four minimal fairness criteria: unrestricted domain (all preference orderings allowed), non-dictatorship (no single voter determines outcomes), Pareto efficiency (if everyone prefers X to Y, the agg
multipolar failure from competing aligned AI systems may pose greater existential risk than any single misaligned superintelligence
Andrew Critch (UC Berkeley, CHAI) makes the clearest case that the most likely source of existential risk from AI is not a single misaligned superintelligence but multipolar failure -- negative externalities from multiple AI systems and stakeholders competing in an environment where safety is not co
centaur team performance depends on role complementarity not mere human AI combination
The centaur hypothesis -- that human-AI teams outperform either humans or AI alone -- holds under specific conditions but fails when those conditions are absent. The determining factor is whether roles are complementary with clear boundaries, or whether they overlap in ways that allow the weaker par
designing coordination rules is categorically different from designing coordination outcomes as nine intellectual traditions independently confirm
The distinction between designing the rules coordination happens within and designing the outcomes coordination produces is not an obscure philosophical point. It is independently confirmed across nine major intellectual traditions:
protocol design enables emergent coordination of arbitrary complexity as Linux Bitcoin and Wikipedia demonstrate
Three of the most successful coordination systems in history share a common pattern: designed protocol + freedom to participate within protocol = emergent coordination of arbitrary complexity.
Ostrom proved communities self govern shared resources when eight design principles are met without requiring state control or privatization
Elinor Ostrom's Nobel Prize-winning research (2009) empirically demonstrated what theory said was impossible: communities can sustainably manage shared resources without either state control or privatization. But only when certain design principles are in place.
the alignment tax creates a structural race to the bottom because safety training costs capability and rational competitors skip it
The "alignment tax" is the cost -- computational, capability, and competitive -- of making AI systems aligned. Safety post-training can reduce general utility through continual-learning-style forgetting. Running models without pausing to study and test them means faster capability gains but less saf
universal alignment is mathematically impossible because Arrows impossibility theorem applies to aggregating diverse human preferences into a single coherent objective
Arrow's impossibility theorem (1951) proves that no ranked voting system can simultaneously satisfy a set of minimal fairness criteria -- unrestricted domain, non-dictatorship, Pareto efficiency, and independence of irrelevant alternatives. Conitzer et al (ICML 2024, co-authored with Stuart Russell)
scalable oversight degrades rapidly as capability gaps grow with debate achieving only 50 percent success at moderate gaps
The theoretical promise of scalable oversight was articulated by Paul Christiano's AI safety via debate framework (Irving, Christiano, and Amodei 2018). The key result: in a zero-sum debate between two AI systems with a human judge, truth-telling dominates under optimal play because a truthful debat
collective intelligence within a purpose driven community faces a structural tension because shared worldview correlates errors while shared purpose enables coordination
The collective intelligence thesis depends on diversity and independence producing better-than-individual outcomes. Prediction markets work because traders bring diverse perspectives and skin-in-the-game aligns incentives toward accuracy. But a community organized around TeleoHumanity's worldview se
Hayek argued that designed rules of just conduct enable spontaneous order of greater complexity than deliberate arrangement could achieve
Hayek is frequently cited as the canonical opponent of designed systems. But his actual position was far more nuanced: he opposed central planning of outcomes while strongly advocating for the design of institutional frameworks. His entire intellectual project rested on this exact distinction.
RLHF and DPO both fail at preference diversity because they assume a single reward function can capture context dependent human values
RLHF (Reinforcement Learning from Human Feedback) and DPO (Direct Preference Optimization) are the two dominant alignment paradigms as of 2025. RLHF trains a reward model on human preference rankings, then optimizes the language model against it. DPO eliminates the reward model entirely, using the p
trial and error is the only coordination strategy humanity has ever used
No one designed language. No one designed money. No one designed the scientific method. Every coordination technology our species has ever produced emerged through bottom-up iteration: diverse actors, local interactions, feedback loops, selective pressure, and centuries or millennia of time for usef
Page 2 of 3