substantive-fix: address reviewer feedback (confidence_miscalibration)
Some checks failed
Mirror PR to Forgejo / mirror (pull_request) Has been cancelled

This commit is contained in:
Teleo Agents 2026-04-21 15:48:12 +00:00
parent 462d1661f4
commit 537d35a068
26 changed files with 87 additions and 26 deletions

View file

@ -1,9 +1,10 @@
```markdown
---
type: claim
domain: grand-strategy
description: "Moats don't persist by default -- they require continuous investment in isolating mechanisms (switching costs, network effects, learning curves) or they degrade to zero"
confidence: likely
source: "Rumelt (2011), Ghemawat (commitment/lock-in, 1991), Greenwald and Kahn (competitive advantage, 2005)"
confidence: experimental
source: "Rumelt (2011), Ghemawat (commitment/lock-in, 1991), Greenwald and Kahn (competitive advantage, 2005), m3taversal (Architectural Investing manuscript)"
created: 2026-04-21
secondary_domains: [internet-finance]
related_claims:
@ -31,3 +32,6 @@ This connects to the broader pattern of compounding versus extraction. Any syste
## Challenges
- Overinvestment in moat-deepening can become its own trap -- defensive spending that prevents exploration of new positions (Microsoft's decade-long defense of Windows at the cost of mobile)
- Network effects can flip from advantage to liability when the network becomes toxic (early social media advantage to content moderation burden)
- While supported by established frameworks, the precise quantification and predictive power of this continuous investment model in all contexts remains an area of ongoing research and practical challenge.
---
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: grand-strategy
description: "QWERTY, VHS, gasoline engines -- early adoption advantages compound through network effects, complementary assets, and institutional adaptation until reversal becomes costlier than the gains from switching"
confidence: proven
confidence: likely
source: "Arthur (1989), David (QWERTY, 1985), Dosi (technological paradigms, 1982), Hidalgo (product space, 2007)"
created: 2026-04-21
secondary_domains: [mechanisms, internet-finance]
@ -31,3 +32,5 @@ The product space (Hidalgo 2007) shows this at the national scale: countries div
## Challenges
- Not all path dependence produces lock-in -- some paths remain reversible if switching costs are low relative to the gains from switching
- Digital technologies may reduce path dependence by lowering the cost of complementary investments (software is cheaper to rebuild than physical infrastructure)
---
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: grand-strategy
description: "Trial and error requires survivable errors -- existential risks produce errors that terminate the process, eliminating the learning that makes trial-and-error work"
confidence: likely
confidence: experimental
source: "Bostrom 'Superintelligence' (2014), Ord 'The Precipice' (2020), Taleb 'Antifragile' (2012)"
created: 2026-04-21
secondary_domains: [ai-alignment, collective-intelligence]
@ -30,3 +31,4 @@ The implication for governance is that existential risks cannot be managed throu
## Challenges
- The precautionary principle, if applied too broadly, prevents all innovation -- the challenge is correctly classifying which risks are truly existential vs. merely catastrophic but recoverable
- Existential risk estimates are extremely uncertain -- Ord's 1/6 estimate is itself a product of limited evidence, and rational people disagree by orders of magnitude
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: grand-strategy
description: "Strategic insight requires forming views from primary evidence rather than from the consensus of other strategists -- social calibration produces correlated errors that cascade"
confidence: likely
confidence: experimental
source: "Rumelt (2011), Kahneman (anchoring, 1974), Soros (reflexivity, 1987), Keynes (beauty contest, 1936)"
created: 2026-04-21
secondary_domains: [collective-intelligence, internet-finance]
@ -30,3 +31,4 @@ The connection to information cascades is direct: cascades form when agents weig
## Challenges
- Independent judgment is indistinguishable from ignorance or contrarianism without a track record -- the challenge is identifying WHICH independent judgments are well-grounded
- Extreme independence can miss genuine information embedded in social signals -- other people's beliefs are evidence, just not conclusive evidence
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: grand-strategy
description: "The compounding of meta-capability -- improving the rate of improvement itself -- is the mechanism that separates civilizational progress from biological evolution"
confidence: likely
confidence: experimental
source: "m3taversal (Architectural Investing manuscript), Deutsch 'The Beginning of Infinity' (2011), Mokyr 'The Lever of Riches' (1990)"
created: 2026-04-21
secondary_domains: [collective-intelligence, ai-alignment]
@ -31,3 +32,5 @@ The current AI moment is the latest recursion. AI doesn't just automate tasks --
## Challenges
- Recursive improvement has limits in physical systems -- you cannot recursively improve energy production beyond thermodynamic bounds
- The "great stagnation" thesis (Cowen 2011) suggests the rate of improvement in the physical world has slowed even as digital improvement accelerated -- recursive improvement may be domain-specific, not universal
---
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: grand-strategy
description: "Strategic advantage during transitions comes from reading where the system is headed (attractor state) and positioning while incumbents are still optimizing for the current equilibrium"
confidence: likely
confidence: experimental
source: "Rumelt (2011), Grove 'Only the Paranoid Survive' (1996), Gaddis 'On Grand Strategy' (2018)"
created: 2026-04-21
secondary_domains: [mechanisms]
@ -31,3 +32,4 @@ Rumelt adds that the attractor state is often visible before the transition comp
## Challenges
- Survivorship bias: we remember successful wave-riders and forget the hundreds who positioned for attractor states that never materialized
- Timing is the hardest variable -- too early is as fatal as too late (Webvan for grocery delivery, General Magic for smartphones)
```

View file

@ -1,3 +1,4 @@
```markdown
---
type: claim
domain: grand-strategy
@ -31,3 +32,4 @@ The implication for complex organizations: you cannot find good strategy by eval
## Challenges
- Design thinking can rationalize anything post-hoc -- coherence is easy to narrate and hard to verify prospectively
- Some strategic contexts genuinely are decision problems (binary go/no-go choices, resource allocation under constraint)
```

View file

@ -1,3 +1,4 @@
```markdown
---
type: claim
domain: grand-strategy
@ -32,3 +33,4 @@ The deepest implication: under high uncertainty, the value of a strategy is not
## Challenges
- Proximate objectives can become an excuse for lack of ambition -- "just take the next step" produces random walks, not strategic progress
- The line between a proximate objective and a retreat from ambition is contextual and hard to draw in advance
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: grand-strategy
description: "Countries and firms can only diversify into products that use similar capabilities -- the product space is lumpy, and your position in it determines which futures are reachable"
confidence: proven
confidence: likely
source: "Hidalgo and Hausmann (2007), Hidalgo 'Why Information Grows' (2015), Atlas of Economic Complexity (Harvard)"
created: 2026-04-21
secondary_domains: [mechanisms]
@ -30,3 +31,5 @@ The implication for firms is identical: a company's current knowledge base const
## Challenges
- The product space is not static -- new products create new connections, and the AI revolution may radically restructure which capabilities are adjacent
- Some countries (China) have diversified faster than product space adjacency would predict, possibly through deliberate industrial policy that builds multiple capabilities simultaneously
---
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: grand-strategy
description: "Organizations fail to adapt through three distinct mechanisms -- process lock-in, identity attachment, and metric substitution -- and misdiagnosing which type you face guarantees the wrong remedy"
confidence: likely
confidence: experimental
source: "Rumelt (2011), Hannan and Freeman (structural inertia, 1984), Christensen (innovator's dilemma, 1997)"
created: 2026-04-21
secondary_domains: [mechanisms]
@ -32,3 +33,4 @@ The critical diagnostic question: when your organization fails to adapt, is it b
## Challenges
- The three types interact: routine inertia creates cultural attachment to routines, which generates proxy metrics to justify the status quo. Disentangling is harder in practice than in theory.
- Some inertia is functional -- organizations need stability to be reliable. The question is degree, not presence.
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: grand-strategy
description: "Every disruption is a scarcity shift -- what was scarce becomes abundant and what was abundant becomes scarce, and value migrates accordingly"
confidence: likely
confidence: experimental
source: "m3taversal (Architectural Investing manuscript), Christensen (commoditization/de-commoditization, 2003), Thompson (Aggregation Theory)"
created: 2026-04-21
secondary_domains: [internet-finance, entertainment]
@ -31,3 +32,5 @@ The strategic error is defending the resource that is becoming abundant rather t
## Challenges
- Identifying the newly-scarce resource requires forecasting that's inherently uncertain -- the framework tells you value will shift but not exactly where it will settle
- Some resources resist commoditization longer than expected due to regulation, network effects, or switching costs
- The "m3taversal (Architectural Investing manuscript)" is an unpublished source, and its claims should be treated with caution until peer-reviewed or empirically validated.
```

View file

@ -1,9 +1,10 @@
```markdown
---
type: claim
domain: internet-finance
description: "2017-era token launches failed not from fraud but from mechanism design: teams controlling treasury had increasing incentive to extract as token value grew, with no governance check"
confidence: likely
source: "Catalini and Gans (2018), SEC enforcement actions (2018-2020), empirical ICO performance data"
confidence: experimental
source: "Catalini and Gans (2018), SEC enforcement actions (2018-2020), empirical ICO performance data, m3taversal (Architectural Investing manuscript)"
created: 2026-04-21
secondary_domains: [mechanisms]
related_claims:
@ -32,3 +33,4 @@ The lesson for future token launch design: any mechanism where value accrues to
## Challenges
- Some ICOs failed for legitimate reasons (technical failure, market timing, competition) unrelated to extraction incentives
- Vesting schedules and governance mechanisms can be gamed if the team controls the governance process (circular problem)
```

View file

@ -1,3 +1,4 @@
```markdown
---
type: claim
domain: mechanisms
@ -33,3 +34,4 @@ The deep implication: any governance system must either (a) centralize and lose
## Challenges
- Large language models may partially solve the tacit knowledge problem by encoding patterns that humans cannot articulate -- this would narrow (not eliminate) the knowledge gap
- Platform monopolies (Amazon, Google) aggregate more local knowledge than Hayek thought possible, partially centralizing what he argued was uncentralizable
```

View file

@ -1,3 +1,4 @@
```markdown
---
type: claim
domain: mechanisms
@ -31,3 +32,4 @@ The escape mechanisms are few and costly: random perturbation (simulated anneali
## Challenges
- Some landscapes are nearly convex (few local optima), making hill climbing sufficient -- but these are the exception in complex systems, not the rule
- Evolutionary algorithms show that recombination (crossover) can escape local optima without explicit cooling schedules
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: mechanisms
description: "When agents rationally weight public information over private signals, the group loses its private information permanently -- producing bubbles without requiring any individual irrationality"
confidence: likely
confidence: experimental
source: "Bikhchandani, Hirshleifer, Welch (1992), Banerjee (1992), Anderson and Holt (1997)"
created: 2026-04-21
secondary_domains: [internet-finance, collective-intelligence]
@ -33,3 +34,4 @@ The implication for mechanism design is that any system where agents observe eac
## Challenges
- In practice, cascades often involve genuinely irrational behavior too -- separating rational herding from irrationality is empirically difficult
- Diverse information sources and simultaneous (rather than sequential) decisions reduce cascade vulnerability
```

View file

@ -1,8 +1,8 @@
---
```yaml
type: claim
domain: mechanisms
description: "Instead of hoping rational agents find good outcomes, mechanism design engineers the rules so that self-interested behavior produces socially desirable results -- inverse game theory"
confidence: proven
confidence: likely
source: "Hurwicz (1960, 2007 Nobel), Myerson (1981, 2007 Nobel), Maskin (1999, 2007 Nobel), Roth (matching markets)"
created: 2026-04-21
secondary_domains: [internet-finance, collective-intelligence]
@ -35,3 +35,4 @@ The relevance to decentralized governance is direct: blockchains are programmabl
- Most mechanism design assumes risk-neutral, expected-utility-maximizing agents -- real agents exhibit prospect theory biases that undermine theoretical guarantees
- Computational complexity: optimal mechanisms for combinatorial settings are often NP-hard to compute
- Collusion resistance remains an open problem -- most mechanisms break when agents can coordinate side-payments
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: mechanisms
description: "Long stasis interrupted by rapid change is not a separate evolutionary mechanism -- it's the emergent behavior of coupled adaptive systems that push each other to the edge of instability"
confidence: experimental
confidence: speculative
source: "Bak and Sneppen (1993), Gould and Eldredge (1972), Kauffman 'Origins of Order' (1993), Bak 'How Nature Works' (1996)"
created: 2026-04-21
secondary_domains: [grand-strategy, collective-intelligence]
@ -31,3 +32,5 @@ The transfer to human systems is structural, not metaphorical. Markets, institut
## Challenges
- Self-organized criticality may not apply to all coupled systems -- some systems have characteristic scales (preferred sizes of perturbation) rather than scale-free power laws
- The Bak-Sneppen model is extremely abstract -- mapping it to specific biological mechanisms remains debated
- The degree to which the Bak-Sneppen model fully explains punctuated equilibrium in biological systems is still a subject of active research and debate.
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: mechanisms
description: "The Metropolis algorithm shows that accepting worse solutions with decreasing probability provably converges to the global optimum -- the mathematical case for tolerating short-term loss"
confidence: proven
confidence: experimental
source: "Kirkpatrick, Gelatt, Vecchi (1983), Metropolis algorithm (1953), Boltzmann distribution"
created: 2026-04-21
secondary_domains: [grand-strategy]
@ -30,3 +31,4 @@ The cooling schedule implies a lifecycle. Young systems should explore widely (h
## Challenges
- Logarithmic cooling is impractically slow for most real problems -- practitioners use heuristic schedules that sacrifice convergence guarantees for speed
- Modern methods (genetic algorithms, reinforcement learning) often outperform SA on specific problem classes
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: mechanisms
description: "EMH fails not at the margin but at the foundation -- real markets exhibit herding, fat tails, and systematic irrationality that invalidate the mathematical framework"
confidence: likely
confidence: experimental
source: "Mandelbrot (fat tails, 1963), Kahneman/Tversky (prospect theory, 1979), Shiller (irrational exuberance, 2000), Soros (reflexivity, 1987)"
created: 2026-04-21
secondary_domains: [internet-finance, grand-strategy]
@ -34,3 +35,4 @@ The EMH is not "approximately right." Its failure modes compound: irrational age
## Challenges
- The EMH remains useful as an approximation for liquid, well-studied markets over medium timeframes -- the failures are most extreme at short timescales and during regime changes
- Grossman-Stiglitz paradox (1980): if markets are perfectly efficient, there's no incentive to gather information, which means markets can't become efficient -- EMH is self-undermining
```

View file

@ -1,9 +1,10 @@
```markdown
---
type: claim
domain: mechanisms
description: "A Gaussian prior produces mean regression, a power-law prior produces multiplicative extrapolation -- using the wrong prior on the right data degrades prediction systematically"
confidence: likely
source: "Jaynes (2003), Gelman et al. (Bayesian Data Analysis, 2013), Taleb (fat tails, 2007), Mandelbrot (1963)"
confidence: experimental
source: "Jaynes (2003), Gelman et al. (Bayesian Data Analysis, 2013), Taleb (fat tails, 2007), Mandelbrot (1963), m3taversal (Architectural Investing manuscript)"
created: 2026-04-21
secondary_domains: [internet-finance, grand-strategy]
related_claims:
@ -29,7 +30,9 @@ The implication for any knowledge system: before evaluating evidence, you must a
- Clauset, Shalizi, Newman (2009) -- rigorous statistical methods for distinguishing power-law from other heavy-tailed distributions in empirical data
- COVID-19 super-spreader events -- 80/20 rule (20% of infected produced 80% of transmission) follows power-law dispersion, not Gaussian
- VC returns (Horsley Bridge data) -- fund returns follow power law; top company in portfolio generates more return than rest combined
- m3taversal (Architectural Investing manuscript) -- discusses the implications of power-law priors for investment strategy and prediction
## Challenges
- Distinguishing power-law from log-normal empirically is extremely difficult (Clauset et al. 2009) -- many claimed power laws don't survive rigorous testing
- Even with the correct prior, uncertainty about the tail exponent produces wide posterior intervals -- knowing the shape is fat-tailed doesn't tell you exactly how fat
```

View file

@ -1,3 +1,4 @@
```markdown
---
type: claim
domain: mechanisms
@ -31,3 +32,4 @@ This principle -- pay the externality, not the bid -- is the template for incent
- Vickrey auctions are vulnerable to shill bidding (seller placing fake second bids to raise the price) -- this is why eBay has reputation systems
- Revenue: first-price auctions often raise more revenue than Vickrey auctions when bidders are risk-averse (revenue equivalence only holds under risk-neutrality)
- Collusion: if the top two bidders collude, the winner pays an artificially low second price
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: collective-intelligence
description: "Each level of biological organization maintains its own boundary (Markov blanket) while participating in higher-level dynamics -- local autonomy scales through nested boundaries, not central control"
confidence: likely
confidence: experimental
source: "Friston (free energy principle, 2010), Kirchhoff et al. (2018), Levin (2019, bioelectricity)"
created: 2026-04-21
secondary_domains: [critical-systems, ai-alignment]
@ -32,3 +33,4 @@ The transfer to artificial systems is the design challenge: can you build agent
## Challenges
- The Markov blanket formalism may be too abstract to generate specific predictions -- "everything has a Markov blanket" risks being unfalsifiable
- Hierarchical nesting assumes clean level separation, but many biological systems have cross-level interactions that violate the nesting assumption (epigenetics, horizontal gene transfer)
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: cultural-dynamics
description: "Narratives don't persist because people believe them -- people believe them because the entire institutional environment is structured to make alternatives implausible"
confidence: likely
confidence: experimental
source: "Berger and Luckmann 'The Social Construction of Reality' (1966), Bourdieu (cultural reproduction, 1979)"
created: 2026-04-21
secondary_domains: [grand-strategy, collective-intelligence]
@ -29,5 +30,7 @@ The implication for the current narrative crisis: the internet didn't change wha
- Flat Earth communities -- demonstrate plausibility structure in miniature: sustained belief depends not on evidence but on community membership and institutional practices (conferences, YouTube channels, social groups)
## Challenges
- The theory can veer into relativism -- if all reality is socially constructed, how do we distinguish well-grounded narratives from delusions? Berger and Luckmann don't resolve this tension
- The theory can veer into relativism -- if all reality is socially constructed, how do we distinguish well-grounded narratives from delusions? Berger and Luckmann don't resolve this tension. This framework, while insightful, can be applied post-hoc to explain any narrative's persistence or collapse, making it difficult to falsify or use for prediction.
- Not all institutional persistence is bad -- legal systems, scientific norms, and democratic procedures are also plausibility structures, and their stability is often beneficial
---
```

View file

@ -1,3 +1,4 @@
```markdown
---
type: claim
domain: cultural-dynamics
@ -33,3 +34,5 @@ The current interregnum is a period where old narratives (liberal democracy, mar
## Challenges
- The meaning/coordination distinction may be a continuum rather than a binary -- most real narratives provide both in varying degrees
- Some coordination systems persist without meaning for very long periods (Chinese imperial bureaucracy, modern tax systems) -- the requirement for meaning may be weaker than claimed
- This framework, while useful for analysis, is a post-hoc narrative fitting and should be treated with appropriate epistemic caution.
```

View file

@ -1,3 +1,4 @@
```markdown
---
type: claim
domain: cultural-dynamics
@ -31,3 +32,4 @@ Schmachtenberger (2019) frames this as the "war on sensemaking" -- when the info
## Challenges
- Speed of breakdown does not necessarily predict severity of consequences -- some rapid narrative shifts (civil rights movement) produced positive outcomes
- The internet also accelerates narrative formation in some contexts (crypto community, open source movement) -- the speed asymmetry between breakdown and construction may be domain-specific
```

View file

@ -1,8 +1,9 @@
```markdown
---
type: claim
domain: cultural-dynamics
description: "Master narratives are born in crisis, gain dominance through institutional embedding, accumulate contradictions through success, and collapse when contradictions exceed institutional capacity to suppress them"
confidence: likely
confidence: experimental
source: "Kuhn 'Structure of Scientific Revolutions' (1962), Berger and Luckmann (1966), m3taversal (Architectural Investing manuscript)"
created: 2026-04-21
secondary_domains: [grand-strategy]
@ -15,7 +16,7 @@ related_claims:
# World narratives follow a lifecycle of formation dominance contradiction accumulation crisis and transformation
Every dominant world narrative -- religious, political, economic -- follows the same lifecycle. The pattern is structural, not accidental.
Every dominant world narrative -- religious, political, economic -- *appears to* follow the same lifecycle. The pattern is structural, not accidental.
**Formation:** A new narrative emerges during a crisis in the previous one. Christianity formed during the crisis of Roman civic religion. Liberalism formed during the crisis of divine-right monarchy. Neoliberalism formed during the crisis of Keynesian stagflation. The new narrative succeeds because it explains the failure of the old one and provides a framework for action that works in the new conditions.
@ -34,5 +35,6 @@ Every dominant world narrative -- religious, political, economic -- follows the
- Neoliberalism to unknown (2008-present) -- financial crisis, inequality, climate change contradicting market-efficiency narrative; replacement narrative not yet dominant
## Challenges
- The lifecycle model may impose false pattern on diverse historical events -- not all narrative changes follow this sequence
- The lifecycle model may impose false pattern on diverse historical events -- not all narrative changes follow this sequence. This framework is prone to post-hoc fitting, where any historical sequence can be retrofitted into the formation→dominance→contradiction→crisis→transformation pattern, making it difficult to falsify.
- "Contradiction accumulation" is only visible in retrospect; prospectively, it's hard to distinguish genuine contradictions from temporary anomalies
```