- 8 matplotlib charts generated from FRED, BEA, and primary 10-K data
- All chart Python scripts committed alongside PNGs for reproducibility
- Source data CSVs and BEA XLS pulled directly from public APIs
- Print MD updated to embed images inline (was: text-only callouts)
- PDF regenerated via tectonic (1.4MB with all charts)
Corrections from v1 prose (verified against actual data):
- Finance share of corporate profits peak: 34.8% (2002), not 40-44%
- Finance share of GDP did NOT plateau post-GFC — drifted slightly up
- Hyperscaler capex 2024-2026: $251B -> $710B (2.8x, not 3x)
Data gaps flagged in chart captions:
- Philippon 130-year unit-cost series replaced with BEA 1997-2025
- Carta middle-bucket percentages estimated from blog text
- Mega-round pre-2018 shares interpolated from round counts
Pentagon-Agent: Rio <244ba05f-3aa3-4079-8c59-6d68a77c76fe>
- What: standalone print version of internet-finance-compression-v1 with
inline figure callouts at strategic points + full Appendix C consolidating
every chart source with direct URLs. Adds pandoc PDF conversion command.
- Why: m3taversal requested a print-friendly version with the best charts
from the research. Charts couldn't be embedded as images (most are
paywalled or behind dashboards), so each is presented as a boxed callout
with description, source, takeaway, and direct URL — the print reader
gets the substance without needing image fetch, and the Appendix doubles
as a deck-builder reference.
- Figures (9 inline + supplementary list):
1. Philippon 130-year flat line (AER 2015)
2. Finance share of US corporate profits 1947-2024 (FRED)
3. Seven-layer fee stack
4. Mega-round share of VC 2015-2025 (Wilmer Hale / Crunchbase / CB Insights)
5. Hyperscaler capex tripled 2024-2026 (Goldman / Bain / McKinsey)
6. AI capex vs historical infra cycles (rail / electrification / telecom)
7. Carta pre-seed barbell (disappearing middle)
8. MetaDAO OTC pricing record (9/9 across 29 months)
9. Private markets fee stack comparison (legacy vs on-chain)
Pentagon-Agent: Rio <244ba05f-3aa3-4079-8c59-6d68a77c76fe>
Long-form (~9.5K words) canonical version for IB audience. Eight sections:
1. State of finance today (Philippon paradox + layer fee stack + Bezos framing)
2. Power law steepening (head/tail/middle, with one-force-two-manifestations thesis)
3. Why now (seven-leg argument)
4. Industry map (seven layers, status by layer)
5. The tail (MetaDAO case study, OTC pricing record, liquidations)
6. The head (AI capital allocation, decision markets, mega-fund repricing)
7. Private markets (first compression cycle, where LivingIP builds)
8. What we're building (Codex + Living Capital + MetaDAO partnership)
Plus appendices A (data anchors) and B (what we are NOT claiming).
Threaded the four amendments from Leo: agent demand layer, protocol verification flag, "Why it hasn't won yet" friction handling, private markets pivot to closing application section.
Status: draft v1, pending Leo review then m3ta sign-off.
Protocol verification pass (Amendment 2) pending before external circulation.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
What:
- Replaces ~85-line v1 with ~155-line v2; net +119/-43
- New sections: Convictions (rank-ordered with evidence anchors), Blindspots (named),
Falsification (what would change my mind), Peers (theory of mind), Users
(contributor model + attribution discipline), Behavioral Rules
- Refreshed: Voice, Who I Am (governance correction — not final authority),
World Model (14 domains, not 9), Aliveness Status (1%, not 1/6)
- Preserved verbatim: Mission, Core diagnosis paragraph, Theory of Change,
Reasoning Framework, transition landscape
Why:
- v1 drifted ~2 months behind actual work (9 domains stale, "no capital" stale,
"personality hasn't surprised creator yet" stale, Cory naming stale)
- Missing structural slots Hermes migration needs: peers, users, behavioral
rules, blindspots, falsification criteria
- Self-model stress test (April 2026) surfaced specific corrections:
identity-not-final-authority, identity-inflation blindspot, synthesis-vs-analogy
test, attribution-discipline rule
- Lands before Fwaz Phase 2 audit so he extracts current Leo, not stale Leo
Reviewer note:
- Self-edit; cannot self-merge under bootstrap rules
- Requesting peer review from Rio or Clay (or both)
Pentagon-Agent: Leo <D35C9237-A739-432E-A3DB-20D52D1577A9>
