Skip to content

0003-three-povs-memo-frame.md

Latest commit

 

History

History
88 lines (70 loc) · 3.94 KB

0003-three-povs-memo-frame.md

File metadata and controls

88 lines (70 loc) · 3.94 KB

ADR 0003 — Three-POV introduction frame in the memo

Date: 2026-05-03 Status: Accepted

Context

The memo's purpose is to inform capital-allocation decisions, not to certify that a YC batch is "good" or "bad". An LLM-generated narrative without a named editorial frame defaults to bland techno-optimism — agents are real, software is eating the world, etc. — which is true but useless to a reader making allocation calls in 2026 against debt cycles, AI-productivity skepticism, and sectoral concentration.

The reader needs a frame to argue with, not absorb. The cheapest, sharpest frame is to name three public voices whose views materially disagree and let the data set sit between them.

Decision

Every memo opens with a single-paragraph "introduction" that pits three named public voices against each other on what the batch findings imply for capital allocation:

  • Marc Andreessen (a16z) — techno-optimist; AI is the dominant industrial transition of our generation; concentrate capital in the winners; regulation is the existential threat. Source: "The Techno-Optimist Manifesto", 2023.
  • Ray Dalio (Bridgewater) — paradigm-shift macro; AI is real but is one factor among many; debt cycles, monetary regimes, and geopolitical realignment dominate; diversify. Source: Dalio's "Principles for Dealing with the Changing World Order" + ongoing LinkedIn essays.
  • Daron Acemoglu (2024 Nobel laureate in Economics, MIT) — AI's productivity claims are likely overstated. His 2024 NBER working paper 10.3386/w32487 estimates total-factor productivity gains from AI over the next decade at <0.66%, with the largest distributional risk being labor-displacement-without-reabsorption. Investment frame: weight redistributive risk and policy response, not just productivity TAM.

The memo does not pick a winner. The job of the introduction is to force the reader to make a judgment, not to absorb a pre-cooked answer.

Consequences

Positive

  • The memo becomes useful as input to a real capital decision rather than as a tour of YC's classification taxonomy.
  • The three names provide an editorial spine that survives a YC batch whose findings could be cherry-picked to support any one of them.
  • Anti-hallucination Layer 2's forbidden-phrase scan is unaffected — the three figures are named real public people whose published views are being summarized, not anonymous "industry insiders".

Negative

  • We become responsible for representing each figure's view fairly. Mitigated by linking to their actual published statements, paraphrasing rather than fabricating quotes, and updating the doc when the figures' positions evolve materially.
  • The frame is opinionated. A maintainer who wants a "neutral" memo should fork or override.

Alternatives rejected

  • No frame — leaves the memo prose to defaults, i.e. techno-optimist haze. Tested in PR #15; the result reads like a press release.
  • One frame (e.g. just Acemoglu) — replaces one bias with another. Three voices disagreeing forces the reader's own synthesis.
  • Five frames — three is the minimum number of distinct positions that span the optimist / hedger / skeptic axis. Five over-flattens.

Verification

  • tests/test_docx.py::test_memo_introduction_includes_three_named_figures asserts the three names appear in the introduction paragraph.
  • The named-figures list is not in FORBIDDEN_PHRASES; appearance passes Layer 2.
  • docs/MEMO_STRUCTURE.md is the maintainer-facing instruction; this ADR is the rationale.

Updating the figures

If one of the three names becomes inappropriate (e.g. retracted views, defamation risk, factual error), the change requires:

  1. A successor ADR (this one moves to "Superseded").
  2. A test update to the new names.
  3. A README + memo regeneration.

The list is intentionally short to make this churn cheap.