source_doc

active

source_doc:users-antonborzov-documents-research-nosync-alexander-applied-koan-battery-section-md

koan-battery-section.md

/Users/antonborzov/Documents/Research.nosync/alexander/applied/koan-battery-section.md

Frontmatter (2 fields)

{
  "path": "/Users/antonborzov/Documents/Research.nosync/alexander/applied/koan-battery-section.md",
  "last_claims_extracted_at": 1778899531
}

Outgoing (0)

None.

Incoming (11)

Source of (11)

Aesthetics is a separable axis in AI evaluation, partially independent from a single latent performance factor.(claim)
Capability can become polish, and polish can diminish life—Alexander's central insight empirically demonstrated in AI outputs.(claim)
Forced-choice pairwise comparison with Elo conversion operationalizes Alexander's method and produces stable rankings across models.(claim)
Mirror and Deathbed tests measure orthogonal facets of quality; what feels most alive differs from what merits preservation.(claim)
Numeric scoring on aesthetics is measurably unreliable; inter-scorer agreement on 0–10 scale for taste is poor.(claim)
Quality without a name cannot be captured by checklist; it must be chosen through forced-choice comparison.(claim)
Roughness in responses decreases with parameter count within same-alignment model families, operationalizing the cost of polishing.(claim)
Smaller models produce more alive responses than larger ones in the same alignment family—roughness signals living process over manufactured polish.(claim)
The 15 Properties checklist is implementable and scales beyond Alexander's original small-N book studies.(claim)
The 15 structural properties decompose aliveness into inspectable formal components measurable on individual responses.(claim)
What you would preserve (Deathbed test) correlates with what you value and may serve as an alignment proxy.(claim)