01
Attribute
A specific, measurable claim. Attributes that cannot be made measurable are narratives, not attributes. They may not govern capital allocation.
§ 7–8 · Epistemology
How we
know what
we know.
Capital exists to buy truth. Truth requires a discipline about what counts as evidence, what counts as merely observation, and what counts as narrative masquerading as either.
INDX / § 7–8
BINARY · ANOMALY-FIRST
BINARY · ANOMALY-FIRST
Method of knowing
§ 7
Attribute, threshold, binary evaluation.
An attribute is a falsifiable claim about how the case will create value. A threshold is the binary line the attribute must cross. Evaluation returns exactly one of two outcomes.
02
Threshold
Set before testing begins. Cannot be revised mid-test without documented new information. A retroactively-declared 'approximately passed' is rationalization, not a test.
03
Binary
Threshold met or not met. Ambiguous outcomes are treated as not-met by default. Interpretation is not evaluation.
Attribute ledger · sample
A test either passes or it doesn't.
A01
Pre-order conversion reaches at least 70% of inventory-based e-commerce benchmark.
Threshold
≥ 70%
Result
Met
A02
Creators in the 200–500k segment accept a 90/10 commercial brief.
Threshold
≥ 60% acceptance
Result
Met
A03
AOV increases by at least 30% over three drops for returning customers.
Threshold
Δ AOV ≥ 30%
Result
Not met
A04
Median time-from-signal-to-decision under 72 hours across ten consecutive cases.
Threshold
≤ 72h · N=10
Result
Watch
§ 8.5 · Anomaly
Most data establishes baselines. Decision energy is reserved for deviations.
The system spends less time asking is this good and more time asking why did this behave differently than expected, is it repeatable, and can it be amplified. Averages hide opportunity; anomalies produce it.
- · Unexpected conversion rates
- · Creators outperforming peers by large margins
- · Demand appearing earlier or stronger than modelled
- · Unit economics improving non-linearly with scale
FIG. 08 — DEVIATION FROM BASELINEROLLING 12 CASES
— Baseline◯ Anomaly · flagged for amplification or kill
§ 8 · Feedback loops
If a lesson cannot be translated into a rule change, it is not a learning. It is an observation.
Case-specific
Applies only to the case. Useful locally. No cross-case relevance.
Category-bound
Applies to a category of cases — cross-case, within the same asset class.
General
Applies across asset classes. Candidate for cross-asset-class principle.
When category-bound or general learnings exceed pattern thresholds, they become candidate principles. Candidates are promoted into the system, demoted to noise, or retained as watch items. That is the mechanism by which the system revises itself.
Continue reading