D
Discovery
Asks
Is this worth investigating further?
Noise is accepted. Small bets. Capital exposure minimal. Most cases here do not proceed — that is the design.
§ 1 · Ontology
What Indicio
is, and what
it is not.
Ontology defines the categories that are allowed to exist inside the system. Anything that does not map to one of these entities is out of scope, by construction.
INDX / § 1
6 ENTITIES · 4 STATES
6 ENTITIES · 4 STATES
Position
§ 1.1
A truth-discovery institution.
Indicio is infrastructure for systematic truth-discovery under uncertainty. It is defined by its ability to reduce time-to-truth and cost-to-truth, repeatedly, with compounding improvement — not by any specific brand, product, or category.
Is
- An operating system for repeatable case creation.
- A machine whose primary output is validated decisions.
- A ledger of attributes, thresholds, and outcomes.
- Category-agnostic by design; DTC is the current execution layer, not the definition.
Is not
- A portfolio of brands.
- A venture studio or holding company.
- A DTC operator.
- A vehicle for founder intuition dressed as strategy.
Entities
§ 1.2
Six things allowed to exist.
Every object flowing through the system is one of these. Anything unclassifiable is a bug in the ontology or a signal that a new entity must be ratified before it may operate.
Reads as event, not idea
Input
An observable change in the external environment — a market event, a cultural shift, a platform change, a creator dynamic, a demand anomaly. Ideation is not an input; environmental change is.
Reads as bounded execution
Case
A time-boxed execution instance carrying its own attributes, thresholds, budget, and state. Cases are not brands. A brand may emerge from a case; the case is upstream of the brand.
Reads as falsifiable claim
Attribute
A specific, measurable claim about how the case will create value. Attributes that cannot be made measurable are narratives, not attributes, and may not govern capital allocation.
Reads as pre-committed line
Threshold
The binary line an attribute must cross to be considered validated. Thresholds are set before the test runs and may not be revised mid-test without documented new information.
Reads as controlled mutation
Branch
A proven success (external or internal) with one or two variables mutated and the rest held constant. Internal branches import validated infrastructure and are systematically more valuable than external branches.
Reads as rule change
Learning
An observation that has been translated into a changed rule, a tightened constraint, a new default, or a removed option. If it does not update the system, it is not a learning — it is a story about one.
States
§ 1.3
Four states. No hybrids.
Every case is always in exactly one state. State constrains which actions are legal, which metrics are meaningful, and how the same signal must be interpreted.
V
Validation
Asks
Do the attributes actually hold in reality?
Hypotheses are made explicit with binary thresholds. Capital is deployed against specific tests, not the case as a whole. Time-boxed.
E
Exploitation
Asks
How fast can we scale what is already validated?
Attribute-testing is largely complete. Capital and attention deploy asymmetrically. Experimentation continues only insofar as it supports scaling.
K/R
Kill · Release
Asks
Recover resources or preserve optionality?
Not failure states. Successful outputs of disciplined decision-making. Kill terminates; Release removes from the core system while keeping upside.
§ 1.4 · Flow
The system reduces to a small diagram.
01
Input
Environmental change
02
Filter
Hard constraints
03
Shape
Attributes + thresholds
04
State
Discovery → Validation → Exploitation
05
Output
Go · Scale · Kill · Release
Anything that does not clearly map to one of the four outputs — Go, Scale, Kill, Release — is not a valid system action.
Continue reading