Accounts for a bounded promotion-risk budget before opt-in statistical promotion.
Contracted statistical risk
in GenOS.
Accounts for a bounded promotion-risk budget before opt-in statistical promotion.
What it means
in GenOS.
Statistical decision procedures spend a declared error budget across repeated tests. A ledger can bound promotion risk only if assumptions, dependence and actual decision rules are explicit.
Pre-register → Spend and inherit risk → Gate promotion
For total risk allowance α, allocate α_i≥0 with Σ_i α_i≤α. A promotion consumes its declared allocation only after its test and evidence conditions pass.
Explore the mechanism: Contracted statistical risk
A qualitative teaching model of this concept. It does not run GenOS or validate its implementation.
The modeled transition may proceed. Change the assumption and move through the stages to inspect the decision path.
How it works
at a glance.
Pre-register
Declare the promotion family, binary criterion, null hypothesis and total risk budget.
Spend and inherit risk
Reserve a summable allocation before evaluation. Bind ordered source units to the immutable protocol; forks, merges and topology changes cannot refund spent risk.
Gate promotion
Refuse an opt-in statistical promotion without an admissible receipt; other evidence gates still apply.
What is connected,
and how far.
Protected graph nodes retain their risk scope across resume and composition. Gates reread provenance as well as the signed receipt. The union-bound guarantee requires valid individual tests; the sequential model assumes conditional win probability at most one half under its null. These gates do not replace deterministic checks, safety or dissent review, and no general field benefit is established.
A transactional risk lineage, preregistered ordered evaluation units, provenance checks and signed receipts guard opt-in topology promotions. Statistical assumptions and field qualification remain separate obligations.
Failure modes have not yet been separately cataloged for this entry; follow the source and evidence links before relying on the mechanism.
The mechanism
behind the analogy.
The design of experiments
Ronald A. Fisher · Oliver and Boyd
Original experimental-design treatment; GenOS comparisons need matched inputs and a declared measurement protocol.
Open publication ↗Primary research describes the cited mechanism; specific technologies may instead cite their original technical documentation. These sources provide context and do not validate the GenOS implementation.
Statistical admission is only one promotion gate
Protected graph nodes retain their risk scope across resume and composition. Gates reread provenance as well as the signed receipt. The union-bound guarantee requires valid individual tests; the sequential model assumes conditional win probability at most one half under its null. These gates do not replace deterministic checks, safety or dissent review, and no general field benefit is established.