Causal Consistency

Causal consistency requires that for any two events where one caused the other, a view contains both or neither — never an effect without its cause, or a cause without its visible consequence when the view claims simultaneity.

Causal mismatch is the cross-stream consistency hazard when independent writers or readers race: a payment visible before the order that triggered it, or inventory deducted before the sale event appears. Clock skew and processing lag do not fix this; event order and write authority do — single writer per invariant, append-only L0 streams, explicit supersession.

Relation to serializability and linearizability

Serializability and linearizability are infrastructure-level names for when causal (and stronger) criteria are enforced by the platform rather than assumed at stitch time. A dashboard that manually joins streams must still satisfy the causal criterion or disclose high epistemic uncertainty.

Corpus stance

A2 — working context: Causal cut across the uncertainty facet; complements temporal and probabilistic components. Not full causal consistency as in some DS memory models (weaker than linearizability, stronger than eventual) — same name, scoped here to the cause-before-effect visibility rule in Why Uncertainty Matters.