Evidence Cache
Evidence Cache is the Institute’s intelligence and signal-discipline surface. It produces Signal Cards, Evidence Notes, claim-status classifications, and admissibility observations for operators who need to act without being captured by speed, narrative, or synthetic confidence. Each output is a Layer A instrument with an explicit verification gate — an observable claim that can be confirmed, contested, or revised against the live record.
What Evidence Cache is
A structured pre-decision evidence layer. Evidence Cache classifies claims, signals, model outputs, market indicators, institutional commitments, AI-era announcements, and civilizational coordination events before they become inputs to decisions. The questions it asks are not descriptive. What is the evidence status of this claim. What is missing. What is being inferred rather than established. What has not yet earned the right to be acted upon. The questions are gate-level, not commentary.
What Evidence Cache is not
Evidence Cache is not a prediction market, trading bot, fact-checking blog, legal compliance service, news aggregator, or oracle. It does not tell the reader what will happen. It tells the reader what the evidence status of a claim or signal is at the moment they are considering acting on it. The function is to slow premature commitment, not to accelerate prediction.
The Signal Card
Each Signal Card produced by Evidence Cache contains a fixed structure. The claim or signal under review is stated in one sentence using only public-domain terms. The current admissibility status is assigned — admitted, conditional, inadmissible, contested, unresolved. The evidence available is summarized in plain reference to public sources. The evidence missing is stated explicitly. The inference gap is named — what the claim depends on that has not been demonstrated. The admissibility note offers the Institute’s Layer C reading of the signal. The card closes with a verification gate — what observable would change the assigned status.
How to read a Signal Card
A Signal Card is not advice and it is not a forecast. It is a structured representation of the current admissibility position of a public claim or signal. Reading it correctly means treating it as a snapshot of evidence topology, not as a recommendation to act. The most important field is often the evidence-missing field — what would need to exist for the claim to cross the threshold into admitted status. Operators who use Signal Cards regularly develop a working sensitivity to the difference between strong narrative pressure and strong evidence position. The two are commonly inverted in the AI execution era.
Coverage
Evidence Cache covers AI and ASI development announcements, model releases and capability claims, agent and multi-agent deployments, compute and energy infrastructure decisions, AI governance and regulatory developments, institutional commitments tied to AI-era infrastructure, prediction market movements relevant to admissibility, civilizational coordination signals such as treaties, protocols, and standards, and selected synthetic evidence events. Coverage is not comprehensive. It is selective by admissibility relevance.
Frequently asked
What is a Signal Card. A structured pre-decision classification of a claim or signal. It identifies admissibility status, evidence available, evidence missing, inference gap, and the verification gate through which the assigned status can be revised.
How is Evidence Cache different from news analysis. News analysis describes what happened. Evidence Cache classifies whether what is being claimed has sufficient evidentiary status to enter decision-making. The questions are not the same.
Is Evidence Cache free. The base Signal Card archive is public. Operator Brief — the recurring intelligence product built on Evidence Cache — is available by subscription.
