Overview

Scarcity is a general online relationship-discovery engine — an automated statistician for multivariate data streams. You feed it one row at a time; it learns which relationships currently hold between the variables and hands that structure downstream as a continuously-updated knowledge graph. It is not specific to any one field; macroeconomics was simply the first domain it was applied to.

Around that engine sits a full runtime for learning under scarcity — few samples, high noise, constant drift: federation so nodes can learn together without sharing raw data, meta-learning so magnitudes can be borrowed across systems, and dynamic resource governance so the whole thing runs within a device's compute budget. "Scarcity" names both regimes — scarce data and scarce compute — and the framework is built to work under each.

Scarcity is also the framework from which Organizational Identity Theory emerged: the finding that a system's form — the structure this engine recovers — survives scarcity while its magnitudes do not.

Key Features

• Relationship-Discovery Engine A living population of 15 typed relational hypotheses (causal, correlational, functional, temporal, mediating, and more). Each must survive the data stream or die; survivors are calibrated against permutation nulls with false-discovery control, so noise is not reported as structure.

• Multi-Path Inference Engine (MPIE) Online bandit-based path exploration with UCB/Thompson sampling. Automatically finds the best calculation path.

• Federated Learning Decentralized model aggregation with differential privacy preservation. Learn from data without seeing it.

• Meta-Learning Cross-domain adaptation using online Reptile optimization. Transfer knowledge between different environments.

• Dynamic Resource Governance (DRG) Adaptive resource allocation based on system telemetry. If CPU usage spikes, the model shrinks.

• Real-time Simulation Agent-based modeling with 3D visualization to stress-test policies.

• Stream Processing Continuous data ingestion with backpressure control (PI-Controller).

• Event-Driven Architecture Asynchronous pub/sub communication fabric for non-blocking operations.

Version Information

• Version: 0.4.0-beta
• Author: Omega Makena
• License: Apache-2.0 (See LICENSE file)