Reference · v2.0.0 · Reference Implementation

AetherOrchestrate

A Falsification-First Investigation of Decentralized Orchestration in the Aether Family

A structured investigation of whether a decentralized orchestration gap genuinely exists in the Aether family, or whether it is already covered by AetherFlow (integration), AetherGovern (authority), or AetherMesh (distribution). v2.0.0 adds a reference implementation (7 modules, 39 tests, 3 experiments) alongside the v0.1.0 hypothesis.

7
Python Modules
39
Tests Passing
3
Experiments
5
Failure Conditions

Installation

# From PyPI (published)
pip install aetherorchestrate

# From source (development)
git clone https://github.com/gitdeeper15/aetherorchestrate.git
cd aetherorchestrate
pip install -e ".[dev]"

# Run tests
pytest

# Run experiments
python experiments/decomposition_accuracy/run.py
python experiments/assignment_fairness/run.py
python experiments/coordination_window/run.py

# CLI
aetherorchestrate info
aetherorchestrate verify
Note: The PyPI package (2.0.0) is the reference implementation. It is not a production system. The v0.1.0 paper remains the authoritative hypothesis record.

Project

FieldValue
Full titleAetherOrchestrate: A Falsification-First Investigation of Decentralized Orchestration in the Aether Family
SubtitleA Falsification-First Investigation of Goal Decomposition, Task Assignment, Scheduling, Dependency Tracking, and Routing in Multi-Agent Software Ecosystems
Version2.0.0 (Reference Implementation) + 0.1.0 (Formal Architecture)
StageReference Implementation with Reproducible Experiments
LicenseMIT (code) · CC-BY-4.0 (docs, manuscript)
DomainDistributed Systems / Software Engineering / Multi-Agent Coordination
DOI (Zenodo)10.5281/zenodo.24095399
DOI (Figshare)10.6084/m9.figshare.34963687
OSF10.17605/OSF.IO/2F7TZ
PyPIaetherorchestrate 2.0.0 (published)
Record IDAETHERORCHESTRATE-02

Central Question

Is there a set of requirements for goal decomposition, task assignment, scheduling, dependency tracking, and routing in multi-agent software systems that (i) is not already specified by the six existing dimensions, (ii) does not reduce to verification or to governance, (iii) can be specified without reintroducing a central authority, and (iv) yields an observable benefit over existing mechanisms?

The Seven Functional Dimensions

#DimensionFocusQuestion
1AetherGitStorage and versioningHow are agent-generated changes stored, signed, and versioned?
2AetherFlowIntegration and post-merge verificationHow are changes integrated into a shared trunk and verified?
3AetherProofSpecification and evidenceHow are claims specified and how is evidence for them recorded?
4AetherMeshDistribution and consistencyHow is state synchronized without a central coordinator?
5AetherGovernAuthority and policyWho is authorized to do what, under which policy, and how is authority revoked?
6AetherRuntimeMonitoring and runtime assuranceWhat are agents doing while they run, and how are anomalies contained?
7AetherOrchestrate (hypothesis)OrchestrationWhat should agents do next, and how is work decomposed, assigned, scheduled, ordered, and routed?

Figure 1 — The Seven Functional Dimensions of the Aether Family

Figure 1 — The Seven Functional Dimensions of the Aether Family. The seventh dimension (AetherOrchestrate) is a hypothesis under investigation, not a confirmed dimension.

Failure Conditions F1–F5

IDClaim under testStatus
F1Non-redundancyNot falsified (R2, R4, R6 partly covered)
F2Orthogonality to verificationNot falsified (narrowly)
F3Boundary with governanceNot falsified (decentralized setting)
F4DecentralizabilityNot falsified (with residuals)
F5Observable benefitUntested
"Not falsified" is a weaker status than "confirmed". F5 (observable benefit) is untested. The paper does not promote "not falsified" to "confirmed".

Propositions

#PropositionStatus
P1Determinism of Local DecompositionProof sketch · implemented in code
P2Irreducible Coordination WindowProof sketch · implemented in code
Both propositions are stated with proof sketches; neither is machine-checked. They bear on F4 (decentralizability). Both are implemented in the reference implementation (v2.0.0) and covered by the test suite.

The Five Orchestrate Layers

LayerQuestionStatus
L1 — DecompositionHow is a goal broken into tasks?Implemented
L2 — AssignmentWhich agent executes each task?Implemented
L3 — SchedulingWhen does each task run?Implemented
L4 — Dependency TrackingWhat are the dependencies between tasks?Implemented
L5 — RoutingHow is each task routed to its executor?Implemented

Figure 2 — The Five Layers of AetherOrchestrate

Figure 2 — The five layers of AetherOrchestrate. Ordering reflects conceptual dependency, not an enforcement hierarchy.

Figure 3 — A Task Dependency Graph

Figure 3 — A goal decomposed into a DAG of tasks with hypothetical durations.

Figure 4 — The Assignment Challenge

Figure 4 — The assignment challenge in a decentralized setting: centralized (excluded) vs. decentralized (the challenge).

Figure 5 — Orchestration vs. Integration

Figure 5 — AetherOrchestrate acts before work runs; AetherFlow acts after work exists.

Figure 6 — The Centralization Spectrum

Figure 6 — The centralization spectrum of orchestration designs. The replicated scheduler is where AetherOrchestrate must operate (F4).

v2.0.0 Experimental Results

#ExperimentResult
1Decomposition Accuracy (M1)10 goals; mean task count = 1.00; all acyclic
2Assignment Fairness (M4)Jain index = 0.8803 (expected 0.9174; 4% deviation)
3Coordination Window (M8)5 scenarios; all positive; min 0.008 s, max 7.5 s
Important: These are simulation results. No generalization beyond simulation is claimed. Baselines B1–B5 and scenarios S1–S6 are not evaluated in v2.0.0. See the full evaluation report.
Status note (v2.0.0): This is a reference implementation (v2.0.0) of the falsification-first orchestration architecture specified in v0.1.0. 7 modules, 39 tests, 3 experiments. No real-world deployment has been evaluated. F1–F4 are not falsified by argument, with residual weaknesses. F5 (observable benefit) is untested. Propositions P1–P2 remain proof sketches (not machine-checked), though they are implemented in code. The final project structure (single, dual, triple, or quadruple) is undetermined. The paper is not peer reviewed.