idkmesh

Findings Index

This directory holds research and engineering findings: analyses, landscape studies, evidence reviews, and external source notes. A finding records what was observed and what it implied at its stated date. It is not a canonical contract, and it does not by itself authorize an implementation.

Canonical current authority lives in ../decisions/, ../specifications/README.md, ../architecture/README.md, and ../../PROJECT_RULES.md. Where a finding and a canonical artifact disagree, the canonical artifact wins.

The groups below separate findings that still state a working project thesis from findings that are retained as historical or source material, because the directory listing alone does not distinguish them.

Working theses and program framing

These findings still express positions the project actively builds on, and are referenced by open research issues.

• Distributed Agent Coding and Collective Software Engineering — the many-weaker-agents hypothesis, its failure modes, and the quality pipeline it requires. Cited by issue #2.
• Emergence from Vague Goals — nature-inspired argument that variation plus constraints, selection, memory, and verification, not vagueness alone, can produce coherent systems. Cited by issue #22; measured against baselines in experiment E024.
• Current Agent Ecosystem and the IDKMesh Evolution Wedge — external-ecosystem review concluding that IDKMesh should integrate rather than recreate generic agent infrastructure.
• Open-Source Community, Collaboration, and Platform Strategy — how an open contribution surface coexists with a tightly verified canonical product.

Repository-health and navigation evidence

Bounded reviews produced under issue #152. Each records a source revision, a measured observatory delta, and the reviewer judgement behind it. None of them treats a lower warning count as a standalone objective.

• IDKGraph Orphan-Warning Triage — cohort 001 — the first sampled, seeded classification of orphan candidates.
• IDKGraph Architecture Navigation Pass — indexing docs/architecture/ by declared document status.
• Conversation Index Drift Review — the archive index had fallen behind the archive; repair and measured effect.
• Conversation Index Regression — the follow-on drift that motivated the deterministic index guard test.
• IDKGraph Findings Navigation Pass — this directory’s own bounded pass: classification of all 15 findings and the measured effect of adding this index.
• First Production Collaboration-Observables Snapshot
• Ownership Concentration: First Real Measurement
• Five Committed Executables That Nothing Demonstrates Ever Ran
• Replaying the Node Evidence: One Digest Reproduces, Four Cannot — the only production run the observables pipeline has produced, committed before its 30-day artifact retention expired; which of its six zero-valued observables are real measurements and which are declared collector limitations.

Growth, discovery, and free-compute landscape

• Fast Growth and Free Compute Audit — diagnosis that the project was external-attention starved rather than compute starved.
• Repository-Driven Community Growth — the contribution flywheel and the R_c community-reproduction metric. Cited by issue #10.
• Free / Low-Cost Agent Network Landscape — the zero-cost hosted and volunteer-local compute options available to the project.

Historical records and source notes

Retained as project memory and evidence references. They are not current guidance, and the source notes do not imply that the cited authors endorse IDKMesh.

• Initial Landscape Findings — naming decision, the NovaFabric distinction, and the first “scale does not imply quality” principle.
• Reference Map and Lessons from Bitcoin/Crypto — long reference map; reuse the distributed-systems ideas before the monetary layer.
• Issue Resolution and Work Unit Track Closure — record of the pass that converged the Work Unit protocol track.
• Science and Blockchain Source Notes — external citations behind the statistical-mechanics, gossip, and percolation arguments.

Adding a finding

Give the file a dated name, state the date and scope in the document, and add one line to the group above that it belongs to. If a finding later becomes binding, promote its conclusion into a decision, specification, or architecture document rather than upgrading the finding’s authority in place.

This site is open source. Improve this page.