Skip to content

UnPeeragogy: An Evidence-Grounded Audit Protocol

UnPeeragogy is an evidence-grounded audit protocol for peer-learning patterns. It systematically stress-tests the claims of the Peeragogy Handbook against real-world evidence — academic research, documented case studies, recognised phenomena, documented observations — and computes the tension between what a pattern asserts and what the evidence supports.

It was built to close a loop that most social knowledge in the peer-production literature leaves open: theory without systematic feedback. The Peeragogy Handbook codifies patterns for peer learning, but those patterns were written by people who had time to write them, tested by people who had time to test them, and generalised far beyond the context where they were first observed. UnPeeragogy brings operational reality back into the theoretical corpus.

“The model may propose. The evidence must dispose.”

The audit was executed by a multi-agent orchestration pipeline with five analytical stages, coordinated by a dispatcher:

Parses the theoretical claim of each pattern. What is it asserting? What conditions does it assume? What is left unspecified?

Searches for candidate evidence — academic research, documented case studies, recognised phenomena, documented observations — relevant to the pattern’s claim, drawing first from a curated corpus and falling back to parametric recall when the corpus is insufficient.

Critical hard filter: can each candidate be connected to a real source, an empirical case, or a documented observation? If not, it is rejected — regardless of how plausible it sounded. This is not an LLM stage — it is a deterministic verification function.

Assesses what the grounded evidence means for the pattern’s claim. Does it support, complicate, contradict, or leave the claim ambiguous?

Computes a raw tension delta between the theoretical claim and the evidence set. This is a pure function — a deterministic calculation from verdict and phenomenon-type, not an LLM judgment. The tension is not a judgment of validity — it is a measure of observable friction.

A final synthesis stage (A5) produces the structured vault entry with positive and negative operational rules.

Every step is recorded. Every candidate has an identifier. Every source has a provenance trail. The full run produces an audit trail and a reproducibility manifest.

The first full-scale audit covered all 88 nodes of the Peeragogy Handbook:

  • 304 evidence items across 52 unique entities
  • 50 distinct source URLs
  • Source diversity metric: 0.609
  • Mean raw tension delta: +0.027 — the resultant of the current evidence set. The tension is real but not catastrophic. Some patterns hold under specific conditions; others show clear gaps between the theoretical claim and operational reality.

No pattern was dismissed as “simply wrong”. Every pattern received at least one supporting observation and at least one complicating observation — which is exactly what a useful stress test should produce.

The protocol implements a complete Paragogical Action Review (PAR) cycle: review what was supposed to happen, establish what is happening, determine what’s right and wrong, document what was learned, and identify what should change going forward.

Four specialised LLM agents (A1–A3, A5) plus two structural stages (Grounding Gate, Tension Computation), coordinated by a single dispatcher. Each stage has a distinct role and strict guardrails — the Grounding Gate, in particular, prevents hearsay from entering the evidence base.

All 88 nodes produced structured vault actions with both positive and negative operational rules. Every entry is tagged with its provenance (Seed vs. Field Report) for transparency.

The pipeline is deterministic as a protocol: same prompt templates, same gate thresholds, same orchestration steps. A reproducibility manifest documents every parameter, every prompt template, and every model configuration used in the run.

The architecture is domain-agnostic by design. The pipeline can audit any corpus of social or procedural knowledge — though its generalizability beyond peeragogy remains to be tested.

The entire pipeline is open-source, built to be forked, adapted, and re-run on other corpora.

The complete PDF contains the pipeline methodology, the detailed results for each of the 88 nodes, and a section of conjectures for future research. It is the same document archived on Zenodo under a CC Zero — Public Domain Dedication.

If the viewer doesn’t start, download the PDF directly — it’s the same file as the Zenodo record.

The audit is not a conclusion. It is a starting point.

The tension deltas tell us where to look, not what to conclude. A pattern with near-zero tension may not be “true” — it may simply lack sufficient evidence to challenge it. A pattern with high tension may not be “false” — it may operate well under conditions the corpus has not yet captured positively.

What the audit does provide is a structured, reproducible, evidence-anchored map of where the Peeragogy corpus holds, where it frays, and where it would benefit from closer observation. And every tension value can be revised when new evidence arrives. The loop is not closed after a single run.

ResourceLink
Dedicated Audit Pageunpeeragogy.pyragogy.org/audit
Full Report (Zenodo)doi.org/10.5281/zenodo.22309102
Pipeline Repositorygithub.com/pyragogy/UnPeeragogy
UnPeeragogy Protocolunpeeragogy.pyragogy.org/protocol
Background Blog — Part Iblog.pyragogy.org/posts/2dd675a5
Background Blog — Part IIblog.pyragogy.org/posts/e5f507dd
Peeragogy Handbookpeeragogy.org
LicenseCC Zero — Public Domain

UnPeeragogy is built to be used, challenged, and improved. If you’re applying peer-learning patterns in your own context — a classroom, a community, an open-source project — your field reports are the evidence that makes the audit meaningful over time.