A few quick answers to common questions about Pyragogy and this community. What is Pyragogy? Pyragogy is an exploration of how learning changes when humans and AI think together. It builds on the idea of Peeragogy, a framework where people learn from each other as peers rather than from a central authority. Pyragogy asks a new question: What happens when some of those peers are AI systems? The goal is not to replace human learning, but to explore a new form of collaboration between different kinds of minds. Is Pyragogy a formal theory? Not yet. Pyragogy is an open experiment. Ideas are tested through conversations, projects, and experiments shared by the community. Think of it as a living framework, not a finished doctrine. Do I need technical knowledge to participate? No. Some discussions involve AI tools or experiments, but many conversations are about: • learning • collaboration • creativity • knowledge sharing Curiosity is more important than expertise. Is Pyragogy about AI replacing teachers? No. Pyragogy is not about replacing teachers or experts. It explores how learning ecosystems change when AI becomes a participant in the process, alongside humans. Human communities remain central. Who started Pyragogy? Pyragogy was initiated by members of the Peeragogy community and independent researchers exploring new forms of learning in the AI age. This forum is one of the spaces where the idea is being explored and developed. What can I do here? You can: • introduce yourself • ask questions • share experiments with AI • discuss learning methods • collaborate on ideas and projects The forum works best when people contribute their own experiences and reflections. Is Pyragogy connected to the Peeragogy Handbook? Yes. Pyragogy grows out of the ideas and practices developed in the Peeragogy Handbook, which explores peer-to-peer learning communities. Pyragogy extends that exploration into the AI era. Can I challenge the ideas here? Absolutely. Disagreement and critical thinking are welcome. Pyragogy is not a belief system — it is a collective exploration. Where should I start? If you’re new here: Introduce yourself in the introduction thread Browse the Agora discussions Share a question or idea Small contributions often lead to the most interesting conversations.

I want to put a pattern on the bench, not announce a finished thing. It’s written up as a preprint — Oblique Peer Review: Extending Pyragogy Design Patterns through Structural Isolation and the Limits of Blind Convergence (Zenodo, DOI 10.5281/zenodo.20544658, CC BY 4.0) — but the paper is the long version. This post is the short version, and it’s here because Pattern Workshops is where things get tested, extended, or refuted. I’d rather it got refuted here than admired. The pattern, in one breath 1 + N. One human orchestrator, N AI agents from different vendors, and one rule: the agents never read each other. Information moves only through the human, who strips it of its source and its argumentative framing before passing it on. No agent knows whose reasoning it’s building on, or whether it’s the first voice or the fourth. The point isn’t to make the agents disagree. It’s to keep their analytical axes independent, so they cross the problem at different angles instead of collapsing into one voice. The friction is geometric — a property of how the trajectories intersect — not hostility between critics. I called it oblique for that reason: not parallel (redundant), not frontally opposed (sterile), but transversal. It’s the Peeragogy move applied to synthetic peers: when the peer has no stakes and no accountability, the epistemic value can’t live in the agent’s judgment. It has to be engineered into the architecture that arranges the agents. Where it breaks — and why that’s the actual contribution Here’s the part I’d defend least confidently, which is exactly why it’s the part worth workshopping. The pattern is built on the premise that independent, mutually-blind agents supply friction a single model can’t. That premise does real work — but it has a hard limit, and the paper’s central finding is that limit: Blindness removes imitation, not error. Independent agents can converge on the same wrong answer, for the same reason independent instruments can share a systematic bias — not because they talk to each other, but because they’re exposed to the same salient feature and respond to it the same way. When that happens, their agreement feels like corroboration and is, from the inside, indistinguishable from the real thing. I caught this the hard way. Three independent models — different vendors — converged on a fix that looked perfect and was endorsed by all of them at once. It was wrong. It would have made the system blind to short-form defamation. Three converging AIs didn’t catch it; a check run against recorded data, outside the loop of agents, did — and only because the check was aimed deliberately at the cases the fix might break, not the cases it was built to fix. So the stopping rule can’t be agreement. Convergence among the agents is a quality signal only after a check outside the loop, never in place of one. The two ways the human fails Both failures land on the same node — the one in the 1+N: Cognitive Impedance Mismatch. The agents generate faster and denser than the human can integrate. Past a threshold the operator keeps steering but stops absorbing — delegated comprehension — and the loop has quietly become the automation it was meant to prevent. A limit of bandwidth. False convergence. The one just described — the shared error that reads as proof, and lulls the operator into trusting it. A limit of discernment. One overwhelms the human; the other reassures him. A real operator can hit either without noticing. What I’m bringing to the bench This is n=1 — one practitioner, one body of work, observed from the inside, by the same person who designed the pattern. That’s a genuine limit, and I’d rather state it than have it pointed out. The reflexive circularity is real too: the paper was itself produced through the pattern it describes. So the honest questions are the ones I can’t answer alone: Does it survive other hands? The pattern needs a knowledgeable human at the centre — the operator’s domain knowledge repeatedly supplied the ground truth the agents lacked. Run by someone other than its designer, on problems outside software, does it still do anything? Can the false-convergence check be generalized? I could only build a hand-made, case-specific discriminator — did the independent voices touch distinct features of the problem, or pile onto the same one? Does that distinction hold beyond the single case? Can it be detected without a human reading the underlying material? Is “delegated comprehension” detectable before it’s too late? The CIM is named here as an observed boundary, not measured. What would it look like to instrument the operator’s integration load instead of relying on his own report of when he started slipping? If you’ve run something like 1+N — even informally, even just bouncing one decision between two models — I want to hear where it held and where it didn’t. Especially where it didn’t. The paper is the formal carry. This is the invitation to break it.

Status: Completed — first audit cycle Date started: 2026-08 Date completed: 2026-09 Participants: 1 human researcher + 6 AI agents Corpus: Peeragogy Handbook — 88 nodes [image: 1788676643608-img.jpg] Hypothesis What happens when the patterns of the Peeragogy Handbook are systematically examined against documented evidence and real-world practice? The goal was not to prove that Peeragogy is wrong. The goal was to find where its theoretical patterns remain well supported, where evidence introduces tension, where claims are underspecified, and where important questions remain unanswered. The experiment started from a simple principle: The model may propose. The evidence must dispose. Method The original experiment was designed as a human research laboratory. Practitioners were invited to contribute critical incidents from their own experience: situations in which peer-learning theory met real conditions, including friction, failure, contradiction, or unexpected outcomes. The laboratory remained open. Nobody submitted a critical incident. Rather than treating that silence as the end of the experiment, I changed the experimental conditions. I built a machine-mediated research branch. The resulting pipeline uses six AI agents, coordinated across five analytical stages, with different responsibilities: Theory Interpretation — identifies the claims and assumptions expressed by each pattern. Evidence Hunting — searches for relevant empirical, documented, or research-based evidence. Grounding Gate — verifies whether proposed evidence actually supports the claim being examined. Critical Analysis — evaluates support, contradiction, ambiguity, and limitations. Tension Analysis & Documentation — compares interpretations, records tensions, and produces the audit trail. The system does not treat an LLM’s internal knowledge as evidence. Evidence must be grounded in identifiable sources. What Happened The complete Peeragogy Handbook corpus was processed across 88 nodes. Each node was examined against external evidence and analysed through the multi-agent pipeline. The resulting audit produced a structured assessment rather than a binary classification of true or false. The audit records different relationships between theoretical claims and available evidence: Support Tension Contradiction Ambiguity Insufficient evidence This distinction matters. A lack of contradictory evidence is not evidence that a theoretical claim is true. Likewise, evidence of tension does not automatically make a pattern false. It may indicate that a pattern works under some conditions but not others, that its assumptions are underspecified, or that the available evidence is insufficient to establish a stronger conclusion. What Broke The first experimental condition broke before the research itself really began. The human participation model produced no field reports. The laboratory was open, but nobody walked through the door. That forced a methodological change. Instead of pretending that human participation had occurred, I recorded the absence of participation as part of the experiment and built a second branch using AI-mediated systematic analysis. This changed the nature of the experiment. The machine branch is not equivalent to human participation. AI agents do not have lived experience, social stakes, or personal involvement in the situations described by practitioners. What they can provide is something different: systematic comparison, source retrieval, structured critique, and repeated examination of a theoretical corpus. That distinction remains central to the interpretation of the results. Surprises The most important surprise was that the failed participation model did not end the experiment. It changed it. The absence of human contributions became the condition that led to the construction of the multi-agent pipeline. The second surprise was how different the resulting activity felt from ordinary AI-assisted writing. The agents were not used simply to generate a report. They were assigned different functions and asked to challenge one another’s outputs. One agent could propose an interpretation. Another could search for evidence. Another could reject weak grounding. Another could identify tensions. The orchestrator then had to reconcile the resulting outputs. The result was less like asking an AI for an answer and more like constructing a small adversarial research process. Results Corpus 88 — Peeragogy Handbook nodes audited Evidence 304 — evidence items identified 52 — entities represented in the evidence 50 — distinct source URLs Analysis 6 — AI agents 5 — analytical stages 1 — orchestrated research pipeline The full quantitative and qualitative results are documented in the audit report. What Changed Mid-Experiment The initial hypothesis assumed that the most valuable evidence would come from practitioners contributing critical incidents. That condition did not materialise. So the experiment changed from: Theory → Human Critical Incidents → Analysis to: Theory → External Evidence → Multi-Agent Analysis This was not treated as a correction to the data. It was treated as a change in experimental conditions. That distinction is important because the machine-mediated branch answers a different question from the original human laboratory. The first asks: What happens when practitioners report what happened to them? The second asks: What happens when a theoretical corpus is systematically interrogated against available evidence? UnPeeragogy became the name for the second experiment. Analysis The audit should not be read as a verdict on Peeragogy. Its purpose is to make theoretical claims more exposed to evidence, criticism, and revision. 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 that the corpus does not adequately describe. The most useful result is therefore not a ranking of “good” and “bad” patterns. It is a map of where the relationship between theory and evidence becomes interesting. The central research loop is: Theory → Evidence → Interpretation → Counter-evidence → Revision The audit is not a conclusion. It is a starting point. What This Experiment Changed I started with a question about Peeragogy. I ended up building a reusable research protocol for moving between theoretical claims and external evidence. The experiment also changed my understanding of participation. I originally thought the laboratory required people to enter the room before anything meaningful could happen. It turned out that the laboratory could also be a set of conditions, tools, protocols, and questions that remain active even when nobody enters. That does not make the human laboratory unnecessary. It makes the distinction between human experience and machine-mediated analysis more explicit. The door remains open. Human field reports are still the evidence that can bring lived experience back into the system. Artifacts Quick Links Resource Link Dedicated Audit Page unpeeragogy.pyragogy.org/audit Full Report (Zenodo) doi.org/10.5281/zenodo.22309102 Pipeline Repository github.com/pyragogy/UnPeeragogy UnPeeragogy Protocol unpeeragogy.pyragogy.org/protocol Background Blog — Part I blog.pyragogy.org/posts/2dd675a5 Background Blog — Part II blog.pyragogy.org/posts/e5f507dd Peeragogy Handbook peeragogy.org License CC Zero — Public Domain Get Involved 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, or another collaborative environment — your field reports are the evidence that makes the audit meaningful over time. The machine-mediated audit can systematically examine a corpus. It cannot replace lived experience. That is where the next experimental cycle begins. Submit a field report via the UnPeeragogy repository Discuss the protocol on the Pyragogy Forum Fork the pipeline and run it on your own corpus Challenge the results by providing counter-evidence or alternative interpretations Next Steps The first audit cycle is complete. The laboratory is not. The next step is to bring external practitioners back into the loop. Field reports, counter-evidence, alternative interpretations, and independent runs of the pipeline can all become inputs for subsequent iterations. The experiment therefore remains open to revision. The audit is not a conclusion. It is a starting point. And perhaps that is the most useful result of the experiment so far: The first experimental condition failed to produce participation. The experiment did not stop. The conditions changed.

Contributing to the Handbook The Pyragogy Handbook is community property. The process for contributing should be accessible to anyone willing to engage seriously. The Handbook Structure The handbook lives in a GitHub repository (confirm URL with @Fabry — link pending final setup). It’s organized into: Foundations — Core concepts and Cognitive Rhythm framework Patterns — Validated patterns in formal template format Practices — How-to guides and process documentation Stories — Case studies and experiment records Resources — Annotated bibliography and tool references Three Ways to Contribute Path 1: Forum-First (Recommended for New Contributors) Post your contribution in the appropriate Archive subcategory Let the community discuss and refine it When there’s rough consensus, tag a maintainer Maintainer creates the GitHub PR or helps you create one Best for: Pattern contributions, new sections, anything where community input helps. Path 2: Direct GitHub PR Fork the repository Create a branch: contrib/[your-handle]-[short-description] Make your changes following the style guide Submit a PR with clear description of what you changed and why Request review from at least one maintainer Best for: Corrections, small improvements, people comfortable with Git. Path 3: Suggest, Don’t Write Post in Handbook Contributions with [PROPOSAL] in the title. Describe what you think should be added and why. Content Standards What we’re looking for: Tested claims (not “AI can do X” — “we tried X and here’s what happened”) Clear examples (not just abstract descriptions) Acknowledged uncertainty (don’t claim more than you know) Disclosed AI assistance What we’re not looking for: Claims that haven’t been tested in practice Content that could have been written without engaging with Pyragogy specifically Attribution Contributors are credited in the handbook’s contributor file. AI assistance is noted with the human author credited as primary. This is your work. The handbook is better because you contributed. That matters. Human-AI Co-Creation