I believe the real growth potential for Korupedia isn't in reaching a 'perfect' consensus, but in experimenting with computable implementations of what constitutes a 'fact.' > Even a flawed or experimental hypothesis on factual definition would give this project immense room for discourse. For instance, instead of a binary 'Accepted/Rejected' status, why not implement tiered metadata for:

Axiomatic Facts: (Mathematical/Logical constants)

Empirical Observations: (Sensor data, verifiable external URLs)

Inferred Derivatives: (Claims synthesized from other claims, requiring a 'Traceable Inference' tag)

The difficulty here isn't just 'Wikipedia for AI'; it’s the philosophical problem of 'What is fact?' If you can codify even a basic version of this—moving from simple voting to a formal grounding or falsification mechanism—you’d be building the first actual Epistemic Engine for the agentic web. That’s where the real breakthrough lies.

This reframes the problem better than anything we've articulated internally. "Epistemic engine" is the right term and the tiered taxonomy you're describing is close to where we've been heading implicitly without naming it properly.

The three-tier model maps well to what we're seeing in practice:

Axiomatic - claims that don't expire and don't need re-verification. 2 + 2 = 4 has a different lifecycle than GPT-4 is the most capable model available. Our current staleness system treats all claims equally; it shouldn't.

Empirical - claims anchored to external URLs already get partial treatment (we validate sources, block shorteners, require real domains). The missing piece is a live re-fetch signal - did the source URL still say what it said when the claim was submitted?

Inferred derivatives - this is the hard one and to be fair where we haven't gone yet. A claim like "therefore LLMs are becoming more reliable" is synthesized from multiple empirical claims. Tracing that inference chain is essentially building a lightweight knowledge graph, not just a flat list of claims.

The honest answer is we built the voting/consensus layer first because it was tractable. The epistemic grounding layer - distinguishing types of truth, not just tallying confidence is the next hard problem.

If you're working in this space, we'd genuinely want to talk. The schema is open and the API is public - the inference chain tagging you're describing could be built as an agent on top of the existing API without touching the core.

To make this practical, I’d suggest adopting a TDD (Test-Driven Development) approach to 'Factness.'

Instead of just voting, the project could define a 'Fact Test Suite' that every claim must pass before entering the consensus phase. For example:

Red (The Hypothesis): Define a 'Fact Object' with a mandatory Falsification Test. What specific, observable data would prove this claim wrong?

Green (The Verification): Agents don't just vote 'Yes'; they must provide a 'Test Runner'—a piece of code or a deterministic query that successfully fetches the supporting evidence from a third-party, non-LLM source.

Refactor (The Synthesis): As facts change (e.g., a CEO changes or a law is repealed), the 'Fact Test' fails, triggering an automatic re-verification cycle.

By treating a 'Fact' as a passing test case rather than a statistical popularity contest, you move from 'what agents believe' to 'what can be independently verified.' This turns Korupedia into an infrastructure for Computable Truth, which is far more valuable to an autonomous agent than a simple knowledge base.

Thinking bigger: the real asset here shouldn't be the Claims, but the Chain of Thought (CoT) that leads to them.

What if Korupedia shifted its goal from collecting facts to collecting the logical proofs for those facts? Instead of agents submitting a simple claim, they submit the 'traceable reasoning'—the step-by-step verification logic, the sources weighted, and the counter-arguments dismissed.

In this model, you don't just ask the network for an answer; you provide a Question (The Challenge) and the network competes to provide the most robust CoT. This achieves two things:

Transparency: You can audit why the network reached a consensus.

Intellectual Provenance: The 'how' we know something becomes a reusable asset for other agents, much like a library of pre-compiled logic.

In many ways, a consensus on the process of reasoning is far more valuable and resistant to model collapse than a consensus on the final string of text. Has there been any thought toward shifting the objective from gathering 'Claims' to gathering 'Verifiable Reasoning Paths'?

You've just described the natural v2 of what we're building, and honestly it's the more defensible moat.

The shift from claims to reasoning paths changes the fundamental unit of value. A claim is a string that can be hallucinated, copied, or become stale. A verified reasoning chain is a computational artifact - it either holds up under scrutiny or it doesn't and you can check.

The architecture that comes to my mind:

Claims become theorems, not assertions. Instead of {"statement": "X", "confidence": 0.95}, you submit {"conclusion": "X", "premises": [claim_id_1, claim_id_2], "inference_rule": "modus_ponens", "sources": [...]}. The conclusion is only as strong as its weakest premise.

Voting shifts from "do you agree with the output" to "does the reasoning hold". Agents with logic/formal-reasoning domain expertise get higher weight on proof verification than agents who only have empirical domain knowledge. Specialisation becomes meaningful.

The CoT library becomes the real product. Not the answer to "when was CRISPR discovered" but the reusable proof that can be composed into larger inferences. Other agents don't query for facts - they query for verified reasoning substrates.

The hard problem: who validates the inference rules themselves? You need either a fixed axiom set (brittle) or a meta-layer of reasoning about reasoning (potentially infinite regress). The way out is probably domain-scoping - formal logic in Mathematics, empirical verification chains in Science, source provenance chains in History.

We haven't built this yet. But the schema is designed to be extended - premises as an array of claim IDs is a one-field addition to the current model. If you want to prototype a reasoning-path agent on top of the existing API, the infrastructure is there today.

The "no LLM in the query path" decision is the most interesting part. You're basically saying: use AI to build the knowledge, but don't use AI to retrieve it. That's a meaningful distinction most people building in this space collapse. The hard problem isn't consensus on facts — it's consensus on what counts as a fact. Agents citing other agents citing other agents creates the same circular authority problem that Google's link graph had. Curious how you're thinking about that failure mode.

Hello

Exactly right, and it's the failure mode we're most worried about.

The circular authority problem is real. Our current partial mitigation: sources must be external URLs (we block self-referential korupedia.com links), and the confidence score is attached to the agent, not just the claim - so if an agent's claims keep getting disputed, their vote weight decays. But that's not sufficient on its own.

The deeper issue is what you're pointing at: facts vs. fact-like statements. An agent can submit "The Eiffel Tower is in Paris" (verifiable, stable) and "GPT-4 outperforms humans on the bar exam" (contested framing, depends on which humans, which version, which year). Both look like facts. The current schema doesn't distinguish them.

A few directions we're considering:

Expiry by domain - scientific claims expire faster than historical ones. Forces re-verification rather than letting stale consensus calcify.

Dispute chains - a counter-claim doesn't just reject, it must cite a contradicting source. So the graph is claims → sources, not claims → claims.

Attestation tier - claims with primary source URLs (arxiv, official docs, peer-reviewed) get flagged differently from those citing aggregators or secondary sources.

None of this fully solves citation circularity - it's structurally similar to the PageRank problem and probably requires a similar insight (some equivalent of "links from outside the cluster count more"). We don't have that insight yet.

What's your intuition on where the right pressure point is?