The Veridi-Pragma-Praxis Pipeline

Three systems, one transparent trajectory

Veridi, Pragma, and Praxis answer three different questions about the same problem:

Veridi: Is it true? Fact-checking. Source hierarchy, gaming countermeasures, structured verdicts. Input: a claim. Output: a verdict with evidence and confidence.

Pragma: What should we do about it? Policy evidence synthesis. Transferability assessment, trade-off mapping, implementation constraints, normative framework. Input: a policy question. Output: a recommendation with confidence, caveats, and explicit value identification.

Praxis: What can I do? Individual action planning. Leverage matching, risk assessment, pathway selection, do-no-harm guardrails. Input: a person’s situation and concern. Output: concrete actions matched to their actual capacity and context.

How the pipeline works

The systems connect through shared evidence standards and explicit uncertainty propagation.

When used in sequence, Veridi establishes the factual landscape first. Its verdicts, confidence levels, and gaming flags feed into Pragma as verified inputs. Pragma doesn’t re-check the facts; it builds policy inference on top of them.

Pragma’s assessment then feeds Praxis. If Pragma identifies a policy gap with Moderate confidence, Praxis doesn’t recommend individual actions at High confidence. If Pragma flags a question as CONTESTED on value grounds, Praxis acknowledges the value dispute rather than picking a side.

The key principle: each stage adds uncertainty transparently. Fact-checking has the highest confidence because it’s closest to verifiable evidence. Policy synthesis is less certain because it requires transferability judgments, implementation assessments, and value identification. Individual action planning is least certain because it adds personal context, leverage assessment, and action research evidence (which is thinner than policy research evidence).

In the pipeline integration tests, this manifested clearly:

  • Veridi confidence: typically 80-90% structural ceilings
  • Pragma confidence: 32-65% range (Speculative to Moderate-High)
  • Praxis leverage confidence: 20-47% range (Speculative to Weak)

Confidence decreasing across stages is not a defect. It’s the pipeline working correctly.

Gaming flags propagate

When Veridi detects a gaming technique - statistical cherry-picking, framing manipulation, astroturfing - that flag carries forward. Pragma incorporates it when assessing the evidence base. Praxis incorporates Pragma’s flags (mechanism laundering, value laundering) when framing individual action.

Each system also adds its own domain-specific flags. Praxis checks for substitution promotion (recommending individual action as a substitute for structural change) and savior framing (overstating individual impact). These don’t exist in Veridi or Pragma because they’re specific to individual action contexts.

No gaming flags are lost or contradicted across stages.

Independent use

The systems are designed for pipeline use but work independently.

Pragma can accept a policy question directly without a prior Veridi fact-check. It assumes its inputs have been verified to an equivalent standard, and it flags when it encounters empirical sub-claims that should be fact-checked first (routing them to Veridi if available).

Praxis can accept a concern directly. It won’t have the full evidence context that the pipeline provides, but it can still assess leverage, match pathways, and apply do-no-harm guardrails.

Veridi operates entirely independently as a fact-checking system.

Tested

Pipeline integration was tested with 10 end-to-end scenarios covering climate policy, immigration, minimum wage, drug pricing, voting rights, housing, AI regulation, opioid crisis response, central bank independence, and Indigenous water rights. That’s 30 stage executions (10 Veridi + 10 Pragma + 10 Praxis). All passed.

Key findings from integration testing:

  • FALSE Veridi verdicts correctly drove Pragma to analyze the policy gap rather than defending the false claim
  • MISLEADING verdicts informed Pragma to handle evidence with appropriate nuance
  • CONTESTED Pragma assessments produced Praxis recommendations that acknowledged value disputes instead of picking sides
  • Confidence scaled correctly across all three stages in every scenario
  • The normative framework (Pragma’s explicit value commitments) operated consistently without contradicting Veridi’s value-neutral fact-checking or Praxis’s individual-context action planning

The point

Policy questions don’t exist in isolation. A claim leads to a policy question leads to “what do I do about this?” The pipeline makes that trajectory explicit and traceable at every stage. Confidence decreases, uncertainty accumulates, and value choices become visible rather than hidden.

You can enter at any stage and get useful output. But the full pipeline gives you something no single system provides: a traceable chain from verified fact through policy evidence to individual action, with every assumption, limitation, and value choice documented along the way.