[ MPS/0.1 · learning center ]
Research should not lose its boundaries when it travels.
These short guides explain the practices behind the Maha Provenance Standard: how a substantive claim can retain its source, epistemic status, scope, and revision history when people—or AI systems—reuse it.
What is claim-level provenance?
The minimum record that lets a claim keep its source, status, scope, and review history when it is quoted or reused.
Read guide →How should AI-assisted research be cited?
A practical distinction between citing the work, disclosing the instruments, and tracing the sources behind individual claims.
Read guide →How do source, interpretation, and speculation differ?
A compact reading and writing method for keeping evidence, inference, and possibility from being flattened into one voice.
Read guide →[ MPS implementation library ]
Decide where AI belongs before deciding what it should say.
This practical library extends the Learning Center from claim provenance into deployment choices. It compares on-device, cloud, and hybrid AI without treating any location as an automatic privacy, security, performance, or sovereignty outcome. Start with a workload, map its data and dependencies, and test the real device and network conditions.
AI implementation decision framework
Compare privacy, capability, cost, latency, resilience, and device requirements before selecting a local, cloud, or hybrid boundary.
Open guide →On-device AI vs cloud AI
The existing canonical workload-level comparison: where inference belongs, what each option requires, and why hybrid is often appropriate.
Open guide →Implementation guides
Practical starting points for individuals, schools, small organizations, and developers.
Open guide →Reference architectures
Bounded field, school, and internal-search patterns with data flows, measurements, failure paths, and sources.
Open guide →[ MPS implementation library ]
Choose a boundary, then test it.
Decision framework
Compare privacy, capability, cost, latency, resilience, and device fit.
On-device vs cloud
The canonical workload-level deployment comparison.
Individuals
Choose a bounded, reversible personal workflow.
Schools
Plan classroom use with student-data and access boundaries.
Small organizations
Start with a managed, testable operational use case.
Developers
Build and measure a local, cloud, or hybrid boundary.
Glossary
Plain-language deployment and provenance terms.
Methodology & sources
What this library measures, assumes, and does not claim.
What these guides are
A public explanation of one project’s methodology and tools. They use examples from Maha work, including the Research Context Registry and the De Sitter Atlas, to show the difference between a visible source trail and a bare assertion.
What they are not
They are not peer-reviewed research, legal guidance, a general certification scheme, or a substitute for reading primary sources. MPS records what was checked and how a claim is framed; it does not make a claim true.
[ MPS learning center ]
Learn the practice before using the tool.
Claim-level provenance
What must travel with a claim for it to remain inspectable.
Citing AI-assisted research
How disclosure, citations, and source limits work together.
Source, interpretation, speculation
A practical way to separate evidence from judgement.
MPS is a self-published framework and audit aid. It does not certify truth, replace primary-source review, or make an AI output authoritative.