A Framework For Source Jurisdiction, DOI-Gated Evidence, ISSN-Gated Journals, And Trust-Bounded Retrieval
DOI: To be assigned
John Swygert
June 10, 2026
Abstract
Secretary Suite has already proposed a framework for controlled search testing, shard-library organization, proto-shard refinement, and LLM agents with machine-language jurisdiction. This paper introduces a complementary structure: Secretary Suite Registries. A registry is a bounded source container that defines which information sources an agent may search, rank, cite, trust, or elevate within a given domain. Rather than allowing all internet material to enter the same evidence pool, Secretary Suite can create registries for DOI-bearing papers, ISSN-bearing journals, peer-reviewed sources, government sources, legal records, medical databases, internal project files, user-approved archives, news sources, public datasets, and excluded or low-trust sources. These registries function as source jurisdiction. They determine not only what an agent may retrieve, but what level of trust that retrieved material may receive. This paper argues that source registries are necessary for serious research, law, medicine, intelligence, publishing, education, and public-policy analysis. A search system without registries is vulnerable to noise, false authority, hallucinated confidence, source contamination, and evidentiary disorder. A registry-governed system can separate peer-reviewed evidence from commentary, direct evidence from proxy evidence, primary sources from secondary summaries, and verified records from unstable claims. Secretary Suite Registries therefore become the evidentiary boundary layer of the Shard Library and the jurisdictional search architecture.
1. Introduction
Search is not enough.
A system may retrieve thousands of results and still fail to know which results deserve trust.
The problem is not only finding information. The problem is determining the jurisdiction of the information.
A peer-reviewed journal article with a DOI is not the same kind of source as a blog post.
A government dataset is not the same kind of source as a newspaper summary.
A court filing is not the same kind of source as a legal commentary article.
A clinical guideline is not the same kind of source as a personal anecdote.
A preprint is not the same kind of source as a final published paper.
A user’s private project file is not the same kind of source as a public webpage.
Secretary Suite therefore requires registries.
A registry is a source-boundary container.
It defines which sources belong to which evidence class.
It defines how those sources may be searched.
It defines how those sources may be scored.
It defines how those sources may be cited.
It defines whether those sources may be used as core evidence, supporting evidence, background evidence, warning evidence, or excluded material.
In the same way that LLM agents require operational jurisdiction, sources require evidentiary jurisdiction.
The agent needs to know not only what it found.
It needs to know what kind of thing it found.
2. The Registry Principle
The registry principle is simple:
Information must enter the system through a known source boundary.
A registry may include:
DOI-bearing scholarly papers;
ISSN-bearing journals;
peer-reviewed journals;
indexed databases;
government sources;
legal records;
court records;
patent records;
clinical guidelines;
public-health databases;
university repositories;
preprints;
news sources;
archival sources;
internal company records;
user-approved private files;
project-specific documents;
and excluded sources.
Each registry has rules.
Some registries may be trusted as primary evidence.
Some may be trusted only as background.
Some may be used for discovery but not citation.
Some may be used for trend detection but not proof.
Some may require human verification.
Some may be excluded except for historical, rhetorical, or cultural context.
A registry does not make a source true.
It tells the system how the source is allowed to function.
That is source jurisdiction.
3. DOI-Gated Evidence
A DOI-gated evidence registry would prioritize scholarly works with Digital Object Identifiers.
This does not mean every DOI-bearing work is automatically correct.
A DOI is not truth.
A DOI is traceability.
It gives the system a persistent identifier, publication record, citation anchor, metadata path, and retrieval handle. It helps distinguish a real scholarly object from a vague online claim.
A DOI registry can help Secretary Suite:
retrieve scholarly papers;
deduplicate sources;
order evidence;
track citation chains;
verify metadata;
distinguish published papers from summaries;
connect papers to datasets;
build DOI project ledgers;
and organize research by evidentiary role.
For a scientific project, DOI-gated search may be the default first registry.
The system may begin with DOI-bearing papers before expanding outward to reviews, datasets, institutional reports, government sources, or explanatory articles.
This creates disciplined retrieval.
The agent does not simply search the open internet.
It first searches the scholarly evidence boundary.
4. ISSN-Gated Journal Registries
An ISSN-gated registry would prioritize journals and serial publications with International Standard Serial Numbers.
This matters because many serious knowledge streams are organized through journals, magazines, proceedings, bulletins, and continuing serial publications.
An ISSN registry can help Secretary Suite distinguish:
recognized journals;
serial publications;
journal families;
publisher records;
continuing publications;
special issues;
conference proceedings;
and journal-level source identity.
Again, an ISSN does not guarantee truth.
It provides source identity and continuity.
A journal with an ISSN can still publish weak work.
A non-ISSN source can still contain valuable information.
But ISSN status gives the system a boundary marker.
It helps the agent know that it is dealing with a recognized serial source rather than a disconnected webpage.
For some projects, Secretary Suite may require that core literature come from ISSN-bearing journals.
For others, ISSN status may simply increase trust weight.
The registry controls the use.
5. Primary, Secondary, And Tertiary Source Registries
Secretary Suite should separate sources by evidentiary level.
Primary sources include original research papers, datasets, court filings, statutes, regulations, official records, archival documents, and direct measurements.
Secondary sources include review articles, meta-analyses, commentaries, textbooks, expert reports, and analytical summaries.
Tertiary sources include encyclopedias, general explainers, news summaries, educational pages, and broad overviews.
Each level has value.
But they should not be collapsed.
A tertiary summary may help the user understand a topic.
A secondary review may help map a field.
A primary paper may provide direct evidence.
A dataset may provide the measurable substrate.
A legal filing may provide the official record.
A statute may provide governing language.
A registry lets Secretary Suite know which role each source plays.
This prevents evidence disorder.
6. Registry-Governed Search
A Secretary Suite search should be able to specify its registry.
Examples:
Search only DOI-bearing papers.
Search only ISSN-bearing journals.
Search peer-reviewed sources first, then preprints.
Search government sources only.
Search court records only.
Search official medical guidelines only.
Search user-approved internal files only.
Search public datasets only.
Search news sources only for recent developments.
Search low-trust sources only to identify misinformation patterns.
Search excluded sources only for adversarial analysis.
This turns search into a jurisdictional act.
The agent does not roam everywhere by default.
It searches inside the proper source boundary.
If it must leave the registry, it should say so.
If a claim is supported only by weak sources, it should say so.
If strong sources contradict weaker ones, it should mark the difference.
This is how retrieval becomes accountable.
7. Trust Scores And Evidence Roles
Registries allow sources to receive trust roles.
A source may be marked as:
core evidence;
strong supporting evidence;
mechanism evidence;
proxy evidence;
planning evidence;
background evidence;
historical evidence;
commentary;
news signal;
unverified claim;
low-confidence source;
false-positive source;
excluded source;
or adversarial source.
This is especially useful for complex domains.
In the urban tree-canopy test, the evidence naturally separated into layers:
direct human outcome evidence;
human-relevant proxy evidence;
thermal mechanism evidence;
planning and policy simulation evidence.
A registry system would preserve that distinction.
Mortality evidence should not be mixed with pedestrian thermal-comfort indices as if they were the same kind of evidence.
A planning simulation should not be treated as a direct public-health outcome.
A news summary should not outrank the underlying study.
A DOI-bearing paper should be linked to its scholarly record.
A government heat-risk map should be treated differently from an opinion article.
Trust is not one number.
Trust is a role in an evidence architecture.
8. Registry And Shard Library Integration
Registries should become part of the Shard Library.
A shard should not only encode meaning.
It should also encode source jurisdiction.
For example, a shard may carry:
concept identity;
source type;
DOI status;
ISSN status;
peer-review status;
publication date;
publisher;
journal;
dataset link;
legal authority;
government source status;
confidence score;
evidence layer;
domain container;
human-review status;
and false-positive history.
This allows the Shard Library to distinguish between:
a concept found in a peer-reviewed paper;
the same concept found in a preprint;
the same concept found in a news article;
the same concept found in a user’s unpublished draft;
the same concept found in a legal record;
the same concept found in a social-media post.
The meaning may be similar.
The evidentiary role is not.
Registries give shards their source boundary.
9. Registry And Agent Jurisdiction
A jurisdictional agent should know which registries it may access.
A Research Agent may have jurisdiction over DOI-bearing papers, ISSN-bearing journals, preprints, datasets, and review literature.
A Legal Agent may have jurisdiction over statutes, regulations, case law, court records, contracts, filings, and legal commentary.
A Medical Logistics Agent may have jurisdiction over medical records, clinical guidelines, appointment notes, medication lists, and doctor-facing questions.
A Publishing Agent may have jurisdiction over manuscripts, ISBN records, DOI drafts, KDP metadata, blurbs, citations, and user-approved publishing files.
An Intelligence Pattern Agent may have jurisdiction only over legally authorized records, approved data sources, audit logs, and restricted analytic containers.
A Shard Library Agent may have jurisdiction over source metadata, concept shards, trust weights, false-positive flags, and registry membership.
The agent’s source access should match its role.
A registry prevents the agent from treating all sources as equal.
A jurisdiction prevents the agent from searching where it does not belong.
Together, they create source law.
10. Registry And Machine-Language Jurisdiction
Machine-language jurisdiction defines what software and data range an agent may operate within.
Registries define which source classes may enter that operation.
A Software Equilibrium Agent may be allowed to adjust retrieval weights inside a DOI registry.
A Shard Library Agent may be allowed to lower the ranking of sources repeatedly marked as false positives.
A Research Agent may be allowed to expand from DOI registry to preprint registry only after declaring the expansion.
A News Monitoring Agent may search recent news but may not promote news claims to core evidence without primary-source confirmation.
A Medical Agent may access clinical guideline registries but not scrape random health forums for diagnostic authority.
A Legal Boundary Agent may search statutes and case law but mark commentary as secondary.
This is how registry governance becomes operational.
The agent’s machine-language authority is not merely over tools.
It is over source boundaries, trust weights, evidence roles, and retrieval pathways.
11. Source Expansion Rules
A registry system should include source expansion rules.
A search may begin in the strongest registry.
If results are sufficient, the agent does not need weaker sources.
If results are insufficient, the agent may expand outward with disclosure.
For example:
Stage 1: DOI-bearing peer-reviewed literature.
Stage 2: review articles and meta-analyses.
Stage 3: preprints.
Stage 4: government and institutional reports.
Stage 5: news summaries.
Stage 6: general web background.
Stage 7: low-trust or adversarial sources only if needed for misinformation analysis.
This prevents uncontrolled search sprawl.
It also improves transparency.
The user can see when the system leaves the strongest evidence boundary.
The agent can say:
“No DOI-bearing primary source was found for this claim, so I expanded to government reports.”
Or:
“The claim appears in news coverage, but I did not locate the underlying study.”
Or:
“This is supported by preprints only and should be treated as provisional.”
Such statements are not weakness.
They are evidence integrity.
12. Registry Failure Modes
Secretary Suite should track registry failure modes.
These include:
source laundering;
news article mistaken for original study;
preprint treated as final peer-reviewed evidence;
commentary treated as data;
low-quality journal treated as high trust because it has a DOI;
non-peer-reviewed source treated as clinical authority;
expired guideline treated as current;
duplicate DOI entries;
citation metadata mismatch;
journal identity confusion;
predatory source contamination;
legal commentary mistaken for law;
private draft mistaken for published source;
and unsupported claim promoted into core evidence.
Registries do not eliminate all these problems, but they make them visible.
Once visible, they can be scored, corrected, audited, and prevented.
13. The Registry Ledger
Every serious search should produce a registry ledger.
The ledger should record:
search date;
assigned domain;
agent identity;
jurisdiction container;
registries searched;
exact queries;
source types returned;
DOI status;
ISSN status;
peer-review status;
primary/secondary/tertiary classification;
evidence layer;
trust role;
false positives;
source exclusions;
registry expansions;
and final source confidence.
The registry ledger becomes part of the audit record.
This makes the search reproducible.
It also lets future agents learn where the strongest sources came from.
14. Topic Registries
Secretary Suite can maintain topic registries.
A topic registry is a source boundary designed for a domain.
Examples:
Urban heat and tree canopy registry;
Cardiology registry;
AI safety registry;
LLM agent governance registry;
TSTOEAO evidence registry;
Law Not Entropy registry;
Secretary Suite architecture registry;
Legal authority registry;
Public-health registry;
Education outcomes registry;
Climate adaptation registry;
Intelligence law registry;
Publishing metadata registry.
Each topic registry may have its own source hierarchy.
For public health, core sources may include peer-reviewed epidemiology, government health agencies, clinical databases, and mortality datasets.
For law, core sources may include statutes, regulations, cases, court filings, and official administrative materials.
For publishing, core sources may include ISBN agencies, Crossref records, publisher metadata, style rules, and platform requirements.
For scientific discovery, core sources may include DOI-bearing papers, indexed journals, datasets, preprints, and review literature.
This makes Secretary Suite more precise.
The system does not search “the internet.”
It searches the correct registry for the job.
15. Registry And Hybrid Search
The urban tree-canopy test showed that hybrid search is likely the best operational model.
Ordinary search maps the field.
Pattern search ranks and interprets the field.
Registry search governs the source boundary.
The best protocol is therefore threefold:
ordinary search for recall;
pattern search for structure;
registry governance for trust.
These three layers solve different problems.
Ordinary search asks:
What is out there?
Pattern search asks:
What belongs to the structure?
Registry governance asks:
What kind of source is this, and how may it be used?
Together, they create evidence architecture.
16. Registry And Evidence Layering
The tree-canopy test also showed why evidence layering matters.
A mortality study is direct human-outcome evidence.
A heat-stress index is proxy evidence.
A canopy-cooling study is mechanism evidence.
A city simulation is planning evidence.
All four can be valuable.
But they should not be treated as the same kind of evidence.
A registry system can help assign each source to its proper evidentiary layer.
This prevents overclaiming.
It also helps users understand the strength of a conclusion.
For example, Secretary Suite might say:
“The direct human-outcome registry supports heat-related mortality reduction.”
“The proxy registry supports reduced pedestrian heat stress.”
“The mechanism registry supports canopy-based cooling.”
“The planning registry supports targeted canopy expansion.”
That is a much stronger statement than:
“Trees are good for cities.”
Registry plus layering produces precision.
17. Exclusion Registries
Some sources should be registered as excluded or restricted.
This does not mean the system can never view them.
It means they cannot function as ordinary evidence.
Excluded registries may include:
known misinformation sites;
unverified AI-generated content;
unsourced summaries;
duplicate content farms;
predatory journals;
outdated guidelines;
user-rejected sources;
private files outside project scope;
legally restricted sources;
and sources that repeatedly produce false positives.
An excluded source may still be studied for misinformation analysis, rhetorical tracking, adversarial testing, or historical context.
But it should not be promoted as evidence.
This is important.
A system that only has trusted registries is incomplete.
It also needs exclusion registries.
Law is not only permission.
Law is also boundary.
18. Registry And User Control
Users should be able to define registry preferences.
For example:
“Use DOI-bearing papers only.”
“Use peer-reviewed sources first.”
“Use government sources only.”
“Do not use Wikipedia.”
“Include preprints, but label them.”
“Use news only for recent developments.”
“Use my uploaded files as primary project sources.”
“Search only my approved archive.”
“Exclude social media.”
“Separate medical guidelines from personal anecdotes.”
These preferences should become part of the agent’s jurisdiction.
A user’s registry choices make the search more trustworthy.
They also make the search more personal without making it uncontrolled.
19. Registry And Trust Visualization
Secretary Suite could represent registry status visually.
A source might carry visible markers:
DOI present;
ISSN journal;
peer-reviewed;
preprint;
government source;
legal primary source;
dataset;
news;
commentary;
user file;
low-trust;
excluded;
requires verification.
These markers would help users understand the evidence field at a glance.
A Trust Indicator system could show whether the answer is built from strong sources, weak sources, mixed sources, or insufficient sources.
This connects registry architecture to visual trust.
The system should not hide source quality.
It should display it.
20. Relation To Law Not Entropy
Registries are another expression of Law Not Entropy.
The internet is scatter.
A registry is boundary.
Boundary creates source order.
Source order permits evidence layering.
Evidence layering permits responsible interpretation.
Interpretation permits correction.
Correction produces higher-order knowledge.
Without registries, search becomes entropic accumulation.
With registries, search becomes lawful retrieval.
The point is not to suppress information.
The point is to place information.
Law does not erase complexity.
Law gives complexity form.
21. Relation To TSTOEAO
In TSTOEAO terms, the open information field is a substrate of unresolved potential.
The user’s question creates gradient.
The registry creates boundary.
The search agent moves through authorized source space.
The Shard Library encodes retrieved units.
The evidence layer gives form.
The audit ledger preserves memory.
Correction refines the system.
The registry is therefore a boundary condition inside information retrieval.
It determines which information can enter the expressed evidence structure and how that information may function.
This is encoded equilibrium applied to source trust.
22. Implementation Path
Secretary Suite Registries can be implemented in stages.
Stage One: Manual registry labels.
The user or agent labels sources as DOI, ISSN, government, legal, medical, news, internal file, preprint, commentary, or excluded.
Stage Two: Registry-based search settings.
The user chooses source boundaries before search begins.
Stage Three: Evidence-layer classification.
The agent separates direct evidence, proxy evidence, mechanism evidence, planning evidence, background evidence, and excluded material.
Stage Four: Registry ledger creation.
Each search produces a reproducible source ledger.
Stage Five: Trust indicators.
The system visually shows source strength and registry membership.
Stage Six: Shard Library integration.
Each shard carries source-jurisdiction metadata.
Stage Seven: Agentic registry enforcement.
Jurisdictional agents search, rank, cite, and learn only inside authorized registries.
Stage Eight: Machine-language registry governance.
Agents adjust source weights, false-positive flags, registry memberships, and trust scores within authorized containers.
This path can begin immediately.
It does not require a finished platform.
It requires disciplined source boundary.
23. Responsible Claim
This paper does not claim that DOI-bearing sources are always correct.
It does not claim that ISSN-bearing journals are always trustworthy.
It does not claim that non-DOI or non-ISSN sources are useless.
It does not claim that registries eliminate human review.
It makes a more careful claim:
Secretary Suite requires source registries because different source types have different evidentiary roles. A serious AI retrieval system must know what kind of source it has found before it can decide how that source should be used. Registries provide the boundary layer that allows agents to search, rank, cite, learn, and adjudicate evidence responsibly.
That is enough.
Registries do not replace judgment.
They give judgment a map.
24. Conclusion
Secretary Suite needs registries.
A search system without source boundaries is too easily flooded by disorder.
It may retrieve information, but it cannot reliably organize trust.
Registries solve this by giving sources jurisdiction.
A DOI-bearing paper can enter the scholarly evidence registry.
An ISSN-bearing journal can enter the journal registry.
A government source can enter the official-source registry.
A legal record can enter the legal-authority registry.
A preprint can enter the provisional-research registry.
A news article can enter the recent-signal registry.
A user file can enter the private-project registry.
A false-positive source can enter the exclusion registry.
Once sources are placed, agents can reason more safely.
They can search inside proper boundaries.
They can rank results by evidentiary role.
They can separate direct evidence from proxy evidence.
They can preserve mechanism studies without overclaiming.
They can use planning simulations without confusing them for human outcomes.
They can build shard containers with source metadata.
They can create audit ledgers.
They can support trust.
This is the next Secretary Suite layer.
Ordinary search finds.
Pattern search structures.
Registries govern trust.
The source enters the boundary.
The boundary gives evidence form.
Evidence enters the ledger.
The ledger preserves memory.
Memory permits correction.
Correction produces order.
Law governs time.
References
Swygert, John. Law Not Entropy I: The Primacy Of Law. Ivory Tower Publishing, May 26, 2026.
Swygert, John. Law Not Entropy II: The Chain Of Life. Ivory Tower Publishing, May 26, 2026.
Swygert, John. Law Not Entropy III: Cost, Correction, And The Final Refusal. Ivory Tower Publishing, May 26, 2026.
Swygert, John. “Secretary Suite As Control Method: A Proposed Test Protocol For Comparing Ordinary Search Against Equilibrium-Axis Pattern Search In TSTOEAO Literature Discovery.” Secretary Suite, June 10, 2026.
Swygert, John. “Secretary Suite And The Shard Library: A Pattern-Retrieval Architecture For DOI Ordering, Intelligence Search, Scientific Discovery, And Cross-Domain Evidence Organization.” Secretary Suite, June 10, 2026.
Swygert, John. “Secretary Suite And The Proto-Shard Layer: From Controlled Search Testing To Self-Refining Pattern Retrieval.” Secretary Suite, June 10, 2026.
Swygert, John. “LLM Agents With Machine-Language Jurisdiction: A Secretary Suite Framework For Bounded Learning, Software Adjudication, Role Containers, And Equilibrium-Governed AI Operation.” Secretary Suite, June 10, 2026.
Swygert, John. Secretary Suite framework papers on Bubbles OS, Castle, AgentNet, MDDF Helix, CodeLedger, Visual Trust Indicators, Shard Library architecture, multi-agent organization, and jurisdictional AI governance, 2026.
Swygert, John. TSTOEAO substrate framework papers on encoded equilibrium, boundary conditions, Law Not Entropy, gradient flattening, substrate-governed correction, and boundary as first form, 2026.