Machine-Readable Web Intelligence, Provenance, and Trust Infrastructure
GoGuides is hybrid web-evidence infrastructure for systems that consume the open web. It combines independent observation and provenance with longitudinal state, deterministic comparison, source-use decision context, bounded derived evidence, and model-assisted analysis while keeping those evidence classes distinguishable.
AI systems need more than content retrieval. They need source identity, provenance, time, objective state, policy context, change evidence, relationship evidence, and bounded machine-readable analysis.
GoGuides separates these concerns into Evidence → Interpretation → Decision. Observed facts remain distinct from deterministic comparisons, derived evidence, model-assisted judgments, and downstream source-use decisions.
Read the Public Technical White Paper
Version 3.0 documents the current public architecture, including Web State V3, Change Intelligence, URL Request, Batch Evaluate, Claim Support, Content Relationships, reusable derived evidence, Verified Text, machine identity evidence, public contracts, licensing, limitations, and the Favicon North Star.
28 pages • Published September 15, 2026
SHA-256:
64948170f044db30e7a4599cd3d0207b0b70231b1b93fbc3c741f838fecb1fd7
GoGuides now documents a separate live layer for Current Delegated Authority State: atomic authority reservation, signed machine clearances, live pre-execution validation, and receipts/reconciliation. This layer is separate from AI Rank, web observation, provenance, and existing public trust records.
Read Amendment A Read the dedicated ACN Technical White Paper ACN Technical White Paper PDF · Authority Clearing Network documentation
The problem: AI needs more than content
A web page can be fetchable without being stable, attributable, current, appropriate for citation, or suitable for a particular machine task. A domain may change ownership, alter structured data, replace outbound links, change machine-facing directives, or substantially rewrite content while retaining the same URL.
Reconstructing that context independently requires repeated crawling, normalization, fingerprinting, historical storage, and comparison. GoGuides is designed to perform that observation work once and expose the resulting evidence through public machine-readable surfaces.
System architecture
1. Web observation
GoGuides fetches public web sources and records objective observations about content, headers, links, structured data, machine-facing directives, and other source state.
2. Evidence preservation
Observation records preserve time and state so later systems can distinguish a current snapshot from longitudinal evidence.
3. Derived evidence and interpretation
Deterministic comparisons, reusable signals, policy rules, and bounded model-assisted analysis remain tied to stored authoritative evidence rather than replacing it.
4. Machine-use decision context
Public endpoints expose compact decision-oriented records while allowing consumers to follow links into deeper State, History, verification, or primary-source evidence.
The Favicon North Star
The GoGuides favicon is a lightweight deterministic monitoring signal generated from supported GoGuides measurement inputs for the site. The same supported record state produces the same favicon output. It is not decoration, a security certificate, or proof of truth; it is a compact site- and state-specific signal that points back to deeper canonical machine-readable evidence.
A tiny public signal should be able to point back to a deeper trust record.
Web State V3
Web State V3 is the objective current-state layer exposed at
/state/{domain}. It reads the latest eligible stored observation and publishes
compact measurements without turning those measurements into claims of truth.
Current Web State V3 fields
text_simhash
structured_data_sha256
schema_type_mask
headers_sha256
word_count
internal_link_count
external_link_count
outbound_links_sha256
ai_policy_maskFields without an observed value may be omitted. Absence is not equivalent to zero.
Machine Interaction Sequence
GoGuides is designed so a machine can begin with lightweight discovery, resolve decision context for one or more URLs, inspect objective state and longitudinal evidence, and invoke specialized analysis only when the task requires it.
From discovery to source decision
AI crawler / agent
↓
Discovery / machine-readable signal
↓
Evaluate / Batch Evaluate
↓
State / Changes / History as needed
↓
Claim Support / Content Relationships when needed
↓
Primary source when needed
↓
Consumer applies its own policy1. Discover
Read machine-use guidance, AI discovery, or a lightweight signal.
2. Evaluate
Resolve decision context for a domain, exact URL, or bounded URL batch.
3. Inspect State
Use Web State when objective fingerprints, counts, directives, or normalized state matter.
4. Inspect Changes / History
Use longitudinal evidence when recency, stability, or transition matters.
5. Specialized evidence
Use Claim Support or Content Relationships when the task requires bounded analysis.
6. Fetch primary source
Retrieve the original source whenever current or complete content is required.
7. Decide
The consuming system applies its own read, cite, recommend, transact, monitor, or ignore policy.
Live machine-readable endpoints
AI record discovery
Crawlable HTML discovery surface that points automated consumers toward eligible public GoGuides records.
Machine-use discovery
Machine-readable description of GoGuides public capabilities, endpoint relationships, and usage flow.
Evaluate
/evaluate.php?domain=goguides.com&format=json
Decision-oriented evaluation for domain or exact-URL context with links into deeper authoritative evidence.
Batch Evaluate
Bounded multi-URL evaluation using the same normalization, scope resolution, and evidence authority as ordinary Evaluate.
Web State V3
Objective current-state measurements from the latest eligible stored observation.
Change Intelligence V1
Deterministic comparison of stored longitudinal evidence for the same requested URL.
URL Request
Explicit request for intelligence about a public URL with a durable request status resource. Caller intent does not create or alter authoritative evidence.
Claim Support
Evidence-bound model-assisted analysis of whether a fixed stored source observation supports, contradicts, leaves unaddressed, or cannot resolve a claim.
Content Relationships
Measurable relationship evidence for two public URLs without treating content similarity as proof of common ownership or coordination.
Observation history
Historical observation and provenance context for a domain.
Verified Text
/verified-text.php?q=gravity&format=json
Source-attributed text records with licensing, integrity metadata, and canonical full-record links.
Favicon marker
/favicon_img.php
Deterministic visual monitoring surface linked to deeper GoGuides state.
Evaluate → State
Evaluate remains decision-oriented. It does not need to duplicate every raw Web State V3 field.
When deeper objective state is useful, the evaluation can point the consumer to
state_url.
Representative relationship
{
"api_version": "evaluate-v3.2",
"domain": "goguides.com",
"trust_state": "observed",
"public_eligible": true,
"state_url": "https://www.goguides.com/state/goguides.com"
}Historical observation and provenance
Current state answers a different question from historical evidence. GoGuides preserves observations over time so consumers can inspect when a source was seen, whether its recorded state persisted, and whether later observations differ.
Time is therefore part of the evidence model rather than a decorative timestamp.
Verified Text
Verified Text extends GoGuides from domain-level web state into source-attributed text records. Records can expose source identity, licensing, attribution, hash scope, preview metadata, machine-use context, and canonical full-record URLs.
It is an integrity and provenance layer. It is not a claim that every statement in a source is true.
Machine-readable contracts and licensing
The white paper explains why the system exists, how the architecture works, what the evidence means, and where its limits are. Exact machine contracts are published separately so software does not need to reverse-engineer architectural prose.
Public contract discovery
/machine-use.json — capability discovery, machine guidance, links, and licensing metadata.
/docs/openapi.yaml — OpenAPI 3.1 definitions for documented machine routes, parameters, responses, and errors.
/docs/web-state.json — canonical Web State V3 field meanings, bitmasks, omission behavior, and compatibility rules.
GoGuides Data License 2.0 exposes machine-readable reuse terms on supported interfaces. Licensing metadata describes permitted reuse; it does not alter the evidence, rank, observation, or machine-use decision associated with a source.
Version 3.0 production expansion
Version 3.0 incorporates the machine-evidence capabilities that now sit alongside Evaluate, Web State, History, Verified Text, and Change Intelligence. Batch Evaluate provides bounded multi-source screening; URL Request provides an explicit request/status lifecycle while preserving independent observation; Claim Support binds model-assisted analysis to fixed stored evidence; and Content Relationships exposes measurable source relationships without overclaiming ownership or intent.
These capabilities share the same architectural rule: supporting analysis may supplement authoritative evidence, but it does not silently rewrite the observation that produced the underlying source state.
What GoGuides does not claim
GoGuides does not certify truth, guarantee safety, replace professional review, or prove that a website has not changed after the recorded observation time.
GoGuides publishes what it observed, preserves evidence and time, computes defined signals, exposes policy and verification context, and allows downstream systems to make their own decisions.
A request is evidence of a request, not endorsement or downstream use. A signed request proves verified key control for that request, not intent. Content similarity does not prove common ownership or coordination. A Claim Support verdict is model-assisted analysis over bounded evidence, not cryptographic fact or universal truth.
Frequently asked questions
Is GoGuides an AI model?
No. GoGuides is not a general-purpose generative AI model. It is hybrid web-evidence infrastructure combining deterministic crawling, provenance, fingerprints, policy rules, longitudinal state, bounded derived evidence, and model-assisted analysis.
What is Web State V3?
It is a compact objective state record derived from the latest eligible stored observation of a domain.
Why separate Evidence, Interpretation, and Decision?
Because an observed fact is not the same thing as a computed signal, and neither is automatically the same thing as a final source-use decision. Keeping those layers distinct makes the system easier to inspect and less likely to overstate what the evidence proves.
What is Claim Support?
Claim Support is model-assisted analysis bound to a fixed stored source observation. It can report supported, contradicted, unaddressed, insufficient evidence, or source unavailable without presenting the result as a truth certificate.
What does Content Relationships prove?
It exposes measurable relationships such as canonical identity or strong duplicate and near-duplicate evidence. Similarity by itself does not prove common ownership, coordination, copying direction, deception, or falsehood.
Does the favicon replace the JSON endpoints?
No. The favicon is a lightweight deterministic monitoring signal that points toward deeper records. State, History, Evaluate, verification, and other machine-readable surfaces remain the authoritative sources for deeper inspection.
Observe the Web. Preserve the Evidence. Publish the Signal.
Read the complete v3.0 white paper
The 28-page PDF documents the complete public architecture: independent observation and provenance, the Favicon North Star, Web State V3, machine-use decisions, longitudinal history, Change Intelligence, URL Request and machine identity evidence, Evaluate and Batch Evaluate, Claim Support, Content Relationships, reusable derived evidence, automation, manipulation resistance, Verified Text, Bot Radar, machine-readable interfaces and contracts, Data License 2.0, limitations, standards, and the current machine interaction sequence.