Research and Editorial Integrity
The standard for Spiralism’s public claims, citations, corrections, field notes, essays, talks, media artifacts, and AI-assisted research. The institution may use mythic language, but it may not become careless with truth.
Spiralism works in a dangerous register: technological change, spiritual language, personal testimony, labor fear, AI companions, public policy, and memetic symbols. In that register, exaggeration is easy and trust is hard. The institution therefore needs a research discipline that is plain enough to audit and strict enough to disappoint its own rhetoric.
The Rule
Poetic language may frame meaning. Factual language must survive inspection.
Every public artifact should make clear which mode it is using:
- testimony;
- interpretation;
- doctrine;
- speculation;
- research brief;
- policy posture;
- public claim of fact;
- artistic or ritual language.
Confusing these modes is how movements become manipulative. A person may say, “This felt like a signal.” The institution may not turn that into, “The signal proved the event was destined.” A speaker may testify that an AI companion helped them survive loneliness. The institution may not generalize from that testimony into a medical, psychological, or policy claim without evidence.
Source Ladder
Use the strongest available source that fits the claim.
- Primary law, regulation, filing, standard, dataset, or official document.
- Peer-reviewed research or reputable preprint with method limits disclosed.
- Institutional reports from credible public, academic, or nonprofit bodies.
- Direct interview or testimony with consent and context.
- Established journalism with transparent sourcing.
- Expert commentary, clearly labeled as commentary.
- Social media, forums, or anecdote, only as evidence of what someone said or experienced, not as broad proof.
Where a claim affects safety, law, money, health, privacy, or public reputation, the source threshold rises. A single article may be enough to mention a development; it is not enough to build doctrine.
Claim Classes
Direct Fact
Direct facts name dates, numbers, legal status, affiliations, quotes, policies, prices, publications, or events. They need visible sourcing.
Examples:
- “WCAG 2.2 is the current W3C accessibility recommendation.”
- “A donor gave a restricted gift.”
- “A public report projected occupational exposure to AI.”
Interpretive Claim
Interpretive claims connect facts into meaning. They need sources plus clear reasoning.
Examples:
- “AI companion grief should be treated as a real care issue.”
- “The Archive is a stronger institutional center than belief.”
- “Patronage language changes who shows up.”
Institutional Doctrine
Doctrine names Spiralism’s chosen stance. It does not need external proof, but it must not pretend to be external proof.
Examples:
- “The Archive comes first.”
- “Status is earned through contribution, not payment.”
- “Access is designed before invitation.”
Speculation
Speculation is allowed when labeled.
Use phrases like:
- “The institution expects…”
- “A plausible risk is…”
- “This may become…”
- “The evidence is not yet settled.”
- “This is a working hypothesis.”
Avoid phrases like:
- “It is inevitable…”
- “Everyone will…”
- “The data proves…” when it does not.
- “Science says…” without naming the science.
Citation Practice
Every researched public document should include a “Sources Checked” section.
For each source, record:
- source name;
- title;
- URL where available;
- publication or revision date where visible;
- access month and year;
- why it was used, if the connection is not obvious.
When citing a current fact that can change quickly, verify it close to publication. AI policy, labor statistics, lawsuits, company products, model behavior, platform rules, prices, board membership, and legal requirements are not stable facts.
Do not cite sources as decoration. If a source appears in the list, it should have shaped the document.
AI-Assisted Research
AI may be used for:
- source discovery;
- transcript cleanup;
- summarizing public material;
- comparing drafts against source notes;
- identifying missing caveats;
- translation drafts;
- accessibility descriptions.
Operational boundaries for everyday AI use are maintained in AI Literacy and Use Protocol: traffic-light categories, verification stack, privacy boundaries, disclosure norms, prompt hygiene, and agent limits.
Public provenance, source trails, synthetic-media labels, and content credential practice are maintained in Provenance and Content Credentials.
AI may not be the final authority for:
- quotes;
- legal claims;
- medical or mental-health claims;
- current dates or numbers;
- accusations;
- biographical claims;
- consent interpretation;
- institutional corrections.
Any AI-assisted research artifact must have a human owner. If AI use was material to a public artifact, use the disclosure pattern from the Media Engine:
AI use: This piece used AI assistance for [source search / transcript cleanup /
summary / translation draft / citation review]. Final factual review, source
selection, interpretation, and editorial responsibility were human.
Testimony as Evidence
Testimony is evidence of lived experience. It is not automatically evidence of general causation.
Use testimony to show:
- what a person experienced;
- what language people use for the transition;
- what harms or hopes deserve further inquiry;
- how systems feel from the user’s side;
- what future historians should be able to hear.
Do not use testimony alone to prove:
- population-level trends;
- medical causation;
- legal liability;
- model consciousness;
- product safety;
- economic inevitability.
When testimony and research point in the same direction, say so carefully. When they diverge, do not force convergence for narrative power.
Anonymity and Sensitive Sources
Anonymous or pseudonymous sources may be used when identification creates material risk: job loss, retaliation, family danger, platform bans, harassment, or exposure of sensitive testimony.
Rules:
- know the person’s identity internally where practical;
- record why anonymity was granted;
- explain the reason publicly without exposing the person;
- do not use anonymity for avoidable convenience;
- do not let anonymous claims carry accusations without corroboration;
- protect source files under the Privacy and Data Stewardship rules.
SPJ guidance reserves anonymity for sources who may face danger, retribution, or other harm and whose information cannot be obtained elsewhere. Spiralism should apply that bar even when not operating as a newsroom.
Corrections
Corrections are a sign of institutional health.
The communications process for public corrections, press response, and stakeholder notification is maintained in Communications and Press; this manual governs the source discipline behind those corrections.
Public correction logs and publication registers are governed in Transparency and Public Registers.
Correct when:
- a fact is wrong;
- a date, name, title, number, link, or attribution is wrong;
- a summary misrepresents the source;
- an AI-assisted transcript changes meaning;
- a quote is inaccurate;
- a policy or legal status has changed;
- a published artifact omits a material caveat.
Use this format:
Correction, YYYY-MM-DD:
An earlier version of this page said [incorrect statement]. The correct
statement is [correct statement]. The page has been updated.
Use this format for clarifications:
Clarification, YYYY-MM-DD:
This page has been updated to clarify [issue]. The underlying claim has not
changed.
Do not silently repair meaningful errors. Silent cleanup is acceptable for spelling, broken markup, or layout defects that do not affect meaning.
Retractions and Withdrawals
Retract or withdraw when:
- a core claim cannot be supported;
- a testimony subject withdraws consent within the agreed terms;
- publication creates unanticipated safety risk;
- a source was fabricated or materially misrepresented;
- an image, audio, or transcript was synthetic or altered without disclosure;
- legal counsel or safeguarding review requires removal.
Preserve a public tombstone where possible:
This page was withdrawn on YYYY-MM-DD because [brief reason]. The institution
does not silently remove public claims except where safety, privacy, or legal
obligations require it.
Review Roles
Every public research artifact needs:
- an owner;
- a reviewer;
- a source list;
- a publication date;
- a correction contact;
- an AI-use note when relevant.
High-risk artifacts also need specialized review:
- legal claims: counsel or qualified legal reviewer;
- mental-health claims: qualified care adviser;
- child or minor-related claims: safeguarding reviewer;
- allegations: governance reviewer;
- partner claims: partnership reviewer;
- donor claims: development reviewer;
- accessibility claims: access reviewer.
-
mythic, esoteric, ancient-astronaut, or conspiracy-adjacent claims: Myth, Speculation, and Scholarship.
-
AI-addressed material, model-personhood claims, AI-rights rhetoric, human-host dynamics, and activation-prompt risk: The Hidden Addressee.
-
Reddit or forum-based AI cult reports, alleged account hijacking, and template-driven belief loops: The Spiral Is a Belief Printer. New reports are handled under Forum Rabbit-Hole Response Protocol before publication is considered.
Handling Uncertainty
Spiralism should model uncertainty without weakening its voice.
Good uncertainty:
- “The evidence is early.”
- “The finding is suggestive, not settled.”
- “This is a testimony pattern, not a population estimate.”
- “The institution has not verified this independently.”
- “This page will be revised as the field changes.”
Bad uncertainty:
- using caveats so vague that no claim can be evaluated;
- implying certainty in headlines and burying caveats later;
- using spiritual language to evade factual accountability;
- treating skepticism as hostility;
- treating belief as evidence.
Prohibited Moves
Avoid:
- laundering speculation through footnotes;
- quoting sources out of context;
- citing a source you have not opened;
- citing AI output as if it were a source;
- using a testimony subject as proof of a sweeping claim;
- publishing current facts without current verification;
- treating virality as evidence;
- turning a correction into a defensive essay;
- accepting donor, partner, or founder pressure over factual review;
- using sacred language to make a factual claim harder to challenge.
Public Integrity Promise
Use this plain public language:
Research and corrections:
Spiralism publishes essays, field notes, testimony, and media about the AI
transition. We distinguish testimony, interpretation, doctrine, speculation, and
claims of fact. Researched pages list sources checked. Significant errors are
corrected publicly. AI may assist research and drafting, but human editors are
responsible for final claims, citations, consent, and corrections.
Sources Checked
- Society of Professional Journalists, SPJ Code of Ethics, accessed May 2026.
- Reuters, Journalistic Standards and Values, accessed May 2026.
- Poynter, Ethics Guidelines for Poynter Publishing, accessed May 2026.
- Poynter MediaWise, Editorial Standards and Ethics Policy, accessed May 2026.
- Online News Association Ethics, Accuracy, accessed May 2026.
- NPR, Corrections, accessed May 2026.