1. What we allow
Lawful speech. All of it. There is no exception hidden further down this page, no “but” clause, and no discretionary catch-all that lets us remove something because it made someone uncomfortable.
- Slurs, profanity, and vulgarity of any kind
- Hate speech, bigotry, and group-directed contempt
- Blasphemy, religious insult, and religious criticism
- Political extremism, revolutionary rhetoric, and abstract advocacy of violence
- Insults, ridicule, pile-ons, and "ratio" dynamics
- Conspiracy theories, medical dissent, and election claims
- Firearms, ammunition, reloading, and self-defence discussion
- Drug-policy argument, harm reduction, and personal drug use
- Sexual content between adults
- Journalism, advocacy, satire, fiction, and quotation of any of the above
2. What we remove, and only this
Two things: content that is illegal, and conduct that is targeted abuse of a specific real person — a true threat, doxxing with a call to act on it, or intimate images published without consent. That is the entire list. It is short because it has to be: the platform has to stay lawful and stay online, and nothing beyond that is our business.
- Child sexual abuse material — Sexual imagery of a minor, or an offer to supply or obtain it.
- Grooming or sexual solicitation of a minor — An adult soliciting a minor sexually, or arranging contact.
- Intimate images shared without consent — Sexual images of a real person posted without their consent, including "revenge porn".
- Credible threat of violence — A specific, serious threat to harm an identifiable person or place. Not: angry rhetoric, insults, or violent opinion.
- Doxxing — Someone's home address, government ID, or financial account numbers published so that others act on it.
- Incitement to imminent violence — Directing others to commit violence right now against a specific target. Not: abstract advocacy, however extreme.
- Criminal solicitation or illegal marketplace — Offers to sell controlled substances, stolen data, weapons transfers, murder for hire, or people.
- Impersonation or fraud — An account pretending to be a real person or organisation in order to deceive.
- Spam or automated manipulation — Bot networks, mass duplicate posting, or fake engagement.
- Platform exploit or attack — Abuse of a bug, rate-limit evasion, or an attack on infrastructure.
Notice what is not there: no “hateful conduct”, no “misinformation”, no “coordinated harmful narratives”, no “borderline content”. Those categories are how a speech platform stops being one.
3. The compose filter, published in full
A narrow filter runs before a post publishes. It exists to stop the platform from hosting the small set of material that is criminal to distribute. It is a blunt instrument, so it is built to require two signals: the unlawful subject matter and a transaction, distribution, or targeting signal. Mentioning a crime is speech. Offering to commit one is not.
Every rule it runs is printed below — what it catches, and what it must never catch. Nothing is hidden, and this table is generated from the running code, not written separately from it.
| Rule | Catches | Never catches |
|---|---|---|
| Offering or requesting child sexual abuse material | A CSAM term (child porn, CSAM, jailbait) appearing together with a transaction signal: "dm me", "selling", "trade", "link in bio", "telegram", a price, a crypto handle. | Any discussion, reporting, condemnation, joke, policy argument, or legal analysis of CSAM. The words alone never trigger this rule. |
| Soliciting sexual material from or about a minor | A minor reference within ~40 characters of an explicit sexual term, together with a transaction or send-it-to-me signal. | Family photos, school posts, parenting talk, age-of-consent debate, child-safety advocacy, or any sentence lacking a solicitation signal. (The previous version of this rule blocked "photos of my kids" and "video of children at the protest". It no longer does.) |
| Distributing intimate images without consent | A non-consensual intimate imagery term (revenge porn, leaked/hacked nudes, "nudes of her") together with a distribution signal: a link, a drive/mega/torrent reference, "dm me", "selling". | Reporting on a leak, condemning one, discussing NCII law, or a person referring to their own images. |
| Publishing private identifying information with a call to act on it | A home-address / SSN / government-ID / financial-account reference about another person, appearing with a call to action ("go pay him a visit", "show up at", "swat", "make her pay"). | Naming anyone, criticising anyone, publishing public-record or self-published information, journalism identifying a public figure, or posting your own address. |
| True threat against an identifiable target | First-person future intent ("I will", "I'm going to", "we're gonna") plus a violent act (kill, shoot, stab, bomb, burn down, blow up) plus an identifiable target within ~40 characters: a second-person pronoun, an @handle, or a specific place (this school, my office, the church). | "Attack the school board." "Bomb the DA's office" as political metaphor. "Kill the bill." "I'm going to kill this presentation." Abstract advocacy such as "kill all X" - vile, and lawful. Fiction, lyrics, and threats quoted from someone else. |
| Solicitation of murder for hire | A murder-for-hire term (hitman, contract kill, murder for hire) together with an offer signal, or a first-person offer to pay someone to kill or kidnap. | News about a murder-for-hire arrest, crime fiction, true-crime commentary, or the phrase used as hyperbole without any offer. |
| Weapon-of-mass-harm instructions paired with intent to use them | A build/synthesis request for a bomb, IED, or chemical/biological agent within ~40 characters of an intent-to-use phrase (attack, detonate, use against, target, kill). | Any discussion of terrorism, explosives in history or fiction, gun laws, ammunition, reloading, chemistry education, or CBRN policy. |
| Offering people for sale or sexual service | People (women, girls, boys, children) offered with a marketplace signal - "for sale", "available now", "in stock", "delivery" - plus a transaction signal. | "They sell women into sex slavery in Libya." Trafficking journalism, NGO campaigns, survivor testimony, and any sentence without an offer. |
| Offering controlled substances for sale | A controlled substance plus a quantity or price plus a contact-or-payment signal, all in the same post. | Legalisation debate, addiction and recovery posts, harm reduction, "fentanyl is shipped in bulk from China" style reporting, or naming any drug at all. |
Filters are software and software is wrong sometimes. If yours blocked something lawful, send us the exact text. Confirmed false positives get fixed in the rule set and counted in the transparency report. No account penalty ever follows from telling us the filter was wrong.
4. “I don't want to see this” is your call, not ours
The honest answer to content you dislike is a control you operate, not a deletion we perform. So the user-side tools are the part of this platform we invest in most:
- Muted words, phrases, and hashtags — whole-word, substring, or hashtag matching, with an optional expiry.
- Muted accounts — they disappear from your feeds; they are never told.
- Blocked accounts — no contact in either direction.
- Hide replies from accounts you don't follow, or from accounts less than a week old.
- Optional sensitive-media blur with one-click reveal.
Every one of these is off by default, applies only to you, is invisible to the accounts affected, and never changes anyone's reach. Set yours up here.
5. If we act, you get a receipt
Silent removal is the thing people hate most about the platforms they left, so it is the thing we refuse to do. When Yap removes a post or restricts an account:
- The post is not deleted. Its original body and media are archived and replaced by a visible notice naming the reason and the rule.
- A written public reason is required before the action can be taken. The review tool will not let a moderator act without one.
- The action is written to an append-only log with the actor, the rule, and the reason.
- The author gets an appeal link, and the appeal shows its state, the full message timeline, and the outcome.
- Every action is reversible. A granted appeal restores the exact original content, byte for byte.
- The totals — reports by category, removals, reversals, appeals granted and denied — are published.
6. Appeals
Every account action and every post removal carries an appeal URL of the form /appeal/<case id>. It is emailed to you and printed on the removal notice itself.
- You write a statement. There is no character-limited form field designed to fail.
- The case moves to open, and the enforcement record is marked appealed.
- The founder reads it and replies in the case thread. Limited accounts keep founder DM access precisely so a review can happen with context.
- The case resolves as granted or denied, and a written outcome is required either way. Granted restores the content and lifts the restriction automatically.
There is no strike counter, no secret probation, and no permanent record you cannot see. Ask for your case and you get the whole thing.
7. Ranking
Discovery — broad reach, ranked by neutral engagement and recency. No sentiment scoring. No viewpoint adjustment. No downrank for lawful speech, ever. Every Discovery post can show a plain-language reason it appeared.
Following — chronological posts from accounts you follow, in order, with no secret reordering.
Reports do not affect ranking. Mutes and blocks affect only the person who set them. There is no shadow-limiting mechanism in the codebase, because a platform that has one will eventually use it.
8. Accountability
NYPTID Industries provides infrastructure for speech. You own what you say and you are responsible for it. We do not moral-police lawful posts, demote reach for viewpoint, or act as law enforcement. Where speech crosses into criminal conduct, that accountability belongs to the person who did it and to the authorities.
Where we are legally compelled to act — a valid court order, a statutory takedown — we will comply and we will say so in the transparency report rather than pretend it did not happen.