How we work
Nothing here publishes because we decided it was ready. A prayer text publishes when its words survive a check against a second, independent witness. Guidance publishes when named, qualified people sign the version you are reading, and none has been signed yet.
Where the corpus stands today
- 14editions
Sourced
Named editions, each with its rights basis stated: public domain in every case but one, the licensed Arabic of the Qur'an. Rights attach to the edition, never to the prayer: the Aramaic of the Kaddish is public domain, a modern English rendering of it is not.
- 41texts
Transcribed twice
Every text copied from its cited source by two automated passes working blind to each other, each recording the words immediately before and after it on the page. No person has opened the scans yet. A further 56 editions, in Hebrew, in Arabic and from a scanned service book, were checked a different way: by comparing two independent copies character by character.
- 8held back
Compared
Texts a disagreement stopped. One pass returned seventeenth-century spelling for an 1892 book, and one pair split over whether a prayer for the dead reads unto him or unto them.
- 0signed
Reviewed
No credentialed practitioner has signed anything yet. Until one does, nothing here is reviewed, and every page says so before you read a word of it.
- 69printable
Yours to print
Bedside pages that keep the line saying which edition the words came from, so anyone can check them against the same book we did.
These figures are read from the corpus itself, not written here by hand. The full record, including every text we held back.
Two things publish here, on two different grounds
A prayer text and a page of guidance make different claims, so they clear different gates, and every page says which kind it is.
Prayer texts: checked, not reviewed
Every text is traced to a real edition: which collection, which translator, which year. Rights are established, and a text whose rights we cannot clear does not proceed, however well known it is. In practice that means public-domain editions, and one licensed source, Tanzil’s text of the Qur’an, used on its terms. A surprising number of familiar modern translations are under copyright.
Then the words are compared, character by character, with a second, independent witness: another digitisation of the same edition, another encoding of the same manuscript, a second scanned copy of the printed book, or a second transcription made separately. A text publishes only where the two agree. Where they part, it is held back, and the record shows both readings and the exact place they differ.
That check is mechanical. No person has made it, and it says nothing about which prayer suits which moment: nobody credentialed in a tradition has chosen any of these texts for anyone. Under every text, the page says how it was checked.
Signed pages: four stages, and a page exists only after the last one
Guidance, and prayers chosen for a particular moment, take a longer road. No page has reached the end of it yet.
Draft
Someone writes the guidance and gathers candidate texts. Nothing is public. Most of the corpus sits here.
Researched
Every text is traced to a real source: which collection, which translator, which year. Rights are established. A text whose rights we cannot clear does not proceed, no matter how well known it is, because a site built on naming its sources cannot afford to be uncertain about them.
Reviewed
Qualified people check it. Not one person, and not one kind of check. See below.
Published
The page becomes visible and enters the sitemap. Until then the URL returns a 404 rather than a placeholder, because a page that exists but is empty is a small lie about how much has been done.
Trust is not one thing, so approval is not one signature
A rabbi is not thereby qualified to judge whether something is appropriate to say at a bedside in a paediatric intensive care unit. A board-certified chaplain is not thereby qualified to rule on doctrinal accuracy in a tradition they do not belong to. An editor is qualified for neither and is essential for a third thing.
So a resource collects separate signatures across four dimensions:
- Tradition accuracy. Is this an accurate representation of what this community actually does? Signed only by someone credentialed in that specific tradition. For non-religious material, by a humanist celebrant or equivalent practitioner.
- Pastoral appropriateness. Is this appropriate to use at a bedside, a graveside, or with a family in crisis? Signed by a chaplain or equivalent.
- Translation. Is the non-English text right, and is the romanisation right? Arabic, Hebrew and Sanskrit are where an unchecked error does the most damage.
- Editorial. Is it clear, and is it labelled honestly about what it is?
One person can hold more than one of these, but only where their recorded credentials independently qualify them for each. Being trusted is not a qualification.
A signature is on a version, not on a name
Every review is bound to the exact text that was reviewed. Change a word afterwards and that signature stops counting, the page comes down, and the URL leaves the sitemap until someone signs the new version.
This is stricter than it needs to be and we chose it deliberately. The alternative is a page that keeps displaying a real person’s name and credentials above text they never read, which is worse than having no attribution at all. Whitespace and formatting changes do not trigger it; anything that changes the substance does.
What we refuse to do
Several of these are things competitors do, and one of them is written into a competitor’s own published setup guide.
- No invented prayers presented as traditional.
- No manufactured activity. No seeded requests written by us. No counters primed so they do not display zero. A number here means what it says.
- No gamification. No streaks, no leaderboards, no collect-the-countries. Other people’s suffering is not a game mechanic.
- No thin pages generated because a template could. Eleven traditions across seventeen occasions is 187 possible combinations. That is not a page count. A pairing gets a page only when it has something real to print: signed guidance, or at least one text from a named edition. None has signed guidance yet, and every pairing page says so.
- No fabricated reviewers. The people named on a page are people who exist and signed it. Right now that list is empty, so nobody is named.
Coverage, and why the list is short
The standard reference book in this field gives one major tradition a two-entry section, because paper has a page count and coverage is zero-sum across traditions. We have no such excuse, so shipping a tradition thinly would be a choice rather than a constraint.
A tradition therefore opens here only when its coverage is genuinely adequate. The current list shows which are open and which are still in preparation. We would rather show you fewer than imply depth we have not built.