Omnism is the position that every religious path deserves to be taken seriously, that none is wrong merely for speaking differently, and that what repeats between them is worth attending to. An omnist belongs to no single tradition and rejects none.
This differs from syncretism, which assembles a new tradition out of several. It also differs from indifference: if everything is equally true, there is nothing to look at. Omnism demands the opposite, namely looking closely at where traditions say the same thing and where they do not.
This site does not merely claim that, it measures it. Sermons, addresses and teachings from many traditions are analysed by the same standards, and every figure leads back to the text it came from.
There is something greater than us. Call it God. No one sees it directly. Religions are the protocols people built to reach it: prayer, ritual, scripture, law. Each protocol carries the language, the culture and the knowledge of its time. That is why they differ, and why none of them holds the whole. But they are all trying to do the same thing: to tell the truth about what is greater, and to live by it. An Omnist calls none of them wrong for speaking differently. An Omnist listens to all of them, including those who search without God, compares what they say, and pays attention to what repeats. Where many paths meet, truth is most likely to be found.
Sermons, addresses and stories are collected regularly: as text from a website, from a feed, from a PDF, as the subtitles of a video, or as a machine transcript of a recording. Where a text came from is recorded on every item. A transcript is not the same thing as a written text, and that should be visible.
A language model summarises each item, assigns it one to three values from a fixed vocabulary of twelve, and measures it on several scales. It also records which of sixteen areas of life appear and how the text frames them, and answers eight questions about its view of the world. Everything runs on our own machines; nothing is sent to outside services.
The individual measurements are aggregated by period, tradition and region. The shared value is not the loudest one, but the one no tradition leaves out.
A language model does not always read nuance correctly. The classification is pattern recognition, not a theological claim. Every item stays linked to its source and can be checked word for word.
The evidence base matters more. The collection is small and unevenly spread, and some traditions are represented by a handful of texts. A reading drawn from two sermons says nothing about a religion. Every figure therefore states what it rests on, and where the basis is thin it is visibly held back rather than smoothed over.
What is missing is missing visibly. The coverage page names the religions for which not a single source exists yet, and the catalogue names the sources that are connected but deliver nothing, with the reason.
The site runs on a rented server, the domain costs money every year, and the analysis runs on our own hardware rather than at a provider. That is the expensive part: machines bought for the purpose, and the electricity they draw while working through sermons for days. In return the texts and their analysis stay in house and go to no outside service.
If you would like to chip in, you can. Buy me a coffee