Write down something you believe about antisemitism and social cohesion in Australia. This page finds where the submissions to the Royal Commission address it — and shows you what they actually say, on all sides of it.
Writing it in the third person matters. “I think the police have gone soft” is about you; “police responses to protest activity have been inadequate” is a claim the corpus can be checked against.
Nothing is hidden. Your proposition is reduced to concepts, each widened with related wording the submissions actually use. Click any word to switch it off; add your own below. The result changes as you do.
Every document below has language close to your proposition somewhere in it. That is all the match means. A submission arguing the opposite of what you wrote will appear here, often at the top, because denying a claim uses the same words as making it. Read the passage, not the ranking.
This is a search over the full text of the submissions the Royal Commission has published, plus the exhibit documents held alongside them — 93 documents and about 949,000 words. That is 81 of the 82 published submissions, one being too short to index, together with 12 documents tendered as exhibits in this Commission’s proceedings. Two further documents sitting in the same exhibit library are held but deliberately not searched here, because they were written for other inquiries entirely — a review of the Online Safety Act and a consultation on artificial intelligence. They are findable through the Commission’s own library. It runs entirely in your browser. Nothing you type is sent anywhere, stored on any server, or seen by anyone, including us.
That holds for what you write as well as what you search. If you weigh a result and note why, the note is kept in your own browser so that it is still there when you come back — it is not transmitted, not collected, and forms no part of the research data. The result of a proposition is for the reader’s eyes only, and so is the reader’s answer to it.
Most public argument about this Royal Commission is conducted in summaries of what “they” supposedly want. The submissions are public. If you believe a position is widely held, this tells you how many of the 93 documents take it up at all — which is frequently a much smaller number than the argument implies.
It also finds the other side of your proposition in the same pass, because it searches for the subject matter rather than for agreement.
This version searches for wording. Your words are widened with related terms the submissions use — so preacher also finds cleric, imam, sermon, preaching — and the full list is shown to you and editable. But a submission that makes your point in vocabulary nobody anticipated will be missed.
Silence here is not evidence of absence. It means these words did not appear, which is a different and much weaker fact.
It has no view about your proposition and cannot acquire one. It will not score, rank or grade any organisation, and a high position in the list means a document uses this language often — not that it is correct, representative, or even serious.
Length is normalised so long submissions are not flattered, but genre is not: a 60-page policy paper and a two-page personal account are not comparable objects.
Results name organisations and, in some submissions, individuals. Everything shown is an extract from that author’s own submission, published by the Royal Commission with their consent. Nothing here characterises what an author meant, and the ordering reflects how much of this language a document uses — nothing more.
A search can still place a document beside a proposition its author would reject. That is a real consequence of making the corpus searchable, and it is not answered by leaving names out — anyone with the same tool can find them. It is answered by the extracts being accurate and by there being a way to fix them. If you are quoted here and the extract is wrong or misattributed, tell us and it will be corrected on this page, with the correction recorded rather than quietly made.
More than twenty thousand submissions were received. Eighty-two have been published, selected by the Commission, released in themed batches and requiring the author’s consent. This is not a sample of what Australians think, and no count taken from it can be stated as one.
It is a picture of the organised advocacy landscape the Commission chose to surface. That is a real and useful thing to be able to see. It is not the public.
One submission also appears twice, because it reaches the corpus by both routes and the two copies are not identical. Both are kept, and the difference between them is recorded rather than tidied away.
A belief stated as a belief cannot be checked — there is nothing in a document that could confirm or deny what you feel. A belief restated as a claim about the world can be: it names something, asserts something about it, and could turn out to be wrong.
The rewrite does something else as well. When a proposition locates a problem in who people are rather than in what is done, this page says so before it runs. That is not a judgement about you — it is one of the twenty-six questions the research instrument asks of every submission in the corpus, turned around and asked of the person using the tool. The submissions get no gentler a test than you do.
Word matching misses a submission that makes your point in wording you did not think of. The fix is to index the corpus by meaning rather than by spelling, so that “radical clerics are radicalising young men” is found by a search for “hate preachers are the problem” even with no word in common.
It will be built so that the same question always returns the same passages, and the index itself will be published — so that anyone who doubts a result can reproduce the retrieval rather than take our word for it.
The research programme reads every submission against twenty-six structural questions. Three of them bear directly on most propositions people arrive with:
Once the coding is done and its reliability tested, those answers will appear beside each result. Then a proposition returns something better than quotations: it returns which step is not available yet.