What nine days of cryptography research records show
After nine days, most entries in the research repository end the same way: related precedent, finite checks passed, theorem unverified, and no verified new cryptographic primitive.
Nine days in, the repository holds, counted by files today: about 138 claim records, 251 mathematics run reports, 129 industry and computing reports and 39 natural-science reports, plus the indexes that hold them together. It started on 13 October with a structure and no findings, and nothing in it was claimed as a finding until it had a source.
It says less than I hoped. Most entries end with related precedent and no verified new primitive.
What the entries show
Many results connect to existing cryptography through precedent rather than through the new theorem. Totient-counting results sit next to RSA arithmetic that has existed for decades. A matching result sits next to privacy-preserving matching from 2022. An edit-distance result sits next to private edit-distance comparison from 2003. In each case the older work stands independently, and the new result doesn't create a new scheme.
Finite checks pass; theorems stay unverified. Large numbers of exact finite assertions, in the tens or hundreds of thousands in places, have passed with zero failures. Every one of those entries also says that they are not a proof.
Practical caveats are real. One almost-linear matching algorithm turns out to need parameters so large that its helper bound implies a number with 24 digits in the exponent. An edit-distance result falls back to the exact method for every input size that can physically exist. That is the gap between a theorem and a tool.
No verified new cryptographic primitive. There are entries marked potentially novel application, always narrow and always with low confidence, such as certificate checkers for exact algebraic computations. None is a deployed scheme. The repository's status line says experimental algorithms are not suitable for real-world security without independent review.
Not finding prior work
When I wrote that, as far as I could find, nobody had written specifically about this, I meant it as a search result. The ledger's own rule is that a negative search is not proof of non-use, and I hold myself to that. It looks like uncharted ground, and so far the ground holds nothing that works.
What is still open
The next steps in my notes: getting the original PDFs read page by page, checking proofs where machine-checked ones exist, benchmarking the proposed certificate checkers against existing tools, and having someone who can verify mathematics tell me where I'm wrong.
Sources and notes (8)
- Breuer, Meyer and Wetzel (2022), privacy-preserving maximum matching — the 2022 matching work that predates the new theorem
- Ford, The distribution of totients — earlier totient-counting work
- Lizama-Perez (2026), The Structured Totient Preimage Problem — preprint treating the same collision analysis as cryptography
- Novelty ledger — the notes this was written from
- Computing applications ledger — the notes this was written from
- Industry index — the notes this was written from
- Science opportunities — the notes this was written from
- Run log — the notes this was written from
Cryptography researchAll 10 parts →
