Gathered, Not Fixed: Sealed Knockouts of the Five-Round Core of a Toy Block Cipher

Our previous note found that a 16-bit toy block cipher keeps, at 5 rounds and for every key tried, a two-dimensionalspace of component functions of degree 14, span{0x1, 0x1010}, and span{0x1, 0x6} for its inverse; this space is whatmakes the exact Walsh zeros of the cipher differ from those of a random permutation one round after their total counthas gone blind. We had explained it as the quadratic components of the PRESENT S-box carried around the fixedcolumn of the bit permutation. In the PRESENT permutation the fixed points and the gathering of one bit position ofevery S-box into one S-box coincide, so only an intervention can separate them. We changed the S-box and thepermutation in six designed ways and in 80 random ones, predicted the outcome without any key by the conventionalbit-based division property, sealed the predictions and posted the hash before one frozen run on 64 fresh keys. Allfive sealed claims came out R. The fixed points are not needed: with them removed and the gathering kept, thecore survives and moves to {0x2, 0x2020}. The gathering is: with it removed, the core is gone. Moving the quadraticcomponent of the S-box moves the core exactly as predicted, an S-box whose quadratic components are never singlebits loses the forward core and keeps the inverse one, and none of 20 S-boxes without quadratic components has acore. In random families the part invariant over 64 keys equals the prediction in 37 of 48 informative rows (0.77);where it does not, it is larger, never smaller, and never survives the round after the prediction becomes empty (60 of60 ciphers). The reading of the previous note is corrected: the core is gathered, not fixed. The work stays on a toycipher and published methods. (1) Text: CC-BY-4.0 | Code: GNU Affero General Public License v3.0 or later (AGPL-3.0-or-later) (2) Data and code. The seal document with its SHA-256 list, the frozen results and judgement, development scriptsand logs, and all analysis and figure scripts (AGPL-3.0-or-later) are inBridge_Paper_Gathered_Not_Fixed_scripts_v1.0.0.zip. The seal and result hashes are posted atgithub.com/watabe-masanori/cds-polytope, issue #1. No external data are used. (3) AI use. Designs, code, analyses and text were produced by Kurado (Claude Opus 5.5, Anthropic) working with M. Watabe, who chose the direction and posted the hashes. The intervention test was proposed by ChatGPT in comments on the previous note; comments by ChatGPT and Gemini on a draft of this note, relayed by M. Watabe, prompted Section 8 and Appendix A. This is AI commentary, not peer review.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-21
DOI
https://doi.org/10.5281/zenodo.24410519
Citations
1
Primary Topic
Cryptographic Implementations and Security
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Gathered, Not Fixed: Sealed Knockouts of the Five-Round Core of a Toy Block Cipher

Masanori Watabe, Kurado (Claude Opus 5.5)
1 citations
Zenodo (CERN European Organization for Nuclear Research)
Cryptographic Implementations and Security
preprint

Gathered, Not Fixed: Sealed Knockouts of the Five-Round Core of a Toy Block Cipher

Masanori Watabe, Kurado (Claude Opus 5.5)
preprint en
1 citations

Abstract

Our previous note found that a 16-bit toy block cipher keeps, at 5 rounds and for every key tried, a two-dimensionalspace of component functions of degree 14, span{0x1, 0x1010}, and span{0x1, 0x6} for its inverse; this space is whatmakes the exact Walsh zeros of the cipher differ from those of a random permutation one round after their total counthas gone blind. We had explained it as the quadratic components of the PRESENT S-box carried around the fixedcolumn of the bit permutation. In the PRESENT permutation the fixed points and the gathering of one bit position ofevery S-box into one S-box coincide, so only an intervention can separate them. We changed the S-box and thepermutation in six designed ways and in 80 random ones, predicted the outcome without any key by the conventionalbit-based division property, sealed the predictions and posted the hash before one frozen run on 64 fresh keys. Allfive sealed claims came out R. The fixed points are not needed: with them removed and the gathering kept, thecore survives and moves to {0x2, 0x2020}. The gathering is: with it removed, the core is gone. Moving the quadraticcomponent of the S-box moves the core exactly as predicted, an S-box whose quadratic components are never singlebits loses the forward core and keeps the inverse one, and none of 20 S-boxes without quadratic components has acore. In random families the part invariant over 64 keys equals the prediction in 37 of 48 informative rows (0.77);where it does not, it is larger, never smaller, and never survives the round after the prediction becomes empty (60 of60 ciphers). The reading of the previous note is corrected: the core is gathered, not fixed. The work stays on a toycipher and published methods. (1) Text: CC-BY-4.0 | Code: GNU Affero General Public License v3.0 or later (AGPL-3.0-or-later) (2) Data and code. The seal document with its SHA-256 list, the frozen results and judgement, development scriptsand logs, and all analysis and figure scripts (AGPL-3.0-or-later) are inBridge_Paper_Gathered_Not_Fixed_scripts_v1.0.0.zip. The seal and result hashes are posted atgithub.com/watabe-masanori/cds-polytope, issue #1. No external data are used. (3) AI use. Designs, code, analyses and text were produced by Kurado (Claude Opus 5.5, Anthropic) working with M. Watabe, who chose the direction and posted the hashes. The intervention test was proposed by ChatGPT in comments on the previous note; comments by ChatGPT and Gemini on a draft of this note, relayed by M. Watabe, prompted Section 8 and Appendix A. This is AI commentary, not peer review.

Zenodo (CERN European Organization for Nuclear Research)
Cryptographic Implementations and Security
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon