wai/research: 2030 transport landscape — what the field ratified while we were not looking #80
Loading…
Reference in a new issue
No description provided.
Delete branch "wai/2030-transport-landscape"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
A landscape survey for
wai/research/— not a normative extension, nothing implemented.How it was made. A global sweep across standards bodies (SC 29 WG 2/3/4/5, ITU-T SG21/JVET, IETF AVTCORE/MOQ/SCITT/COSE/GREEN, AMWA, C2PA, Fórum SBTVD/ABNT, Komdigi, WorldDAB) including non-English primary sources, then two adversarial verification lenses — reality (does the designation exist, at the stage and with the content claimed) and bearing (is there something to register, bind or specify against, and is it already in the tree under another name).
71 candidates verified → 19 survived → 52 refuted. The refuted ones are deliberately absent, with their failure modes recorded so they are not re-litigated.
The uncomfortable findings — §3 is the point of the document
determinism-tiers.mdhas a live conformance hole, twice over. H.274nnpfapost-filters change output pixels after a conformant decode; and an object honouring a power-saving hint can still claim byte-exactness. A defect in our own shipped text, no external dependency — ranked #1 for that reason.Energy, stated correctly rather than flatteringly
The field is contested, not empty: ISO/IEC 23001-11 (3rd ed. in force) actuates joules in-band before decode; 23001-19 carries it in ISOBMFF/DASH;
draft-ietf-avtcore-rtcp-green-metadatamakes it a request/commit exchange; IETF GREEN models device energy with a graded accuracy taxonomy. None carries a measured per-object figure bound to a reproducible output under a signature. That remains ours — flanked on both sides, which makes the work reconciliation rather than invention.One item has a deadline
ISO/IEC 23888-6 (Media authenticity and provenance indication) — CfP responses due 13 January 2027 to MPEG 157, Brisbane; then CD/CDAM at 158, DIS at 159, FDIS early 2028. The only square in the survey where our mechanism could enter normative text instead of being registered after someone else's.
Verification status
Five flagged designations resolved; one corrected (there is no confirmable MPEG-AI Part 7); one narrowed (H.274 signatures survive remux, not transcode — they are computed over NAL units). Two flags remain open by design: the NNC signed-data-unit amendment is uncorroborated, and ABNT NBR 25608 is paywalled and was read through public companions only.
Nothing here is a conformance statement about WAI. §9 lists what would falsify each conclusion.
🤖 Generated with Claude Code
A landscape survey for the WAI tree, in the register of the existing research docs. Not a normative extension; nothing here is implemented. Produced by a global sweep across standards bodies (SC 29 WG 2/3/4/5, ITU-T SG21/JVET, IETF AVTCORE/MOQ/SCITT/COSE/GREEN, AMWA, C2PA, Fórum SBTVD/ABNT, Komdigi, WorldDAB) including non-English primary sources, then two adversarial verification lenses: reality (does the designation exist, at the stage and with the content claimed) and bearing (is there something to register, bind or specify against, and is it already in the tree under another name). Seventy-one candidates reached verification; nineteen survived; fifty-two were refuted and are deliberately absent, with their failure modes recorded so they are not re-litigated. The uncomfortable findings, which are the point of the document: * §3.1 "The wire carries instructions" is NOT novel. ISO/IEC 23001-16:2021 ratified derived visual tracks — a track carrying a derivation graph the receiver evaluates — five years ago, inside ISOBMFF. The honest claim is narrower: what Part 16 leaves unspecified is a byte-exact determinism contract per operation and a signed receipt binding the executed graph to measured joules. * §3.2 determinism-tiers.md has a live conformance hole, twice over — H.274 NNPFA post-filters change output pixels after a conformant decode, and an object honouring a power-saving hint can still claim byte-exactness. This is a defect in our own shipped text with no external dependency, and it is ranked first in the queue for that reason. * §3.4 A receipt does not travel with its object. A hard container-scoped binding dies at the first re-encode, which is normal distribution. * §3.6 No transparency log, no key discovery. Every WAI signature presumes the verifier already holds the key. Energy is stated correctly rather than flatteringly: the field is contested, not empty. ISO/IEC 23001-11 (3rd ed. in force) actuates joules in-band before decode; 23001-19 carries it in ISOBMFF/DASH; the AVTCORE RTCP draft makes it a request/commit exchange; IETF GREEN models device energy with a graded accuracy taxonomy. What none of them carries is a measured per-object figure bound to a reproducible output under a signature. That remains our lane — flanked on both sides, which makes the work reconciliation rather than invention. One item has a deadline: ISO/IEC 23888-6 (Media authenticity and provenance indication), CfP responses due 13 January 2027 to MPEG 157. It is the only square in the survey where our mechanism could enter normative text instead of being registered after someone else's. Verification pass recorded in the provenance note: five flagged designations resolved, one corrected (there is no confirmable MPEG-AI Part 7), one narrowed (H.274 signatures survive remux, not transcode — they are computed over NAL units). Two flags remain open by design and are listed in §9.5. Co-Authored-By: Claude Opus 5 <[email protected]>3c2ac875ca584a9bd735