The strongest arguments against this specification, written as their proponents would write them, each with our response or an admission that we do not have one.
This file is maintained by the authors. A draft with one implementation cannot manufacture consensus, but it can argue against itself in public, which is the one thing a real consensus process does that a single author can partly substitute for.
Adding an objection is a pull request. We would rather host the argument than have it made elsewhere.
A specification with a single implementer is a configuration file with an ambitious README. Nothing here has been tested between two independent parties, so none of the ambiguities that only appear at an interface have been found yet. Every interface specification that was written before two implementations existed had to be substantially rewritten afterwards.
Response: accepted, without qualification. This is the strongest objection and we do
not have an answer to it. It is stated in spec/00-status.md in the same words. The
mitigation is that 0.1 is published as a Request for Comments rather than a release, and
that the conformance framework (§8) is built to make a partial or sceptical
implementation expressible rather than forcing a binary claim.
Status: open. Closes only when a second independent implementation exists.
Updated 2026-08-27. A reference implementation of the interlock now exists
(reference/), along with signed conformance vectors in conformance/vectors/. That
does not narrow this objection, because the reference implementation is the author’s
too. What changed is that a second implementer can now check their work against this
specification without talking to us, so the thing that would close this objection got
cheaper for someone else to do.
5 seconds, 1 Hz, 0.1 W/m², 250 ms, ±0.2°. None of these came from hardware. They came from arithmetic performed by people who have not built a power beaming system, and several of them will be wrong for any real one.
Response: substantially accepted. The distinction we draw is between the shape of a requirement and its value. The shape — a dead-man token whose lifetime bounds a keep-out buffer — is what we are actually proposing. The value is a placeholder and we have tried to make that structurally obvious rather than merely admitting it:
profiles/*.json, not in a normative sentence;Where we are least confident: the probe level ratio (20 dB is a guess about loop SNR) and
abortToSafe_ms for tube-based transmitters.
Status: open, structurally mitigated.
No standards body has identified a need. No operator has requested it. Writing a specification for an industry that has not asked for one is a way of appearing important, and the absence of a work item is evidence there is no demand, not evidence of a gap.
Response: partly rejected, partly accepted. Rejected on the facts: the absence of a
work item at ITU-R, IEEE, CCSDS and every regulator surveyed is documented in
gap-analysis.md with dates and sources, and it coexists with at least seven funded
programmes across two incompatible physical layers and one 2027 flight carrying both.
That is the condition in which interfaces get frozen by accident, bilaterally, and
become expensive to unfreeze.
Accepted on the implication: nobody did ask, and this document has no mandate. That is
why spec/00-status.md contains a standing offer to be superseded by any body that opens
a work item, and why we would regard being superseded as success rather than failure.
Status: open.
The geometric corridor check is the second half of the anti-hijack construction, but at GEO a 2 km aperture subtends 0.0032°. Any corridor wide enough to be implementable is vastly wider than the pointing requirement, so the check catches only gross failures and the closed pointing loop is doing all the real work. The two-of-two claim is therefore much weaker at GEO than the text implies.
Response: accepted. The arithmetic is correct and the limitation is stated in §5.6 rather than buried. We do not have a fix. A GEO system probably needs a third, independent check that this document does not specify — plausibly an independently-surveyed ground-side measurement of where the beam actually is, reported back over the interface. That would be a real addition to a future version and we would welcome a proposal.
Status: open. No mitigation offered.
ICNIRP 2020 changes the local-exposure basis at 6 GHz, from absorbed power density above it to specific absorption rate below. The design frequency used throughout the examples is 5.8 GHz, which is below that boundary. A power-density screening therefore does not address local exposure at all, and a specification that presents power-density screening as the exposure treatment is misleading about what has been assessed.
Response: accepted. [R-A-013] requires localExposureBasis to be declared, and
not_assessed is an allowed and honest value. That makes the gap visible; it does not
close it. Closing it requires an RF safety engineer, which this project does not have.
Status: open. Contributions specifically wanted here.
[R-C-022] requires p99 command-path latency below one third of the token lifetime. Many operators do not have continuous command coverage — they have contacts. A specification that assumes a continuous path has assumed away the ground segment economics that make the business hard.
Response: accepted as a real constraint, rejected as a defect. It is a genuine
consequence of the design, and it is the correct one: a beam that cannot be stopped
promptly should not be radiating. [R-C-021] therefore recommends a co-located issuer with
a direct path, which converts the problem into one-way light time. An operator without
that cannot deliver on STANDARD, and we think that is the right answer rather than a
gap to be softened.
Status: closed as intended behaviour. Reopen with a counter-example.
[R-M-012] names IEC 62053-22 classes 0.2S/0.5S and ANSI C12.20 classes 0.2/0.5. These were not read from the published documents. A specification that elsewhere refuses to carry a second-hand table has carried one here.
Response: accepted, and now resolved — with two errors found and a third thing learned. Verified 2026-08-23 against the issuing bodies’ own catalogue entries:
Not closed: several secondary sources report that ANSI C12.20 was merged into ANSI C12.1 in 2022 and is now historical. That was not confirmed first-party — the NEMA and ANSI pages carrying it returned 403 — so the specification continues to name C12.20 without asserting its status. Anyone with catalogue access is invited to settle it.
Status: closed as to the class designations. The C12.20 currency question is open.
The information model uses
receivingSegmentto accommodate space-to-space delivery, but every worked example, the keep-out volume, the intruder budget and the entire aviation treatment assume a ground receiver. The neutrality is cosmetic.
Response: accepted. v0.1 is scoped to space-to-ground and says so. The naming
reserves the slot at zero cost; the RELAXED timing profile is the only real
accommodation. A space-to-space profile needs its own hazard analysis — the intruder is
not a person, the keep-out volume is an orbital volume, and the abort budget is set by
relative velocity rather than by a jet.
Status: open, deliberately deferred.
Annex OPT exists, Annex RF exists, and the core talks about beams without saying which kind. An engineer who builds RF hardware can read this and conclude it was written for optical systems and that its problems are not theirs.
Response: accepted as a writing defect, not a content one. Annex RF is normative,
Annex OPT is a stub that forbids claiming it at 0.1, and no wavelength or photovoltaic
assumption appears in the normative text. But this objection came from an RF engineer
who read the document and concluded otherwise, which makes it real regardless of what
the text technically says. Recorded as DISPOSITIONS.md C-1.
The likely cause is that the physical-layer-agnostic core reads as unspecified rather than as deliberately shared, and the reader supplies the missing PHY themselves. Candidate fix for 0.2: state in §1.1 which annex is normative today, and say plainly in the core that it is shared by construction rather than vague by omission.
Status: open. Fix proposed, not written.
gap-analysis.mdproves no standards body covers beamed power. It does not establish that no methodology covers it. Those are different claims, and a specification that invents a rule where a published textbook already has one has reinvented rather than profiled — which is the failure its own §1.4 is written to prevent.
Response: accepted. Found by the first external reader, who pointed at Power Beaming: History, Theory, and Practice (Jaffe, Nugent, Strassner II and Szazynski, World Scientific, 2024) in response to §6’s efficiency requirement. The book sets out a subsystem decomposition whose product gives the end-to-end figure, and its stated purpose is establishing common nomenclature for assessing power beaming systems.
The survey was scoped to publishers of standards — ITU-R, IEEE, CCSDS, ICNIRP, regulators — because the question asked was “is this standardised?”. The question that should also have been asked is “is there already an accepted way of doing this?”. The answer for efficiency was yes, and nobody inside the project noticed.
The finding does not weaken the gap analysis’s own claim: there is still no standard. It weakens the inference drawn from it, that the field has no conventions worth profiling.
Fix for 0.2: add a literature row to gap-analysis.md for each part, distinct from the
standards rows, and rewrite §6 to profile the published decomposition.
Status: open. Fix identified, blocked on reading the source.