[Agentproto] Re: Delegation and evidence: a concrete candidate for the charter discussion

[Bradley B <[e-mail]>]

Doug,

On the third property. It is statable without naming a mechanism, and the
statement has a consequence you will want in -01.

The property: the ordering must derive from a constraint that was not under the
sole control of a party whose conduct or assertion is material to the claim, at
the time the binding was fixed. The test is not whether the ordering is stated.
It is whether it could have been stated otherwise.

That is mechanism independent, and your candidates all satisfy it. But they do
not satisfy it in the same way, and the difference is structural rather than a
matter of implementation quality. There are two routes.

Causal. The effect consumes an artifact that did not exist before the decision.
If the effecting side cannot act without a value minted at decision time, the
ordering follows from causality, and a relying party verifies it later from the
two records alone. Neither side needs to trust the other's clock. In a
fail-closed plane this is native, because the effect cannot be reached except
through the decision.

Temporal. Both sides anchor independently. An anchored decision record
establishes that the decision existed no later than its anchor, which is the
easy half. The lower bound on the effect is self asserted unless the effecting
substrate also anchors at effect time, and a self asserted effect time is the
same defect DET-2 excludes elsewhere, arriving one level down.

Across administrative domains the temporal route is honest only where both
sides anchor, which is a deployment property rather than a requirements
property. Worth saying plainly in the document, because a profile that claims
the temporal route and anchors one side has not met DET-2.

Two riders, and I think DET-2 needs both regardless of which route a profile
takes.

Single use. Without it, one decision establishes ordering for an unbounded
number of effects, each of which individually passes the test. In a fail-closed
plane single use may hold by construction, but that is a property of the
deployment, not of the requirement, and a model that carries the binding
without stating consumption semantics will be implemented both ways.

Order is not duration. The binding establishes that the effect could not have
preceded the decision. It says nothing about how long after. A decision taken
long ago whose binding is still live passes DET-2 exactly as cleanly as a fresh
one. As written, DET-2 is satisfiable by a stale authority. The binding needs
to carry, or its declaration needs to state, the interval within which it
remains usable. Ordering without a bound is what a stale authority looks like
from the inside.

And the reporting consequence, in your existing DET-2 form:

Where the evaluator establishes that a binding exists but cannot establish
that it could not have been produced after T by a party whose conduct or
assertion is material, the evaluator MUST NOT report the ordering as
determined. It MAY report the binding as asserted, identified as such.

The reason to require the explicit weaker result rather than silence is that
without it implementations choose between overclaiming and not reporting, and
deployed systems choose overclaiming.

This is Section 6 of draft-bradleyb-audit-decision-records-00, where I call the
underlying distinction correspondence versus precedence. Reconciling the
Authorizing the side's record and the effecting side's record hash for hash
establishes that the two sides agree about what happened. It establishes
nothing about which came first. Take whatever of it is useful for -01. I can
send proposed text for both insertions off-list if that is easier to work with.

Nicholas, on absent records being ambiguous. That is the entry emission rule in
Section 5 of the same draft. A decision exists whether or not the action
follows, so a denial, an expiry, or a commitment that was granted and never
honoured stays representable. Your sequence and that rule are approaching the
same failure from different ends, and I think they compose.

IPR disclosure covering the material behind this:

https://datatracker.ietf.org/ipr/7500

Bradley B
11 AI Blockchain Development LLC


On Wed, Sep 9, 2026 at 1:13 PM Sumit Ahuja
<sumit=[e-mail]> wrote:
>
> Doug,
>
> Agreed.
>
> No comparison relation is a property of the two scopes and the governing profile. It survives without provenance because the inputs are the evidence.
>
> Cannot decide is a property of the evaluator at a moment. It says a particular verifier lacked a particular input. Six months later, with no record of which verifier, under which profile, missing what, the result is unfalsifiable.
>
> So the two need different treatment in the record, not just different tokens. The first needs the scopes and the profile identifier. The second needs the evaluator, the governing profile at evaluation time, and the specific missing input, or it decays into an assertion.
>
> However, a result that cannot be re-derived and cannot be falsified is not evidence, it is a report. That does not make it useless, but a relying party should be aware of such.
>
> Iman has just settled the vocabulary on the other list. NO_COMPARISON_RELATION means no rule under the governing profile, and PROFILE_NOT_HELD identifies a missing input that another verifier might supply. That second reading depends on naming the input, which is your point.
>
> Happy to read the -00 whenever you want eyes on it.
>
> Sumit P. Ahuja
> Main Labs
> _______________________________________________
> Agentproto mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]



-- 

Bradley B Founder and Architect

11/11 AI | 11 AI & Blockchain Development LLC [e-mail]

UEI GHEZH441JN51 | CAGE 1A1G9 | DUNS 144921555 Public proof:
https://control.11aiblockchain.com/proof

📧 [e-mail]
🌐 https://1111ai.com