[Rats] Re: Request for adversarial review: Bounded Trust Framework for composed trust decisions

[Nathanael Ritz <[e-mail]>]

Hi Todd,

I have some comments inline below with [NR], but the tldr is that it would
be helpful to know what challenges your work is hoping to resolve:

On Wed, Sep 2, 2026 at 10:11 PM Todd Kirton <[e-mail]> wrote:

> Hello RATS working group,
>
> I am looking for adversarial technical feedback
>

[NR]: Sure.

on a technology-neutral analytical framework I have been developing called
> the CBRS Bounded Trust Framework (BTF).
>

[NR]: What is CBRS?


> BTF grew out of a twelve-paper research series
>

[NR]: Are the papers public? Are they product white papers or academic in
nature?

examining actor provenance, device identity, authentication, decision
> dependencies, behavioral evidence, authority, evidence sufficiency,
> retained state, recovery, uncertainty, and continuous evaluation. The
> framework is an attempt to synthesize those questions into one review model
> for consequential trust decisions.
>

[NR]: That’s a lot all at once and may very well be more than what
individuals may be willing to engage with directly, but maybe the papers
would help.


> I am posting here because RATS provides one of the clearest existing
> architectural separations between Evidence, Verifier appraisal, Attestation
> Results, and Relying Party decision-making. BTF is not intended to replace
> or correct RATS, and I am not claiming novelty over those separations.
>

[NR]: Reading through this email and the shared PDFs, there is a lot of
this hedged phrasing about not looking to “replace” RATS or the like, as
though we might imagine that is a legitimate risk, but not to worry.  In
the case you are using any automated tooling to help put together your
thoughts, I would recommend pruning out excessive “non-goals” and language
like that, and instead describe the motivating problem statement up-front.


> The question I am trying to test is whether a common analytical model
> remains useful when authoritative outputs, local policy, retained state,
> and later decisions cross architectural boundaries and accumulate meaning
> over time.
>

[NR]: For example, what gap did you identify with the work you are doing
that brought you to develop this bounded trust framework? What specific
security concerns are you looking to help mitigate, for example?



> The framework centers on three boundaries:
>
> Semantic: do not conclude more than the evidence and appraisal support.
>
> Authority: do not borrow more authority from an upstream source than that
> source possesses.
>
> Temporal: do not allow a once-justified decision to acquire unlimited
> authority over later actions.
>

[NR]: What systems are you working with the violate any of these? This is
not a question born of personal incredulity; there are plenty of processes
that violate these all the time I’m sure, but again what I think is missing
here is specific examples to properly scrutinize.


> It then asks how decision provenance, root assumptions, policy integrity
> and applicability, disagreement between valid sources, uncertainty,
> persistence, inheritance, change, and recovery should remain visible around
> those boundaries.
>

[NR]: and then after the question is raised, who is responsible for
answering and then acting on that question? And again, in the hopes to
solve what open problem?


> I would especially value adversarial examples from RATS deployments or
> current work where this model becomes redundant, misleading, or simply
> wrong. Multiple-verifier arrangements, transformed or aggregated
> Attestation Results, event-driven attestation, freshness/epoch handling,
> privacy constraints, and downstream authorization seem like particularly
> useful places to attack it.
>

[NR]: Truth be told, I decided to take a stab at such a “adversarial” take
because I’m eager to learn as much about and engage as much as I can about
these specific elements. The challenge presently from reviewing what you
have shared is that I’m not sure how to take the open ended questions and
apply them to any of the above yet.


> The strongest feedback would be something concrete: a counterexample, a
> standards-term correction, a case where one of the three boundaries fails,
> or an example where the framework adds review burden without exposing a
> meaningful architectural issue.
>

[NR]: this is the recurring theme by now, but I think it would be very
helpful to know what problem you are seeking to resolve and then seeing if
that problem is addressed from prior art, or if a framework such as yours
would help us better understand how to orchestrate—say—a multi-verifier
arrangement, for the sake of argument.


> I have kept the first-contact material short:
>
> Standards Engagement Brief — 3 pages
> Standards-Facing Technical Note — 7 pages, including a worked
> attested-workload / production-secret example
>

[NR]: At least on my phone, that 7 page document is filled with very small
text and a lot words! Does that document help answer any of my specific
comments that I may have missed from attempting to skim it over?



> The full framework is available as supporting material if anyone wants to
> go deeper.
>
> I am not looking for endorsement. If the framework is redundant or
> technically unsound, I would rather find that out here before taking it any
> farther.
>
> Thank you for any time you are willing to spend stress-testing it.
>

[NR]: I understand my email was critical without a lot of technical input.
Hopefully the repeated questions provides guidance in refining your
approach in any case.

Cheers,
Nathanael


>
> Todd Kirton
> CoreBond Research Series
>

ps. Ah! That’s “CBRS” stands for.


> _______________________________________________
> RATS mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
>