Full Provenance on the COSS Causal Axis: An Event-Level Closure of Causal Attribution and Confused-Deputy Exclusion

summary

The gist

This paper introduces the COSS model as a generalization of PIC by treating execution occurrences as primitive causal objects and extending authority to a product behavior space…

In short

The COSS model generalizes PIC by treating execution occurrences as primitive causal objects and extending authority to a product behavior space. It proves a universal theorem showing that no lineage-form confused-deputy effect is admissible along ordinary continuation in systems satisfying specific causal completeness and continuity conditions.

Key concepts

COSS Model
A model that treats execution occurrences as primitive causal objects and extends authority to a product behavior space. This allows an occurrence to expose authority, flow, provenance, identity transition, and geography simultaneously without needing a second temporal axis.
Confused-Deputy Effect Exclusion
The core finding is that for any system with causal completeness and ordinary PIC continuity, no lineage-form confused-deputy effect is admissible along ordinary continuation. This means that once causes and protected effects are within the causal semantics, such an effect cannot occur.
Ordinary Continuation
A concept used to define the path of a security system's authority over time. The paper proves that under certain conditions, this continuation identity is stronger than common ancestry, meaning distinct lineages are distinct causal histories even if they share a root.
Product Security State Continuity
A theorem stating that if a lineage has product security-state continuity, then the sequence of states must be non-increasing. This implies that authority cannot introduce a possibility forbidden by the origin state of any carried security system.

Terminology used across episodes

This episode discusses

The paper

Full Provenance on the COSS Causal Axis: An Event-Level Closure of Causal Attribution and Confused-Deputy Exclusion · Read on arXiv

Nicola Gallo

Provenance Identity Continuity (PIC) represents an authority-use event as an operation/resource behavior situated in a causal lineage, (o,r,) in O times R times L. We make the causal occurrence itself primitive: a represented security-relevant causal graph (E, to c) in which each occurrence has a unique causal history, so the consistent event-lineage space is canonically isomorphic to E. The modeled execution has one causal partial order; distinct lineages are distinct causal histories inside it and may share a root while diverging into incomparable branches. PIC's O times R times L is the authority-behavior projection of this space, and Causally Ordered Security-State (COSS) continuity generalizes it to an arbitrary behavior product over the same causal execution. On this semantics we prove an architecture-independent confused-deputy exclusion theorem. PIC continuation requires C(e') C(e) on every ordinary causal edge, so descendants are bounded by their ancestors; at an ordinary join, authority available on only one incoming branch cannot be resurrected, even when branches share a common root. Non-conservative combination must be an explicit authorized composition or relaxation event. With protected-effect mediation, an effect caused by request q through an ordinary authority segment can exercise only authority in C(q), so a confused-deputy state is contradictory. The result holds for any system whose protected request-to-effect causes are faithfully represented and whose ordinary authority transport satisfies PIC. We revisit Rajani, Garg, and Rezk: their value, control-flow, and initial-heap attacks are different witnesses of causal ancestry, so full provenance is an evidence mechanism for recovering edges of the causal graph, not the foundation of the PIC theorem. Finally we lift it to the COSS product of heterogeneous security-state coordinates.

DOI: 10.5281/zenodo.23102373

Transcript

Introduction to the show: ident: Security Radio. Generated commentary on the latest security and cryptography papers.

Nadia: We're looking at the paper: "Full Provenance on the COSS Causal Axis".

Elias: Here's the gist of the paper. This paper introduces the COSS model as a generalization of PIC by treating execution occurrences as primitive causal objects and extending authority to a product behavior space, leading to an architecture-independent theorem excluding confused-deputy effects through ordinary continuation Pruning redundant reasoning tokens The core finding is that for any system satisfying causal completeness, protected-effect mediation, and ordinary PIC authority continuity, no PIC lineage-form confused-deputy effect is admissible along ordinary continuation This universal theorem quantifies over any system whose protected execution has a faithful CDA-complete causal event representation and enforces PIC on ordinary causal continuation The theorem shows that an accepted effect realizing (o, r) must have (o, r) ∈ C(q), which contradicts the requirement for a lineage-form confused deputy state where p /∈ C(q) How the training signal is built The model replaces the privilege tuple as the primitive event by a causal occurrence e ∈ E, and gives the consistent representation Eb ⊆ E ×L from which PIC’s O ×R×L is obtained by authority-behavior projection The behavior is generalized from O × R to B, where B is a typed product of behavior spaces Bjj∈J, giving the event-level basis of COSS without introducing a second temporal axis This allows an occurrence to expose authority, flow, provenance, identity transition, and geography simultaneously Where the gains come from The paper establishes that common ancestry is weaker than continuation identity because distinct lineages are distinct causal histories inside that order and may share the same root while diverging into incomparable branches Proposition 3 proves that common ancestry does not identify causal histories, meaning same root ≠ same causal continuation The theorem is independent of whether the intermediate cause was called explicit designation, implicit designation, value influence, control flow, heap influence, or asynchronous state transfer Relational PIC and Product Continuity The COSS model extends ordinary PIC authority continuity to causal DAGs by proving ancestor-bounded authority on a causal DAG Proposition 2 states that if e is an ordinary causal join, then C(e) ⊆ T p∈Pred(e) C(p), which implies common ancestry cannot restore a privilege that is absent from one of the causal branches participating in the join The product security state continuity theorem proves that if a lineage has product security-state continuity, then Σn ⪯ Σn−1 ⪯ · · · ⪯ Σ0, meaning authority cannot introduce a possibility forbidden by the origin state of any carried security system Heterogeneous Security-State Spaces COSS defines a security-state coordinate Sj = (Xj, ⪯j, Bj, J·Kj), which is semantically monotone when x′ ⪯j x ⇒ Jx′Kj ⊆ JxK The product order (58) ensures that the product state is well-defined as a function of the endpoint occurrence of its represented history Corollary 5 shows that a decision rule cannot enforce a distinction that its observation function erases, which is the general form of lineage-forgetting blindness COSS carries multiple security coordinates on the same causal structure, allowing for distinctions between authority and information flow The final result is that every represented PIC lineage-form CDA instance on an ordinary request/effect segment is incompatible with ordinary PIC continuity This means that once the relevant causes and protected effects are inside the causal semantics, no PIC lineage-form confused-deputy instance is an admissible ordinary PIC continuation The theorem is universal over systems satisfying that representation and enforcement contract, stating that no physical cause omitted from the event model is covered by the theorem References

Paper summary: Nadia: So we've talked about how this paper establishes this universal theorem excluding those confusing deputy effects through ordinary continuation by treating execution occurrences as primitive causal objects and extending authority to a product behavior space. It’s all about closing the loop on causal attribution at the event level.

Elias: I think what’s really interesting here is how they generalize the training signal by replacing the privilege tuple with a causal occurrence e, which lets them build this event-level basis for COSS without needing a second temporal axis. It’s making things much more expressive about flow and provenance simultaneously.

Priya: From my view, the implication for measurement research is that they show how to define security states in a way that respects the causal structure of the data itself, especially with those heterogeneous security-state spaces. The definition of Sj as (X j, sigma j, B j, J K j) being semantically monotone is crucial for ensuring those states are well-defined functions of the endpoint occurrence.

Nadia: That monotonicity means that a decision rule can't enforce a difference between two things if it’s just observing the endpoint occurrence of their history. It also proves that common ancestry doesn't identify causal histories, which is important because distinct lineages are distinct causal histories even if they share the same root and diverge later.

Elias: And that divergence is what allows them to prove ancestor-bounded authority on a causal DAG, showing that ordinary continuation cannot restore a privilege just because the parents have a common ancestor or meet in one process. That’s where they exclude those confused deputy effects in the context of ordinary continuation.

Priya: It really changes how we think about causality in execution tracing; instead of relying on wall-clock distance, you have to respect this branch-sensitive causal structure. The paper essentially provides a way to handle complex branching and concurrency within a single framework by focusing on the event DAG.

Nadia: So looking at the full title, "Full Provenance on the COSS Causal Axis: An Event-Level Closure of Causal Attribution and Confused-Deputy Exclusion," it sounds like they are providing a complete mathematical picture for how authority flows through time when you consider every execution instance as a core causal event.

Elias: It’s about formalizing the relationship between an operation, its resource, and the causal lineage in a way that is architecturally independent. They show how to use this framework to rigorously exclude those tricky confused deputy scenarios when dealing with ordinary continuation on any system satisfying their basic contract.

Conclusion: Nadia: So we've seen how they built this COSS model to treat every single execution event as a primitive cause, and how that leads to this big theorem about excluding those confused deputy effects through ordinary continuation.

Elias: Yeah, focusing on that specific result, the core finding is that for any system meeting certain rules about causal completeness and authority continuity, you can't actually have a lineage-form confused deputy effect along ordinary continuation.

Priya: What that means practically is they’ve quantified this over any system where the protected execution has a faithful causal event representation and enforces PIC on ordinary causal continuation.

Nadia: So if you accept an effect realizing a specific pair, say an operation and a result, that pair has to be in the set C of q.

Elias: That contradicts what’s needed for a lineage-form confused deputy state where that operation is not in C of q. It shows that common ancestry isn't enough to restore a privilege if it wasn't present on one of the branches participating in the join.

Priya: They also proved something about product security state continuity, showing that if a lineage has this continuity, then the sum over all steps from zero up to n is equivalent to just the step before it.

Nadia: That’s telling because it means authority can't introduce a possibility that was forbidden by the starting state of any carried security system.

Elias: They also established that common ancestry is weaker than continuation identity, meaning distinct lineages are distinct causal histories even if they share the same root and diverge later.

Priya: And this whole COSS structure, with its security-state coordinates defined as a tuple involving the product order, is semantically monotone when you look at how information flows through it.

Nadia: So in short, this paper shows that every represented instance of a lineage-form CDA on an ordinary request effect segment is incompatible with ordinary PIC continuity.

Elias: It means once the causes and protected effects are inside the causal semantics, no confused deputy instance is admissible as an ordinary PIC continuation.

Priya: This universal theorem holds for any system that follows those representation and enforcement rules, even if you omit some physical cause from the event model.

Nadia: It really tightens up how we think about authority flow in complex systems by grounding it firmly in a causal structure rather than just timing.

More episodes

← Home