{"id":"7eb83707-44e2-4caf-8f2b-166166c94f08","arxiv_id":"2606.04892","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"ODYSSEY mitigates execution-inference and execution-replay attacks in confidential blockchains via delegated execution to trustees, with a FISCO BCOS prototype achieving ~4k TPS and 0.4-0.5s latency in a 3-node WAN setup.","lead":"ODYSSEY proposes a delegation model for TEE-based confidential blockchains where clients assign execution to trustees and others only synchronize results to reduce attack surfaces from side-channel and state issues. A generalist might read it to see a practical approach for improving privacy in enterprise blockchains without major performance loss.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"Security rests on unverified assumption that trustees execute without leaks/collusion and result sync creates no new inference channels","rationale":"The reader's weakest_assumption matches the load-bearing point exactly. No additional internal inconsistency was found in the abstract-level argument; the concern is the missing verification of the trust shift, consistent with the UNVERDICTED verdict and low confidence due to abstract-only access.","tokens_in":1693,"tokens_out":285,"duration_ms":129222,"concrete_test":"Locate the security analysis or threat model section; extract any formal game or reduction showing that result synchronization leaks no more than the original TEE execution; if the section contains only informal arguments or no reduction, re-evaluate the 'preserving confidentiality' claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The delegation model claims to reduce attack surface for execution-inference and execution-replay attacks by having only trustees execute in TEEs while others sync results. This holds only if (a) trustees introduce no new confidentiality leaks or collusion vectors and (b) observable results alone cannot enable inference. The abstract provides no formal model, reduction, or analysis showing these conditions; formal_verification is listed as none. The two novel techniques (location-aware concurrent execution, delegation failure handler) are asserted to improve security but without evidence that they close the identified gaps.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper taxonomizes execution-inference and execution-replay attacks exploiting TEE side-channel and state-continuity issues in confidential consortium blockchains. It proposes ODYSSEY, whose core is a delegation model in which clients delegate transaction execution to designated trustees (who run inside TEEs) while all other participants synchronize only the resulting state; two supporting techniques (location-aware concurrent execution and a delegation failure handler) are introduced to improve efficiency and security. A prototype implemented on FISCO BCOS is reported to achieve roughly 4k TPS with 0.4-0.5 s latency in a 3-node WAN setting.","tokens_in":1802,"tokens_out":562,"duration_ms":20950,"significance":"If the security properties can be rigorously established, the delegation approach would constitute a practical architectural change that shrinks the TEE-exposed surface while preserving throughput, which could influence the design of enterprise confidential blockchains. The reported prototype numbers indicate that the performance overhead is modest, but the absence of formal verification, reduction proofs, or comparative security evaluations currently limits the result's immediate impact on the literature.","major_comments":[{"comment":"Abstract and security claims section: the central assertion that the delegation model 'significantly reduces the attack surface' while preserving confidentiality rests on the unanalyzed assumption that designated trustees introduce neither new leaks nor collusion vectors and that synchronizing only results creates no alternative inference channels; no formal model, game-based definition, or reduction is supplied (formal_verification listed as none).","section":"Abstract / Security Claims"},{"comment":"Evaluation section: throughput and latency figures are stated for the 3-node WAN prototype, yet the manuscript supplies neither baseline comparisons against prior TEE-based confidential systems, details of the attack-mitigation test methodology, nor quantitative evidence that the two novel techniques close the taxonomized attack vectors.","section":"Evaluation"},{"comment":"Attack taxonomy and mitigation: while execution-inference and execution-replay attacks are defined, the text asserts without supporting analysis or experiments that location-aware concurrent execution and the delegation failure handler suffice to mitigate them; this gap is load-bearing for the paper's primary contribution.","section":"Attack Taxonomy and Mitigation"}],"minor_comments":[{"comment":"The abstract is dense; expanding the one-sentence description of the two novel techniques would improve readability without altering technical content.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The manuscript would be strengthened by the addition of a dedicated security-analysis section containing at least an informal argument or a sketched reduction; without it the central claim remains largely aspirational."},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the constructive feedback and the recommendation for major revision. We address each major comment below with clarifications on our design choices and indicate where revisions will strengthen the manuscript.","responses":[{"response":"The delegation model limits TEE execution exposure to client-designated trustees rather than all nodes, with only final state results synchronized to other participants. Trustees are selected and trusted by clients, reducing the surface for side-channel and continuity attacks. We agree a more rigorous treatment is needed and will expand the security claims section with an informal argument addressing collusion risks and potential inference from result synchronization. However, the manuscript does not include a formal model or reduction proof, as the primary contribution is the practical delegation architecture.","revision_made":"partial","referee_comment":"[Abstract / Security Claims] Abstract and security claims section: the central assertion that the delegation model 'significantly reduces the attack surface' while preserving confidentiality rests on the unanalyzed assumption that designated trustees introduce neither new leaks nor collusion vectors and that synchronizing only results creates no alternative inference channels; no formal model, game-based definition, or reduction is supplied (formal_verification listed as none)."},{"response":"We will add baseline comparisons against prior TEE-based confidential blockchain systems in the revised evaluation. The current prototype evaluation reports performance in a 3-node WAN setting. Attack mitigation is supported by the architectural design rather than direct quantitative attack experiments. We will include additional details on the test methodology and explain how the techniques address the taxonomized vectors. New quantitative attack-simulation results cannot be added without further experimental work.","revision_made":"partial","referee_comment":"[Evaluation] Evaluation section: throughput and latency figures are stated for the 3-node WAN prototype, yet the manuscript supplies neither baseline comparisons against prior TEE-based confidential systems, details of the attack-mitigation test methodology, nor quantitative evidence that the two novel techniques close the taxonomized attack vectors."},{"response":"We will revise the mitigation section to provide explicit step-by-step reasoning showing how location-aware concurrent execution distributes execution to limit side-channel visibility and how the delegation failure handler maintains state continuity to prevent replays. These explanations will be tied directly to the defined attack vectors with concrete scenarios. The supporting analysis will be expanded, though it remains design-based rather than experimental.","revision_made":"yes","referee_comment":"[Attack Taxonomy and Mitigation] Attack taxonomy and mitigation: while execution-inference and execution-replay attacks are defined, the text asserts without supporting analysis or experiments that location-aware concurrent execution and the delegation failure handler suffice to mitigate them; this gap is load-bearing for the paper's primary contribution."}],"tokens_in":1478,"tokens_out":605,"duration_ms":30882,"standing_objections":["A formal game-based security model, definition, or reduction proof for the delegation approach and its mitigations."]},"desk_editor":{"model":"grok-4.3","letter":"The main takeaway is that this paper introduces a delegation model for TEE-based confidential consortium blockchains: clients hand execution to designated trustees who run inside TEEs, while other nodes only sync the results. This is positioned as a way to shrink the surface for execution-inference and execution-replay attacks that exploit side channels and state continuity.\n\nWhat the paper actually does is lay out a short attack taxonomy, then describe two supporting techniques (location-aware concurrent execution and a delegation failure handler) and ship a prototype on FISCO BCOS. The reported numbers—roughly 4k throughput and 0.4-0.5 s latency in a 3-node WAN setup—are concrete and give a sense of practical cost.\n\nThe soft spot is exactly where the stress-test note points: the security argument depends on trustees not introducing new leaks or collusion risks, and on result synchronization not creating fresh inference channels. The abstract supplies no formal model, reduction, or proof that these conditions hold, and there is no formal verification. The evaluation section appears small-scale with no baseline comparisons or direct attack tests mentioned.\n\nThis work is aimed at engineers and researchers already working on confidential consortium chains who need a practical design sketch rather than a broad theoretical advance. The implementation gives it enough substance to merit referee time, even though the central security claims will need more evidence to be convincing.","headline":"ODYSSEY's delegation model cuts TEE exposure by limiting execution to trustees, but the security rests on unexamined assumptions about trustee behavior and result synchronization with no formal analysis provided.","tokens_in":2283,"tokens_out":359,"would_cite":false,"duration_ms":17357,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"ODYSSEY delegates transaction execution to designated trustees so that other nodes synchronize only results and thereby shrink the TEE attack surface.","keywords":["confidential blockchain","trusted execution environment","delegation model","execution-inference attack","execution-replay attack","side-channel mitigation","consortium blockchain","FISCO BCOS"],"falsifier":"A demonstration that an adversary can still recover a transaction’s private inputs by observing only the synchronized results and the public ledger state after a single execution round.","tokens_in":2599,"feed_emoji":"🔒","tokens_out":644,"duration_ms":17931,"temperature":0.7,"pith_summary":"The paper identifies execution-inference and execution-replay attacks that exploit long-lived side channels and state-continuity problems inside TEEs used by consortium blockchains. It proposes a delegation model in which each client assigns execution to its own trustees while every other participant receives only the final results. This change limits the number of nodes that must run sensitive code inside the enclave. The authors add location-aware concurrent execution and a delegation-failure handler to keep throughput and latency acceptable. A prototype built on FISCO BCOS reaches roughly 4 000 transactions per second with 0.4–0.5 s latency in a three-node WAN setting.","feed_headline":"Delegated execution shrinks TEE attack surface in confidential chains","feed_subtitle":"Clients assign work to trustees; everyone else sees only results, cutting exposure while preserving 4k tx/s throughput.","key_machinery":"The delegation model, in which each client chooses its own trustees to execute inside the TEE and every other node receives only the resulting state updates.","core_discovery":"The delegation model lets clients assign transaction execution to designated trustees while all other participants synchronize only the execution results; this arrangement measurably reduces the attack surface for execution-inference and execution-replay attacks without sacrificing confidentiality or system performance.","pith_inferences":["Trust assumptions shift from the entire validator set to a per-client trustee subset, which may change how consortium governance is structured.","If trustees are drawn from the same organizations that already operate nodes, the model may require additional audit or rotation mechanisms not described in the paper.","The approach could be tested on other TEE platforms or in permissionless settings to measure whether the reduced executor set still suffices for liveness."],"forward_implications":["Only the chosen trustees ever run the confidential transaction logic inside the enclave.","Side-channel and state-continuity attacks are confined to the smaller trustee set.","Throughput and latency remain comparable to existing TEE-based systems when location-aware concurrency and failure handling are used.","The same result-synchronization pattern can be applied to any consortium blockchain that already relies on TEEs."],"fun_headline_variants":["ODYSSEY delegates execution to shrink TEE attack surface","Trustees execute txs reducing TEE side channel attacks in ODYSSEY","Delegation model limits inference and replay attacks on TEEs","ODYSSEY cuts attack surface with trustee based tx execution"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"Designated trustees will execute transactions without introducing new confidentiality leaks or collusion risks, and synchronizing only results will not open alternative inference channels.","fun_headline_variants_meta":{"raw":{"variants":["ODYSSEY delegates execution to shrink TEE attack surface","Trustees execute txs reducing TEE side channel attacks in ODYSSEY","Delegation model limits inference and replay attacks on TEEs","ODYSSEY cuts attack surface with trustee based tx execution"]},"model":"grok-4.3","cost_usd":0.004561,"raw_usage":{"total_tokens":2249,"prompt_tokens":634,"num_sources_used":0,"completion_tokens":69,"cost_in_usd_ticks":45612000,"prompt_tokens_details":{"text_tokens":634,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":1546,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":634,"tokens_out":69,"duration_ms":16763,"temperature":1.0,"reasoning_tokens":1546,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-28T05:46:59.906856+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"A demonstration that an adversary can still recover a transaction’s private inputs by observing only the synchronized results and the public ledger state after a single execution round.","supporting_citations":[],"review_version":1}