{"id":"346afb03-9d2a-4993-a7a9-f0ab23acd68a","arxiv_id":"2502.10321","paper_version":1,"verdict":"REJECT","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":5,"one_line_summary":"A proposed optimistic rollup mechanism replaces the long challenge window with a short, randomly sampled approval quorum and exponential timeouts, but the security argument for keeping the single-honest-node guarantee is not established.","lead":"This paper proposes a way for optimistic rollups, systems that assume transactions are valid unless challenged, to settle state changes in under a second instead of waiting a week. It does this by requiring a set of randomly chosen verifier nodes to approve each change, and extending the wait when not enough verifiers sign off.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The c0-approval rule replaces the single-honest-node guarantee with an unproven random-sampling assumption; Section 4.2 computes challenge probability, not the probability that a fraudulent diff is approved by c0 malicious signers.","rationale":"The reader's weakest_assumption identifies exactly the same load-bearing gap: the random approval quorum is never shown to contain an honest party, so the single-honest-node assumption is not preserved. My independent reading confirms this and sharpens it with a concrete counting argument: for realistic N and c0, a single honest node is almost certainly absent from the sample, so c0 malicious signers can approve fraud in fast mode. The paper's Section 4.2 formula does not address this because it computes the probability that at least one challenge is raised, not the probability that a fraudulent commitment obtains the required c0 approvals before any challenge. Since this unsupported assumption is the foundation of the sub-second finality claim, the reader's REJECT verdict remains appropriate; no adjustment is needed.","tokens_in":7523,"tokens_out":7479,"duration_ms":77457,"concrete_test":"Compute the fast-finality fraud probability from first principles: assume N active challengers, exactly one honest node, m=N-1 malicious nodes, and c0 sampled uniformly without replacement; then P(all c0 sampled are malicious) = C(m,c0)/C(N,c0). Plug in N=10^4 and c0=100 to verify the result is about 0.990, and N=10^6, c0=100 to see it remains near 0.9999. If this probability is non-negligible, or if the paper cannot specify the sampling distribution and adversary's selection power, the claim that a single honest node secures sub-second finality fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Load-bearing premise: Section 3 step 4a and Section 4.1 make fast finality conditional on c0 randomly selected active challengers signing off without challenge. For this to preserve the paper's stated single-honest-node assumption (Abstract and Section 3), the selection must guarantee with overwhelming probability that at least one honest, active verifier is in the sample. No such analysis is provided. Section 4.2's P(E)=P(F)*P(D|F)*P(T)*(1-(1-P(Ci))^N) is the marginal probability that at least one of N nodes raises a challenge; it is not the conditional probability that a fraudulent diff is detected and rejected before c0 approvals, and Example 4.2.1 silently sets P(T)=1 despite the target t0=500ms. Quantitatively, if there is exactly one honest active challenger among N=10^4 and the protocol samples c0=100 uniformly, the sample misses the honest node with probability C(9999,100)/C(10000,100) ~ (9999/10000)^100 ~ 0.990; then 100 colluding malicious signers can approve an invalid diff. Thus the mechanism actually assumes at least one honest node in every approval sample, actively monitoring at sub-second speed, which is strictly stronger than the traditional any-honest-party-can-challenge model. The bonding/slashing discussion in Section 4.2.2 is asserted, not derived, to make this probability negligible.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents \"Dynamic Fraud Proof\" (DFP), a modification of optimistic fraud-proof mechanisms intended for Ephemeral Rollups on Solana. The core proposal is to let a state commitment be finalized at the end of a very short challenge window (t0 = 500ms in the running example) if a configurable number c0 of randomly selected \"active challengers\" sign off without raising a challenge; if the threshold is not met, the challenge window grows exponentially (t_n = t0 * rt^n) while the required number of signers decays (c_n = c0 * rc^n), eventually falling back to a traditional long challenge window with no active-challenger requirement. The paper claims this preserves the standard single-honest-node assumption of fraud proofs while enabling sub-second finality, and it presents a probabilistic model in Section 4.2 and an economic-incentive discussion in Section 4.2.2.","tokens_in":7828,"tokens_out":4981,"duration_ms":46311,"significance":"The goal of shortening optimistic-rollup finality from days to seconds with a dynamic, adversarial-awareness mechanism is a relevant research problem, and the paper contains an interesting design intuition (inverting the challenge process by requiring random sign-offs). The paper's central security claim, however, is not supported. The analysis in Section 4.2 computes the marginal probability that at least one node raises a challenge, which is a different quantity from the probability that a fraudulent commitment is denied approval by the randomly selected quorum. As a result, the paper does not establish that the mechanism preserves the single-honest-node assumption; in fact it requires a strictly stronger assumption. Since the security analysis is the foundation of the claimed contribution, the paper is not publishable in its current form.","major_comments":[{"comment":"The model computes P(E) = P(F) * P(D|F) * P(T) * (1 - (1 - P(Ci))^N), which is the unconditional probability that at least one of N nodes raises a challenge. To justify the fast-finality rule of Section 3 step 4a, the paper needs the probability that a fraudulent state diff is NOT approved by the c0 randomly selected signers (i.e., that at least one honest node is selected and challenges, or that the diff is otherwise prevented from finalizing). These are not the same event. With a single honest node among N=10^4 and c0=100, the chance the honest node is excluded from the sample is roughly (1 - 100/10^4)^100 ≈ 0.99, so the proposed mechanism actually assumes that every approval sample contains an honest, actively monitoring verifier, which is strictly stronger than the traditional any-honest-party-can-challenge model. No analysis of this selection probability is provided.","section":"Section 4.2, P(E) formula"},{"comment":"The example sets P(T)=1 while the protocol's fast path uses t0=500 ms; this simply assumes away the detection challenge the paper is trying to solve. Consequently, the derived 99.11% figure is not a guarantee of secure finality; it is just 1 minus the probability that some node challenges, and it does not bound the probability that c0 colluding malicious signers approve an invalid diff. The paper never computes the probability of that bad event, which is the quantity that determines the security of the fast path.","section":"Section 4.2.1, Example"},{"comment":"The transition from the approval-based fast path to the traditional challenge model is unspecified. The text states that when the challenge window extends to 7+ days, the system operates like traditional fraud proofs and the number of active challengers can be zero, but it does not explain how the protocol determines that the fallback condition is reached, nor does it analyze whether an adversary can force acceptance of an invalid diff while the system is in the intermediate regime where both a shortened window and an approval requirement coexist. The exponential formulas t_n = t0 * rt^n and c_n = c0 * rc^n are asserted without a security argument for why this particular schedule is safe.","section":"Section 4.1.2, fallback to traditional challenge model"}],"minor_comments":[{"comment":"The phrase \"trough native delegation\" contains a typo and should read \"through native delegation.\"","section":"Section 2, first bullet"},{"comment":"The defined variable P(R) (participation rate) is never used in the P(E) formula; either remove it or incorporate it consistently into the model.","section":"Section 4.2, model variables"},{"comment":"The notation P(T) is defined as \"the probability that a node detects fraud within the challenge window T,\" but T is also used to denote the challenge window itself; this dual use of T is confusing and should be disambiguated.","section":"Section 4.2, notation"},{"comment":"The paper repeatedly refers to \"ideal scenarios\" and \"optimal conditions\" for fast finality, but never formally defines what these conditions are; a precise definition (e.g., all active challengers are honest and monitoring) would make the claims more testable.","section":"Abstract and Section 1, 'ideal scenarios'"}],"recommendation":"reject","confidential_remarks":"The paper's central security claim is not defensible in its current form: the mechanism's fast-finality rule implicitly assumes an honest node in every randomly selected approval quorum, which is strictly stronger than the traditional single-honest-node assumption. The probabilistic model in Section 4.2 computes a different quantity and cannot repair this gap. Substantial rewriting would be needed to reposition the paper, for instance by explicitly adopting an honest-majority or committee-based trust assumption and analyzing the corresponding sampling probabilities. As submitted, the manuscript does not meet the bar for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: the central security claim doesn't survive contact with the trust model. The paper says it keeps the single-honest-node assumption of traditional fraud proofs while letting c0 randomly selected active challengers approve a diff. With one honest node in the whole network, the chance that a random sample of 100 misses it is about (9999/10000)^100 ≈ 0.99. So the mechanism actually assumes at least one honest node in every approval sample, which is strictly stronger. Section 4.2's P(E) is the marginal probability that at least one node raises a challenge; it is not the conditional probability that a fraudulent diff is rejected before c0 sign-offs. And Example 4.2.1 sets P(T)=1 even though the target window is 500ms.\n\nThat said, the paper does useful work. Dynamic challenge periods are not new — the authors cite [18] — but the specific combination of an exponentially growing window with a shrinking required approval quorum, framed for the SVM account model, is a legitimate new variant. The system description is clear and the extension logic is easy to follow. The paper is honest that dispute resolution is out of scope and that DAS is future work.\n\nThe soft spots are in proportion: Section 4.2's model is the load-bearing part and it is mis-specified. It mixes P(F) and P(D|F) without conditioning on the actual selection of signers. The bonding/slashing discussion in 4.2.2 is asserted, not derived. The claim that the system can reach 'security level nearly equivalent to the base layer' is unsupported. There is no implementation or simulation, so the 99.11% finality figure in the example is just arithmetic from chosen parameters, not a measure of security.\n\nWho this is for: someone working on rollup finality designs might read it as a design proposal and a cautionary example. It is not a validated mechanism. It deserves a serious referee if the venue is open to design papers that state open problems, because the flaw is precisely the kind a good referee can surface. I would not cite it as a security result, but I might point people to it as a clear statement of an approach that needs a proper sampling analysis.\n\nMy recommendation: send it to peer review as a position/design paper, but with a revision request that either proves the sampling guarantee, weakens the trust model, or removes the claim that the single-honest-node assumption is preserved.","headline":"A clear design sketch whose fast-finality claim rests on an unexamined random-sampling assumption; the probabilistic model computes the wrong probability.","tokens_in":8372,"tokens_out":1880,"would_cite":false,"duration_ms":18189,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Dynamic Fraud Proofs claim sub-second finality by requiring random verifier approvals instead of a fixed challenge window.","keywords":["fraud proofs","optimistic rollups","dynamic challenge period","fast finality","Ephemeral Rollups","Solana Virtual Machine","verifier quorum","challenge window extension"],"falsifier":"Compute the probability that a random $c_0$-sized committee drawn from $N$ verifiers contains at least one honest node under the paper's assumption that only one honest node exists. If that probability is materially below 1 for the proposed parameters, then a malicious quorum could approve a fraudulent state diff and the sub-second finality claim would not preserve the single-honest-node guarantee.","tokens_in":7248,"feed_emoji":"⚡","tokens_out":11042,"duration_ms":91877,"temperature":0.7,"pith_summary":"This paper proposes a fraud-proof mechanism that replaces the fixed seven-day challenge window of optimistic rollups with a dynamic window that, under ideal conditions, settles state in sub-seconds. The Dynamic Fraud Proof construction requires a configurable number of randomly selected active verifiers to sign off on a state commitment; if enough approve, the state finalizes, and if not, the window grows exponentially while the required number of approvers shrinks. The paper's motivation is that fast finality is what allows off-chain execution layers such as ephemeral rollups to support real-time interactive applications instead of waiting minutes or hours for state to surface on the base layer. The paper contends that the design preserves the traditional single-honest-node assumption by inverting the challenge process, and it formalizes a probabilistic model of the chance that a challenge is raised.","feed_headline":"Dynamic fraud proofs bring finality down to sub-seconds","feed_subtitle":"Approval by a random verifier quorum can settle state in under a second, with long-window fallback for safety.","key_machinery":"The central object is the Dynamic Fraud Proof construction, an assert/challenge scheme in which finality depends on affirmative approvals rather than on the absence of challenges. The mechanism's two controls are the exponential challenge-window sequence $t_n = t_0 r_t^n$ and the exponential required-approver sequence $c_n = c_0 r_c^n$. These sequences allow the window to start near 500 ms and grow to roughly 6 days after ten steps, while the required number of active challengers decays from 100 to about 2. The probabilistic model $P(E) = P(F)P(D|F)P(T)(1-(1-P(C_i))^N)$ is the formal device used to estimate how often fast finality is achieved and to motivate bonding and slashing as ways to raise the detection and participation probabilities.","core_discovery":"The central claim is that the Dynamic Fraud Proof mechanism achieves fast finality by inverting the traditional optimistic challenge process. Rather than a fixed challenge window during which anyone may raise a challenge, a randomly selected set of active challengers must interactively approve the asserted state diff without raising a challenge. If the required number $c_0$ of approvals is reached by the end of the initial window $t_0$, the state settles; otherwise the window is extended by a factor $r_t$ and the required approvals are reduced by a factor $r_c$, until the window is long enough to behave like a classic fraud-proof system. The paper formalizes the challenge probability as $P(E) = P(F)P(D|F)P(T)(1-(1-P(C_i))^N)$ and argues that with bonding, slashing, and incentives this yields approximately 99.11% probability of achieving the target finality in its example.","pith_inferences":["A natural extension is to derive the probability that the randomly selected committee of $c_0$ verifiers contains at least one honest member as a function of $c_0$, $N$, and the honest-node fraction, since that probability would directly bound the safety of sub-second finality.","Because the commitment granularity is an account diff, the same dynamic-window construction could be applied beyond the paper's base-layer virtual machine, to any layer with fast block times and account-style state, and could also govern cross-chain message passing.","Pairing the approval quorum with data-availability sampling could let light clients serve as active challengers, lowering the hardware barrier to participation and making the ideal-condition assumption more realistic."],"forward_implications":["If the mechanism holds, optimistic rollups can target sub-second finality in ideal conditions while keeping a fallback to long challenge windows when the approval quorum is not met.","The exponential extension and approval-decay rules make the same construction tunable to different values at stake, participant sets, and bond sizes without a protocol-wide change.","Because finality requires affirmative sign-offs from a random set of verifiers, an attacker cannot simply censor challenges for a fixed duration to push through an invalid assertion.","Under the paper's example parameters, the model predicts about 99.11% of commitments would reach fast finality, with the remaining cases automatically delayed until a longer window restores classic fraud-proof safety."],"supporting_citations":[{"why":"Defines Ephemeral Rollups and the delegation model whose committed account diffs the mechanism finalizes.","marker":"[12]"},{"why":"Introduces the dynamic challenge period idea for EVM rollups that the paper adapts to its base layer.","marker":"[18]"},{"why":"Provides the canonical optimistic-rollup fraud-proof construction that the assert/challenge design modifies.","marker":"[1]"},{"why":"Supplies fast dispute-resolution techniques referenced for resolving challenges once they are raised.","marker":"[2]"},{"why":"Supplies the calculated optimal challenge-period baseline that the paper aims to beat with sub-second targets.","marker":"[6]"},{"why":"Explains why standard challenge periods are 7 days, motivating the need for a dynamic alternative.","marker":"[7]"},{"why":"Establishes the assert/challenge construction lineage that the dynamic fraud proof builds on.","marker":"[13]"},{"why":"Describes an earlier optimistic rollup whose challenge mechanism is compared with the proposed design.","marker":"[9]"}],"fun_headline_variants":["Sub-second finality via dynamic fraud proofs","Random verifier quorum speeds fraud proof finality","Dynamic challenge window yields sub-second settlement","Fraud proofs get real-time settlement with dynamic quorum","Approval by random quorum delivers sub-second finality"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a randomly selected approval quorum of size $c_0$ is sufficiently likely to contain an honest actor who would block fraud, even though the analysis quantifies the probability that any node raises a challenge rather than the probability that the quorum's signers are honest.","fun_headline_variants_meta":{"raw":{"variants":["Sub-second finality via dynamic fraud proofs","Random verifier quorum speeds fraud proof finality","Dynamic challenge window yields sub-second settlement","Fraud proofs get real-time settlement with dynamic quorum","Approval by random quorum delivers sub-second finality"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000388,"raw_usage":{"total_tokens":2035,"prompt_tokens":925,"completion_tokens":1110,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":541,"completion_tokens_details":{"reasoning_tokens":1037}},"tokens_in":541,"tokens_out":1110,"duration_ms":9871,"temperature":1.0,"reasoning_tokens":1037,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T18:34:13.434635+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the probability that a random $c_0$-sized committee drawn from $N$ verifiers contains at least one honest node under the paper's assumption that only one honest node exists. If that probability is materially below 1 for the proposed parameters, then a malicious quorum could approve a fraudulent state diff and the sub-second finality claim would not preserve the single-honest-node guarantee.","supporting_citations":[{"cited_title":"Ephemeral Rollups are All you Need","cited_arxiv_id":"2311.02650","evidence_quote":"Defines Ephemeral Rollups and the delegation model whose committed account diffs the mechanism finalizes."},{"cited_title":"Why wait a week? Fast Finality Optimistic Rollups","cited_arxiv_id":null,"evidence_quote":"Introduces the dynamic challenge period idea for EVM rollups that the paper adapts to its base layer."},{"cited_title":"Building Scalable Decentralized Payment Systems","cited_arxiv_id":"1904.06441","evidence_quote":"Provides the canonical optimistic-rollup fraud-proof construction that the assert/challenge design modifies."},{"cited_title":"BoLD: Fast and Cheap Dispute Resolution","cited_arxiv_id":"2404.10491","evidence_quote":"Supplies fast dispute-resolution techniques referenced for resolving challenges once they are raised."},{"cited_title":"Optimizing challenge periods in rollup","cited_arxiv_id":null,"evidence_quote":"Supplies the calculated optimal challenge-period baseline that the paper aims to beat with sub-second targets."},{"cited_title":"Why is the Optimistic Rollup challenge period 7 days?Accessed: 2024-07-07","cited_arxiv_id":null,"evidence_quote":"Explains why standard challenge periods are 7 days, motivating the need for a dynamic alternative."},{"cited_title":"Plasma : Scalable Autonomous Smart Contracts","cited_arxiv_id":null,"evidence_quote":"Establishes the assert/challenge construction lineage that the dynamic fraud proof builds on."},{"cited_title":"Arbitrum: Scalable, private smart contracts","cited_arxiv_id":null,"evidence_quote":"Describes an earlier optimistic rollup whose challenge mechanism is compared with the proposed design."}],"review_version":1}