{"id":"5addf571-5933-484d-8d7c-aa2fe810abd0","arxiv_id":"2607.11496","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":2,"one_line_summary":"A power-weighted randomized lottery with superlinear bid weights lets a user deter front-runners with a single bid that forms a subgame-perfect equilibrium.","lead":"PRECEDE orders blockchain transactions by a power-weighted lottery so a user can post one deterring bid that makes front-running unprofitable. This gives a deployable economic fix for causal ordering without trusted committees or encryption.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"Deterrence closed form and SPNE rest on exact, constant, a-priori knowledge of γ","rationale":"The reader correctly isolates the known-constant-γ premise as the weakest modelling assumption that the closed-form deterrence bid and all subsequent equilibrium statements rest upon. The mathematical development inside that premise is careful and the anti-spam / profitability reductions are tight; the only material gap is the lack of any quantitative robustness check for γ-error. Because the paper already flags the idealizations and the verdict is already CONDITIONAL, the stress-test does not move the recommendation. The concrete numerical probe above would simply make the practical claim more precise without altering the theoretical contribution.","tokens_in":46102,"tokens_out":493,"duration_ms":26223,"concrete_test":"Fix the practical (k,γ) pair of §6.1 (k=3.010, γ=0.259). Recompute max_b u(b;W) of Eq. 2 while the attacker’s true γ' ranges over [0.8γ,1.2γ]; report the smallest relative error |γ'-γ|/γ at which the max becomes strictly positive. If that error is <5 %, the deterrence guarantee is fragile under ordinary pre-execution noise.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (Theorem 1 + Corollary 2) that a single deterrence bid b^{k,γ}_dtr renders every attacker’s expected revenue non-positive, and that this profile is an SPNE, is derived under the premise that every participant knows the exact constant losing-fee rate γ in advance by local pre-execution (§3.2, Lemma 1, Eq. 3). The threshold weight W^{k,γ}_dtr scales as 1/γ; an under-estimate of γ therefore produces a bid that leaves a positive-revenue counter-bid for the true γ. In practice γ is path-dependent (gas consumed on the revert path after a competing transaction has already mutated state) and can only be estimated under a hypothesized order. No robustness margin or sensitivity analysis appears in the main theorems or the numerical k-selection of §6/App. J. If that premise fails, both Causal Ordering and the equilibrium claim fail for the published bid.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper proposes PRECEDE, a power-weighted randomized transaction ordering rule (weight b^k, k>1) that aims to enforce causal ordering against front-running by economic entry deterrence rather than by timestamps or encryption. A user who first discovers a reward R publishes a closed-form deterrence bid b^{k,γ}_dtr = R*((k-1)^{k-1}/(γ k^k))^{1/k}; under the model this makes every later entrant's expected revenue non-positive (Lemma 1, Theorem 1). The authors prove an anti-spam threshold k ≥ ln 2 / ln(1+γ) under which single bids dominate multi-bid strategies (Theorem 2), show that the same bound yields strictly positive user revenue (Proposition 2), and establish that deterrence plus attacker abstention is an SPNE of a multi-step sequential game when k is large enough that deterrence utility exceeds accommodation utility (Theorems 3–4, Corollaries 1–2). They also reduce sandwich defense to the same front-running game with R = V and discuss parameter selection using measured Uniswap gas ratios γ.","tokens_in":46327,"tokens_out":1099,"duration_ms":20027,"significance":"If the results hold under realistic conditions, PRECEDE is a genuine alternative to receive-order fairness and encrypted-mempool designs: it requires only a censorship-resistant collection phase, a public random beacon, and a change to the ordering function, and it targets incentive compatibility rather than cryptographic hiding. Strengths include explicit closed forms (deterrence weight/bid, anti-spam threshold, revenue-optimal k = 1 + 1/γ), full appendix proofs, a careful partial-pay contest model that matches blockchain fee mechanics better than classical all-pay Tullock contests, and independent mainnet-fork measurements of γ. The sandwich reduction and the anti-spam merging argument are particularly clean contributions to the MEV literature.","major_comments":[{"comment":"§3.2 and Lemma 1 / Eq. (3): The deterrence weight W^{k,γ}_dtr scales as 1/γ and the entire equilibrium apparatus (Theorem 1, Corollaries 1–2) assumes every participant knows an exact, constant losing-fee rate γ in advance by local pre-execution. In practice γ is path-dependent: gas on the revert path can change after a competing transaction has already mutated state, so the pre-execution estimate is only valid under a hypothesized order. An under-estimate of γ leaves a positive-revenue counter-bid for the true γ, so Causal Ordering fails for the published bid. The manuscript needs either a robustness margin (e.g., a conservative γ_min with a proved safety factor) or an explicit sensitivity analysis showing how much mis-estimation the deterrence bid tolerates; neither appears in the main theorems or in the numerical k-selection of §6 / App. J.","section":null},{"comment":"Corollary 1 vs §6 / App. J: The only closed-form sufficient condition for the SPNE is k ≥ max{2, exp(1/γ)/γ}, which for the measured Uniswap V2 γ ≈ 0.259 forces k ≈ 184 and leaves the user only ~0.026R. The paper then relies on a numerical search claiming that the much smaller anti-spam bound k ≈ 3.010 already satisfies u^{k,γ}_dtr > u^{k,γ}_acc. That numerical claim is load-bearing for any practical deployment recommendation, yet it is confined to an appendix, uses a bounded search (k ≤ 20), and is not elevated to a theorem or even a formal conjecture with stated precision. Either strengthen the analytic bound or move a fully specified, reproducible numerical certificate into the main body and clearly separate 'formal SPNE' from 'numerically observed SPNE'.","section":null},{"comment":"§5.1.1–5.1.2: Modeling a single colluding attacker is presented as worst-case, but the multi-step game also assumes perfect alternating observation, non-decreasing bids only, and that the attacker always moves last. The paper does not show that the deterrence SPNE survives (i) simultaneous or out-of-order moves, (ii) multiple non-colluding attackers who can free-ride on each other's entry attempts, or (iii) a validator who can both reorder and choose which of her own bids to include after seeing the random beacon. These are standard adversarial capabilities in the MEV setting the introduction targets; without at least a discussion of which of them break the SPNE, the equilibrium claim is narrower than the deployment claim in the abstract and conclusion.","section":null}],"minor_comments":[],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The one thing worth knowing is that this paper gives a deployable ordering rule that removes the economic incentive to front-run without committees, encryption, or trusted parties. PRECEDE weights each bid b by b^k (k>1), draws a random order proportional to weight, and shows a single closed-form deterrence bid makes every later entrant’s expected revenue non-positive. When k clears a modest threshold the user plays that bid and the attacker stays out; that profile is an SPNE of the multi-step game. Causal ordering, anti-spam, and positive user revenue all follow. Sandwich attacks reduce to the same front-running game, so the same bid stops them cheaper than today’s descending-bid auctions.\n\nWhat is new is the adaptation of the Tullock power contest to the partial-pay, multi-action, multi-transaction setting that blockchains actually use. Prior sequential-contest results do not cover this; the paper derives the deterrence weight, the anti-spam threshold k ≥ ln2 / ln(1+γ), the revenue-optimal k = 1 + 1/γ, and the multi-step SPNE from first principles. The score-based realization of the lottery is standard and correctly cited. Appendices contain full proofs; Appendix A measures real γ on a mainnet fork for Uniswap V2/V3; Appendix J shows the analytic sufficient bound on k is loose and the anti-spam bound already works numerically.\n\nThe soft spots are ordinary modeling ones, not load-bearing cracks. The entire closed form and equilibrium rest on every participant knowing the exact constant losing-fee rate γ in advance by local pre-execution. If γ is mis-estimated or path-dependent the published bid may leave a profitable counter-bid. The paper states the assumption clearly and measures practical values, but offers no robustness margin in the main theorems. The single colluding attacker is a worst-case modeling choice; independent attackers would only make deterrence easier. The gap between the huge analytic k and the small numerical value that works should be more prominent in the body. None of these invalidate the math under the stated model.\n\nThis is for anyone working on MEV, transaction ordering, or mechanism design for blockchains. The proofs are careful, the citations cover both the contest-theory and blockchain-ordering literatures without padding, and the contribution is concrete enough that a serious editor should send it to referees. I would bring it to reading group and expect to cite the deterrence construction if I write on ordering incentives.","headline":"Solid mechanism-design paper that turns front-running into an unprofitable entry game via a power-weighted lottery; closed-form deterrence bid and SPNE are clean, practical k is far better than the analytic bound, but everything hinges on known constant γ.","tokens_in":46954,"tokens_out":629,"would_cite":true,"duration_ms":14694,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"A power-weighted random lottery lets users deter front-running with one bid, so attackers rationally stay out and causal order holds.","keywords":["front-running","transaction ordering","mechanism design","entry deterrence","power-weighted lottery","sandwich attacks","blockchain MEV","subgame-perfect equilibrium"],"falsifier":"On a live Uniswap V2/V3 fork with measured γ, post the closed-form deterrence bid for a known arbitrage of size R and check whether any rational counter-bid still yields positive expected profit under the power-weighted lottery; a profitable counter-bid falsifies the claim.","tokens_in":46979,"feed_emoji":"⏱️","tokens_out":947,"duration_ms":10712,"temperature":0.7,"pith_summary":"Front-running lets attackers snatch rewards from useful blockchain transactions once they see them in the mempool, which destroys the incentive to publish those transactions. Existing ordering rules either fail to enforce the true causal order or turn the problem into a pure bidding war. This paper offers PRECEDE: order every transaction by a randomized lottery whose winning probability is proportional to bid^k with k greater than 1. Under that rule a user who finds an opportunity first can post a single, closed-form “deterrence bid” that makes every later entrant’s expected revenue non-positive. When k is large enough the strategy is a subgame-perfect equilibrium, the user still keeps a positive share of the reward, spam is unprofitable, and sandwich attacks are blocked at the same cost. The only on-chain change required is the ordering function itself.","feed_headline":"One bid stops front-running on blockchains","feed_subtitle":"A power-weighted lottery makes later entry unprofitable, so attackers stay out and users keep the reward","key_machinery":"The power-weighted randomized lottery (weight b^k, k>1) together with the closed-form deterrence bid derived from setting the attacker’s expected-revenue function non-positive for all positive counter-bids. Super-linearity squeezes both cheap and expensive counter-bids out of the profitable region, so a single user bid deters entry.","core_discovery":"Under the power-weighted lottery w(b)=b^k the smallest bid that deters every rational attacker is the closed-form b^{k,γ}_dtr = R*((k-1)^{k-1}/(γ k^k))^{1/k}. When that bid is posted, attacker revenue is non-positive for every counter-bid; when k meets explicit thresholds the same strategy together with attacker abstention is a subgame-perfect Nash equilibrium of the multi-step game, simultaneously satisfying Causal Ordering, Anti-Spam and Profitability.","pith_inferences":["The same super-linear weight that deters front-running may also reduce other MEV races whose value is realized only by being first, such as liquidation sniping.","Because the mechanism needs only public bids and a public random beacon, it can be layered on top of existing encrypted-mempool designs without requiring the mempool itself to stay secret.","If chains raise the losing-fee rate γ, user revenue under PRECEDE rises for every k, giving protocol designers a direct lever to improve user surplus.","The multi-step SPNE proof suggests that even continuous-time arrival models with latency advantages for the attacker should still admit a pure deterrence equilibrium once k is large enough."],"forward_implications":["Any censorship-resistant chain can adopt PRECEDE by changing only its transaction-ordering function.","Users keep a strictly positive share of the reward (up to γ/(1+γ)R at the revenue-optimal k) instead of watching it competed away.","Sandwich attacks are deterred at a cost strictly less than the loss they would inflict under pure bid-priority ordering.","Spam is dominated once k ≥ ln 2 / ln(1+γ), so multi-transaction flooding is no longer rational.","A single conservative k can be fixed for a measured range of γ values and still keep deterrence an equilibrium."],"fun_headline_variants":["One deterring bid stops front-running via power lottery","PRECEDE makes later bids unprofitable so attackers abstain","Power-weighted lottery enforces causal order on blockchains","Closed-form bid deters every rational front-runner","Super-linear lottery removes incentive to front-run"],"cache_read_input_tokens":32896,"weakest_assumption_plain":"Every participant must know the exact losing-fee rate γ in advance by local pre-execution, and that rate must stay fixed for the whole competition; if γ is mis-estimated or path-dependent the deterrence bid can fail.","fun_headline_variants_meta":{"raw":{"variants":["One deterring bid stops front-running via power lottery","PRECEDE makes later bids unprofitable so attackers abstain","Power-weighted lottery enforces causal order on blockchains","Closed-form bid deters every rational front-runner","Super-linear lottery removes incentive to front-run"]},"model":"grok-4.5","effort":"low","cost_usd":0.004668,"raw_usage":{"total_tokens":1370,"prompt_tokens":840,"num_sources_used":0,"completion_tokens":84,"cost_in_usd_ticks":46680000,"prompt_tokens_details":{"text_tokens":840,"audio_tokens":0,"image_tokens":0,"cached_tokens":128},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":446,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":840,"tokens_out":84,"duration_ms":4925,"temperature":1.0,"reasoning_tokens":446,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-14T05:11:59.261629+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"On a live Uniswap V2/V3 fork with measured γ, post the closed-form deterrence bid for a known arbitrage of size R and check whether any rational counter-bid still yields positive expected profit under the power-weighted lottery; a profitable counter-bid falsifies the claim.","supporting_citations":[],"review_version":1}