{"id":"32cf750e-c162-4410-93ac-e88e84f58d54","arxiv_id":"2411.11789","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Resonance is a broker-based fee mechanism for heterogeneous blockchain compute markets that achieves efficiency, budget balance, individual rationality, and conditional incentive compatibility at Nash equilibria.","lead":"The paper designs Resonance, a blockchain fee mechanism where competing brokers propose how to match transactions to computing nodes and set individual prices. It proves that in equilibrium the mechanism is efficient, budget-balanced, and simple for users and nodes, though it relies on brokers knowing everyone's value and cost.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 5.2's proof is internally inconsistent: in the equal-margin case it decreases a transaction payment to raise margin, but margin falls; the constructed deviation is not budget-balanced, so the central efficiency claim is unproven as written.","rationale":"The reader flagged the omniscient-broker and zero-cost assumption, which the paper itself discloses in Section 6 and defers to future work. That is a scope limitation, not an internal flaw. More load-bearing is the unacknowledged error in the proof of Theorem 5.2: as written, the contradiction in Case 2 constructs a non-budget-balanced routing and thereby fails to establish the efficiency guarantee even under the paper's ideal assumptions. Since Theorem 5.2 is the paper's central claim, a reader relying on the proof cannot verify the result as presented. The error appears repairable, so this does not necessarily overturn the theorem; the appropriate verdict remains conditional on a corrected proof and, separately, on the acknowledged modeling assumptions. The reader's conditional verdict is therefore preserved, but for a different and more fundamental reason than the one the reader identified.","tokens_in":20318,"tokens_out":24215,"duration_ms":233011,"concrete_test":"Re-derive Theorem 5.2's surplus-maximality proof with the payment adjustment corrected: for an individually rational, budget-balanced R′ with surplus ε > S(R), find an agent with positive utility and increase its payment (or decrease a node's payment) by δ with 0 < δ < min{ε − S(R), u_a}, producing R′′ with positive margin and surplus > S(R). Verify such an agent exists whenever ε > S(R). If the corrected proof fails, construct a concrete type vector and two-broker profile where the output is not surplus-maximizing yet every losing broker deviation that would win is rejected by the individual rationality check.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's headline result (Theorem 5.2) asserts that at any PNE with at least two brokers and truthful reports from transactions and nodes, the output routing is welfare-maximizing and surplus-maximizing. The proof relies on Claim 5.3 (zero margin) and then shows surplus-maximality by contradiction. In the second bullet, Case 2, the proof assumes an alternative budget-balanced routing R′ with strictly higher surplus than the incumbent R. Since R has zero margin (Claim 5.3) and R′ has equal margin, R′ has zero margin. The proof then constructs R′′ identical to R′ except that transaction t's payment is reduced to π′(t)−δ, claiming R′′ has strictly higher surplus and strictly higher margin than R. But margin is defined as Σπ − Σφ; lowering a transaction payment lowers margin. With R′'s margin at zero, R′′'s margin is −δ, so R′′ is not weakly budget-balanced and cannot be selected by the mechanism. A losing broker deviating to R′′ would not win and would earn zero, not γ(R′′). To create a profitable deviation one must instead increase a payment (or decrease a node payment) of an agent with positive utility, choosing δ small enough to preserve individual rationality and keep surplus above the incumbent's. The proof also never establishes that the alternative R′ can be chosen individually rational, even though the mechanism rejects proposals that give any agent negative utility. Although a corrected proof likely exists (an individually rational, budget-balanced routing achieves surplus equal to maximum welfare by paying node costs and charging users a total equal to those costs), the argument as written does not prove Theorem 5.2, and the gap is in the central efficiency result.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes Resonance, a broker-based transaction fee mechanism for a two-sided market with fully heterogeneous users (transactions with private valuations) and nodes (with private cost functions), subject to arbitrary allocation constraints. Brokers submit routings (allocation plus transaction and node payments); the mechanism selects the budget-balanced proposal with highest surplus under the reported types, checks individual rationality, and pays the winning broker the routing's margin. The paper claims budget balance, individual rationality for all parties, welfare/surplus efficiency at pure Nash equilibria (Theorems 5.1 and 5.2), a new incentive property termed DSIC barring B (Theorem 5.3), and computational tractability. An appendix analyzes limitations of multidimensional fee markets, arguing that individualized prices are necessary in heterogeneous settings.","tokens_in":20619,"tokens_out":22079,"duration_ms":226441,"significance":"If the main results were correct, Resonance would be a significant design contribution: it replaces on-chain price oracles and allocation algorithms with competition among sophisticated brokers, while keeping the user/node interface strategically simple. The general two-sided model with arbitrary validity constraints is well motivated by prover networks, AI/ML compute markets, and specialized blockchain infrastructure. The appendix's separation of pricing from allocation and its impossibility results for shared base fees are also useful. The paper is clearly written and self-contained, and Section 6 candidly discusses practical issues such as griefing, private order flow, broker specialization, and collusion. However, the central efficiency theorem (Theorem 5.2) is false as stated, and the proof contains additional gaps; these issues are load-bearing and prevent acceptance in the current form.","major_comments":[{"comment":"Theorem 5.2 is false as stated. Consider T={t} with v_t=1, N={n} with c_n(·)=0, and B={b1,b2}, with b1 first in the tie-breaking order. Let both brokers propose R=(α(t)={n}, π(t)=2, φ(n)=2). This routing is budget-balanced (margin 0) and has surplus S(R|θ)=(1-2)+(2-0)=1, so it is a highest-surplus budget-balanced proposal. The mechanism selects b1 but then, because the transaction has negative utility under the truthful reported type, the IR check fails and the mechanism outputs the empty routing. The profile (θ,R,R) is a PNE: b1 deviating to the IR routing (π=1, φ=1) gets the same surplus and zero margin, so its payoff stays 0; any positive-margin deviation has surplus below 1 and does not win; b2 cannot win any deviation with surplus 1 because of the fixed tie-break, and deviations with surplus below 1 lose; a unilateral report change by the transaction either leaves the output empty or makes R pass IR only if the transaction reports at least 2, yielding true utility -1, while the node alone cannot make R pass. Thus the empty routing is output at a PNE, yet its welfare is 0 while the maximum welfare is 1, contradicting both bullets of Theorem 5.2. The proof's deviation arguments do not account for the fact that a losing broker's higher-surplus proposal may be blocked by another broker's non-IR proposal that wins the surplus comparison but is later rejected.","section":"Section 5.3, Theorem 5.2"},{"comment":"The proof of the equal-margin case constructs R'' from R' by decreasing the transaction payment to π'(t)-δ and asserts that R'' has strictly higher surplus and strictly higher margin than R. This is arithmetically incorrect: margin is defined as Σπ−Σφ, so lowering a transaction payment lowers the margin. Since R' has zero margin, R'' has margin −δ<0 and is not weakly budget-balanced, so the mechanism would not select it. A correct deviation would need to increase a payment or decrease a node payment, with a separate argument preserving individual rationality and surplus dominance; the proof as written does not provide this.","section":"Section 5.3, proof of Theorem 5.2, Case 2"},{"comment":"The proof assumes that if an alternative budget-balanced routing R' has higher surplus and higher margin than the incumbent R, then the winning broker can deviate to R' and collect margin γ(R'). This ignores the mechanism's individual-rationality check: if R' gives any transaction or node negative utility under the true reported types, the mechanism outputs the empty routing and the deviating broker receives zero, not γ(R'). The proof never shows that R' (or a suitable modification of it) is individually rational, so the claimed contradiction is not established even apart from the counterexample in the previous comment.","section":"Section 5.3, proof of Theorem 5.2, Case 1"},{"comment":"The proof constructs 'a surplus-maximizing routing' for the brokers without specifying the feasible set over which surplus is maximized. If the maximum is taken over all routings, surplus is unbounded because node payments can be increased without limit; if it is taken over weakly budget-balanced routings, the resulting routing need not be individually rational, in which case the constructed profile (θ,σ) can be the bad equilibrium described in the counterexample to Theorem 5.2. The assertion that σb' is 'welfare-maximizing with zero margin' is therefore unsupported. The definition of the broker strategy in the proof and the verification of both bullets of Definition 5.3 need to be made precise.","section":"Section 5.4, Theorem 5.3"}],"minor_comments":[{"comment":"The mechanism defines b* as the arg max over budget-balanced proposals, but does not specify what happens when no budget-balanced proposal exists (the arg max is over an empty set). The text should state explicitly that the mechanism then outputs the empty routing, as is implicitly assumed in Lemma 5.2.","section":"Section 4, Mechanism definition"},{"comment":"The sentence 'all routings in σ have the same allocation rule, and therefore the same surplus under any type vector' is not literally correct if the routings have different payment rules; the surpluses then differ by the respective margins. What matters for the argument is that the relative ranking of the proposals is independent of the reported types because all allocations coincide, and this should be stated accurately.","section":"Section 5.4, proof of Theorem 5.3"},{"comment":"Typo: 'budged-balanced' should be 'budget-balanced'.","section":"Section 5.3, proof of Theorem 5.2"},{"comment":"The sentence 'We will denote by A the set A of all transactions and nodes' is redundant; it should read 'We denote by A the set of all transactions and nodes.'","section":"Section 2, notation"},{"comment":"The text says 'As shown in Section 3, Resonance satisfies...', but the formal properties are established in Sections 4 and 5; the reference should be updated.","section":"Section 6, first paragraph"}],"recommendation":"major_revision","confidential_remarks":"The counterexample in major comment 1 is a genuine counterexample to the paper's headline theorem, not merely a gap in a proof. It is, however, repairable within the scope of the paper: for instance, changing the mechanism to select the highest-surplus proposal among those that already satisfy individual rationality (rather than selecting first and checking IR afterward) would eliminate the bad equilibrium. Given the centrality of Theorem 5.2 and the arithmetic error in its proof, I cannot recommend acceptance in the current form; I would encourage a substantial revision along these lines."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nTwo things to know before you read it. Resonance is a genuinely useful step for fee mechanism design in fully heterogeneous two-sided markets, and the broker competition design is clean. But the proof of Theorem 5.2, the paper's headline efficiency result, has a real gap as written, and I think it's fixable.\n\nWhat's new: the model is properly general—arbitrary valuations, arbitrary node cost functions, arbitrary validity constraints—and the mechanism is simple to state: brokers submit routings, the auctioneer picks the highest-surplus budget-balanced proposal that is individually rational, and the winning broker keeps the margin. The paper's new 'DSIC barring B' relaxation, requiring truth-telling by users and nodes only for a particular broker equilibrium, is a sensible fit for settings with sophisticated intermediaries. The appendix's Theorem A.3, showing uniform base fees fail badly with heterogeneous nodes, is a solid contribution. I also give credit for the open discussion of griefing and broker information: those are real limitations, and the authors say so.\n\nThe soft spots, in order. The omniscient-broker assumption is strong, but explicit and deferred. The DSIC barring B equilibrium selection argument is informal, as the authors admit. More serious: in the proof of Theorem 5.2, second bullet, Case 2, the argument constructs R'' from R' by decreasing a transaction payment. Since R' has zero margin, R'' has negative margin and is not budget-balanced, so the mechanism cannot select it, and the proposed broker deviation fails. That is a genuine hole in the central proof. I do not think the theorem is false; increasing a payment (or decreasing a node payment) by a small amount on an agent with positive utility would preserve budget balance and IR and likely give the same contradiction. But as written, the efficiency claim is unproven.\n\nNet: if the authors fix that proof, this is a worthwhile paper that a mechanism design reader should engage with. It deserves a serious referee; the contribution is strong enough that the gap is worth working through.","headline":"A novel broker-based fee mechanism with a real but fixable proof gap in its headline efficiency theorem; worth reviewing after the authors repair Theorem 5.2.","tokens_in":21147,"tokens_out":6312,"would_cite":true,"duration_ms":61077,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["91A10","91B26","91B03"],"pacs":[],"model":"deepseek-v4-flash","headline":"A broker competition mechanism makes blockchain fees efficient and margin-free.","keywords":["transaction fee mechanism","two-sided market","broker competition","pure Nash equilibrium","surplus maximization","heterogeneous computation","blockchain","mechanism design"],"falsifier":"One concrete test: in a small instance with two brokers, two transactions, and two nodes, run a full search over both agents' action spaces and verify that every pure Nash equilibrium has zero margin and maximum surplus. Finding any equilibrium with strictly positive margin or suboptimal surplus would refute Claim 5.3 and Theorem 5.2. A second, easier test is to endow one broker with slightly noisy information about a node's cost; the paper's assumptions no longer hold, so an equilibrium with positive margin would show where the efficiency guarantee breaks.","tokens_in":20104,"feed_emoji":"⚖️","tokens_out":6275,"duration_ms":55696,"temperature":0.7,"pith_summary":"Resonance is a transaction fee mechanism for blockchains in which users and nodes are fully heterogeneous: each user has an arbitrary private value for inclusion and each node an arbitrary private cost for any bundle of transactions, with arbitrary validity constraints on allocations. The paper's central claim is that delegating pricing and allocation to competing brokers — sophisticated agents who know everyone's types and propose complete routings — yields efficient outcomes in equilibrium. Specifically, with at least two brokers, every pure Nash equilibrium is welfare-maximizing among all valid allocations and surplus-maximizing among all weakly budget-balanced routings, with zero broker margin, so users and nodes capture all generated value. The authors also prove that simpler multi-dimensional fee markets with shared unit prices cannot achieve this: even given perfect pricing and allocation oracles, such markets can lose almost all surplus once nodes have heterogeneous costs (Theorem A.3). Resonance is budget-balanced, individually rational, and computationally light, and it minimizes the need for strategization by users and nodes.","feed_headline":"Competing brokers make blockchain fees efficient and margin-free","feed_subtitle":"Resonance proves two brokers at Nash equilibrium route transactions to nodes at zero broker margin with maximum surplus.","key_machinery":"The load-bearing object is the broker proposal combined with a surplus-maximizing selection rule. A routing specifies a valid allocation $\\alpha : T \\to 2^N$, a user payment rule $\\pi$, and a node payment rule $\\varphi$, and its margin is the net cash flow $\\sum_{t\\in T} \\pi(t) - \\sum_{n\\in N} \\varphi(n)$. Brokers compete by proposing routings; the mechanism selects the highest-surplus budget-balanced proposal and transfers the margin to the winning broker. Competition among at least two brokers drives margins to zero, because any positive margin can be undercut by a rival who reduces one transaction's payment slightly, raising surplus and winning the auction. That zero-margin property converts surplus maximization into welfare maximization and is the key identity behind Theorem 5.2.","core_discovery":"The central discovery is that the two hard problems in heterogeneous fee markets — computing a good allocation and discovering individualized prices — can be offloaded to a new class of agents called brokers. Each broker submits a routing, meaning a valid allocation plus a payment from every transaction and a payment to every node; the mechanism keeps the budget-balanced routing with the highest surplus according to the reported types, provided it gives every transaction and node non-negative utility, and pays the winning broker the routing's margin. The paper proves (Theorem 5.2) that with at least two brokers, at any pure Nash equilibrium the winning routing has zero margin and is surplus-maximizing among all weakly budget-balanced routings, hence also welfare-maximizing. With a single broker the outcome is welfare-maximizing but surplus is zero, and Resonance is dominant-strategy incentive-compatible for users and nodes; with multiple brokers it satisfies the weaker notion 'DSIC barring B', meaning users and nodes need not strategize against each other when brokers play a natural equilibrium.","pith_inferences":["The zero-margin equilibrium result depends on brokers having complete and free information about all types; in practice, broker specialization and estimation costs would leave positive margins, so the paper's efficiency guarantee should be read as an idealized benchmark rather than a literal prediction.","The DSIC-barring-B notion is weaker than it might appear: it guarantees users and nodes need not strategize against each other only when brokers coordinate on a common welfare-maximizing allocation with at least two zero-margin proposals, which the paper argues is the outcome of best-response dynamics but does not prove for all equilibria.","A testable extension would be to simulate broker best-response dynamics in a setting with noisy type estimates and see whether margins converge to the cost of information acquisition rather than to zero; the paper's claim would then become that competition reduces margins to broker costs, not to zero.","The grievance-vector attacks identified in Section 6 (single-agent rejection and private order flow) suggest that practical deployment needs additional safeguards; the paper states modifications exist but does not analyze them, so the mechanism in its stated form may not be robust to a single malicious low-reporting agent."],"forward_implications":["If Resonance is used as a blockchain fee market, users and nodes can be told to report their valuations and costs truthfully; strategizing is relegated to brokers, whose competition replaces a centralized price oracle.","The mechanism supports arbitrary validity constraints, so it can express state-conflict rules, capacity limits, and multi-node transaction types such as multi-party computation in one framework.","Theorem A.3 implies that any fee mechanism that charges a shared vector of unit prices per resource can lose almost all surplus when nodes have heterogeneous costs, providing a formal limit on the EIP-1559-style approach and a direct motivation for per-transaction, per-node pricing.","Because Resonance is weakly budget-balanced and individually rational, it can be deployed on-chain without subsidy, and its per-proposal computations scale linearly in the number of agents and brokers."],"supporting_citations":[{"why":"Shows multi-dimensional blockchain fees can be near-optimal under an allocation oracle, the benchmark Resonance is designed to surpass without such an oracle.","marker":"[5]"},{"why":"Designs multi-dimensional fee markets and gives price-update rules; Resonance generalizes this setting to arbitrary node costs and validity constraints.","marker":"[15]"},{"why":"Models zk-rollup prover markets with heterogeneous suppliers, a concrete instance of the two-sided heterogeneous market Resonance addresses.","marker":"[27]"},{"why":"Introduces robust double auctions for resource allocation, another heterogeneous supplier market used as a motivating comparison.","marker":"[20]"},{"why":"Establishes impossibility results for coalition incentive compatibility, invoked to justify the weaker DSIC-barring-B notion.","marker":"[17]"},{"why":"Shows no efficient, individually rational, budget-balanced mechanism exists for bilateral trade with private values, the reason Resonance assumes brokers know types.","marker":"[23]"},{"why":"Optimal auction design, cited for the general difficulty of dominant-strategy incentive compatibility in rich type spaces.","marker":"[22]"}],"fun_headline_variants":["Brokers drive blockchain fees to zero margin","Two brokers suffice for efficient blockchain fees","Resonance: Efficient fees via broker competition","Heterogeneous fees solved by broker bidding","Zero-margin equilibrium for blockchain fees"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The theorems require brokers to know the exact private valuations and cost functions of every transaction and node, and to bear zero cost for proposing and operating routings; if broker information is incomplete or costly, the efficient zero-margin equilibrium is not delivered.","fun_headline_variants_meta":{"raw":{"variants":["Brokers drive blockchain fees to zero margin","Two brokers suffice for efficient blockchain fees","Resonance: Efficient fees via broker competition","Heterogeneous fees solved by broker bidding","Zero-margin equilibrium for blockchain fees"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000111,"raw_usage":{"total_tokens":1059,"prompt_tokens":952,"completion_tokens":107,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":568,"completion_tokens_details":{"reasoning_tokens":42}},"tokens_in":568,"tokens_out":107,"duration_ms":2028,"temperature":1.0,"reasoning_tokens":42,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T18:09:26.328179+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One concrete test: in a small instance with two brokers, two transactions, and two nodes, run a full search over both agents' action spaces and verify that every pure Nash equilibrium has zero margin and maximum surplus. Finding any equilibrium with strictly positive margin or suboptimal surplus would refute Claim 5.3 and Theorem 5.2. A second, easier test is to endow one broker with slightly noisy information about a node's cost; the paper's assumptions no longer hold, so an equilibrium with positive margin would show where the efficiency guarantee breaks.","supporting_citations":[{"cited_title":"Designing mul- tidimensional blockchain fee markets","cited_arxiv_id":null,"evidence_quote":"Designs multi-dimensional fee markets and gives price-update rules; Resonance generalizes this setting to arbitrary node costs and validity constraints."},{"cited_title":"Robust double auctions for resource allocation","cited_arxiv_id":null,"evidence_quote":"Introduces robust double auctions for resource allocation, another heterogeneous supplier market used as a motivating comparison."},{"cited_title":"On coalition incentive c ompatibility","cited_arxiv_id":null,"evidence_quote":"Establishes impossibility results for coalition incentive compatibility, invoked to justify the weaker DSIC-barring-B notion."},{"cited_title":"Efficient mechanism s for bilateral trading","cited_arxiv_id":null,"evidence_quote":"Shows no efficient, individually rational, budget-balanced mechanism exists for bilateral trade with private values, the reason Resonance assumes brokers know types."}],"review_version":1}