{"id":"b55c16cc-4a04-4521-b735-ba96412e85d5","arxiv_id":"2507.16350","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"ADRF adapts the loop-free Precomputed Dominant Resource Fairness algorithm to Ethereum smart contracts using fixed-point arithmetic and epoch-based demand/claim transactions, with measured gas costs linear in the number of resource types.","lead":"Blockchains usually allocate resources as fixed bundles, which wastes idle capacity. This paper presents a smart-contract version of an existing fairness algorithm, called ADRF, and measures its gas costs, finding linear growth in resource count and no cost growth in user count.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The load-bearing assumption is inherited PDRF fairness: ADRF skips the correction step and never measures its deviation from true DRF, so the algorithm's central use-case claim rests on an unverified, self-cited statistic.","rationale":"The reader's weakest assumption is exactly the inherited PDRF approximation, and I agree that it is the most load-bearing gap. The gas-cost scaling is the strongest part of the paper: the code has no loops over users, only deterministic loops over resources, and the reported R^2 values are consistent with that structure. Credit is due for the measured linear scaling and for the clear statement that ADRF omits the excess-reserve correction. However, the algorithm's stated purpose is to provide Dominant Resource Fairness in a blockchain context. Gas efficiency without demonstrated fairness is insufficient. The fairness statistics in Section 3.3 are averages over one distribution, not bounds; they come from a self-authored preprint; and Section 4 explicitly removes the correction mechanism that the statistics were meant to describe. The paper does not show that the 47% one-task-short behavior is bounded over epochs or that the same users are not systematically shortchanged. These issues are addressable through a fairness-oriented simulation, and they do not undermine the gas measurements. The reader's CONDITIONAL verdict is therefore appropriate; no verdict change is needed.","tokens_in":14999,"tokens_out":7060,"duration_ms":82292,"concrete_test":"Run a simulation of ADRF's pure-PDRF allocation on the same distributions used in Section 6 (m in {10,100}, n in {10,100}, demands drawn from U[1,10]) and compare, user by user, against the exact DRF allocation computed by the progressive-filling loop of [9]. Record the fraction of users under- or overallocated by more than one task, the maximum shortfall, and the minimum-to-maximum dominant-share ratio, both after a single epoch and after 10 consecutive epochs with reserve carry-over. If the Section 3.3 statistics do not reproduce for ADRF, or if the shortfall exceeds one task for a non-negligible fraction of users or accumulates over epochs, the inherited-fairness assumption fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing premise is that ADRF actually inherits PDRF's claimed approximation to DRF. Section 3.3 reports the only fairness evidence: under a discrete uniform demand distribution, 'on average, 47% of the users are allocated 1 task short' of DRF and 0.06% are overallocated; these statistics are cited to the author's own preprint [11]. Section 4 then states ADRF is 'an implementation of pure PDRF' that does not distribute excess reserves. Since the correction step described at the end of Section 3.3 is deliberately omitted, the relevant question is whether the remaining approximation error stays acceptable for ADRF's actual use cases. The paper provides no ADRF-specific fairness measurement, no worst-case analysis, and no test over repeated epochs; the carry-over of excess reserves may not compensate the particular users who were underallocated. If, for instance, 47% of users are one task short every epoch and the same users do not catch up, long-run fairness deviates from DRF even though each epoch's per-user shortfall is small. Because the algorithm's purpose is fair multi-resource allocation, this unvalidated inheritance is the weakest point in the argument.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes Autonomous Dominant Resource Fairness (ADRF), a smart contract adaptation of Precomputed Dominant Resource Fairness (PDRF) for blockchain environments. ADRF replaces the central allocation loop of DRF with three functions: updateState, which precomputes the number of PDRF cycles; demand, with which users register their demands and update scaled demand sums; and claim, with which users compute their reserved share and deduct it from the resource reserves. The empirical section reports gas cost measurements on Brownie/Ethereum for up to 100 resource types with 10 users, giving linear regressions such as gas_claim = 15,130·m + 36,486 (R² = 1) and concluding that gas cost is independent of the number of users. The paper claims ADRF is a gas-cost-efficient algorithm that can manage hundreds of resource types for an unlimited number of users.","tokens_in":15057,"tokens_out":12954,"duration_ms":120124,"significance":"If the fairness and efficiency properties are confirmed, ADRF would be a useful building block for multi-resource fair allocation on blockchains, where the block gas limit makes loop-based DRF implementations unattractive. The reported gas measurements are internally consistent, with near-perfect R² values that are plausible for deterministic loops over resource types, and the contract code appears to be available in a public repository. The main unresolved point is the fairness approximation: the paper does not independently validate that ADRF's allocations are close to DRF, and it does not compare against a loop-based baseline, so the central claims in the abstract and conclusion are only partially supported by the presented evidence.","major_comments":[{"comment":"The fairness guarantee of ADRF is not established. Section 4 explicitly states that ADRF is 'an implementation of pure PDRF' and that the problem of distributing excess reserves is not addressed. The only quantitative evidence that PDRF approximates DRF is the statistic in Section 3.3 ('47% of the users are allocated 1 task short... 0.06% are overallocated'), which is cited to the author's own preprint [11] and is obtained under a discrete uniform demand distribution. No ADRF-specific measurement of the deviation from true DRF is provided, nor is there a repeated-epoch analysis to show that the per-epoch underallocation does not accumulate or that the omitted correction step is immaterial. Because the algorithm's stated purpose is fair multi-resource allocation, this missing validation is load-bearing. The cross-check in Section 5 only shows that the Solidity implementation matches a Python implementation of PDRF; it does not test against DRF itself.","section":"Sections 3.3 and 4"},{"comment":"The efficiency claim lacks a baseline comparison. The stated motivation is that DRF's allocation loop is prone to block gas limit exhaustion, but the paper provides no measurements of a loop-based DRF smart contract. The regressions in Section 6 (e.g., gas_claim = 15,130·m + 36,486, gas_demand = 13,616·m + 47,245, gas_update = 11,295·m + 23,539) report absolute gas costs for ADRF only. Without demonstrating that a conventional DRF implementation is more expensive or infeasible for the tested ranges of users and resources, the conclusion that ADRF is 'gas-cost efficient' (Abstract) is only partially supported.","section":"Sections 1 and 6"},{"comment":"The claim of 'unlimited number of users' is not supported by the data presented in the manuscript. All tests described in Section 5 use 10 users; the paper argues from the absence of user loops that gas cost is independent of n, but it does not report experiments with larger user counts in the paper (only a pointer to a repository). The wording in the Abstract and Section 7 should be qualified, e.g., to 'independent of the number of users in the tested range' or be backed by additional experiments with substantially larger n.","section":"Sections 5 and 7"}],"minor_comments":[{"comment":"Typo: 'On avarage' should be 'On average'.","section":"Section 3.3"},{"comment":"The phrase 'deduces the assigned amount from the total reserve' should use 'deducts' instead of 'deduces'.","section":"Section 3.4"},{"comment":"The table header includes an 'Avg.' column, but the rows appear to contain only five values (number of resources, calls 1–3, and standard deviation). Clarify what 'Avg.' denotes and provide the corresponding values, or remove the column.","section":"Table 1"},{"comment":"The notation 'r← r1−s + er' is ambiguous; use a distinct subscript or label to clarify that this is an element of the cyclic buffer.","section":"Algorithm 1, line 13"},{"comment":"The repository link points to a project titled 'blockchainFaucet'; please clarify that the ADRF contract is contained there or provide a dedicated link to the ADRF code.","section":"Reference [44]"},{"comment":"With precision factor p = 1,000,000, the intermediate computation in Algorithm 1, line 14, multiplies r, ds'_*, and p; a brief comment on potential integer overflow limits would be useful, especially if the code is intended for larger reserve values.","section":"Section 4.2"}],"recommendation":"major_revision","confidential_remarks":"The paper's fairness evidence rests on a self-cited preprint (ref. [11]) that is not included in the submission. Although footnote 1 discloses the order of development, the committee may want to ask the author to reproduce the PDRF approximation statistics in this paper's setting or provide independent validation. Also, the repository link in [44] points to a project with a different name; please ensure the ADRF code is clearly identifiable so reviewers can verify the measurements."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The one thing you should know: this is a solid, modest engineering paper, not a theory advance. The author takes his own PDRF approximation of DRF, instantiates it as a Solidity smart contract with fixed-point arithmetic and cyclic epoch buffers, and measures gas costs. The results—gas linear in resource count, flat in user count—look internally consistent, and the near-perfect R² values are plausible for deterministic loops.\n\nWhat is actually new: the ADRF-specific formulas (the precision-scaled k′ and share calculations), the two-parity cyclic buffer design for parallel demand/claim epochs, and the empirical gas regressions. The paper is commendably honest: footnote 1 openly discloses that PDRF emerged from the same blockchain-adaptation project, so the reader knows exactly what is inherited.\n\nThe soft spots are real but not fatal. The central fairness claim rests entirely on PDRF’s approximation statistics (47% of users one task short, 0.06% overallocated), cited to the author’s own preprint [11]. ADRF explicitly skips the correction step, and the paper never measures ADRF’s actual deviation from true DRF. The stress-test concern is legitimate: if the same users are short every epoch and the excess reserve is just carried over, long-run fairness could drift from DRF. That needs an answer. Also, \"unlimited users\" is an overclaim—storage and per-epoch transactions grow with user count, even if a single call’s gas does not. There is no baseline comparison against a loop-based DRF smart contract, so \"gas efficient\" lacks a benchmark. And the GitHub reference is a 2020 repo labeled blockchainFaucet with no commit hash; the ADRF artifacts are not locatable from the paper.\n\nNone of this sinks the central gas-scaling result, which is likely correct. The fixes are clear: publish the ADRF-specific code with a commit hash, add a fairness-error measurement against true DRF on the test distributions, include a baseline, and temper the scalability claim.\n\nWho is this for? People working on blockchain resource markets, token economies, or smart-contract-based fair allocation. It is a genuine engineering data point. I would send it to peer review—conditional accept/revise—rather than desk reject. The referee should push on the fairness measurement and code availability, not on the gas data.","headline":"A modest, honest engineering paper: ADRF puts PDRF on-chain with clean gas measurements, but the fairness claim is inherited from a self-cited preprint and never re-measured.","tokens_in":15757,"tokens_out":1304,"would_cite":false,"duration_ms":15362,"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":"A smart-contract adaptation of Dominant Resource Fairness lets a blockchain allocate many resource types fairly at per-call gas costs that scale with the number of resources but not with the number of users.","keywords":["blockchain","smart contract","dominant resource fairness","resource allocation","precomputed dominant resource fairness","gas cost","max-min fairness","decentralised resource management"],"falsifier":"Deploy the published contract, draw the same discrete uniform demands as the paper, and run a reference DRF implementation on the same demand vectors: if noticeably more than 47% of users end up more than one task short of their DRF allocation, or if overallocation exceeds 0.06%, the inherited approximation bound fails for ADRF. The 'unlimited users' claim is separately checkable by holding resources fixed, doubling the number of registered users, and recording claim gas, which the paper claims stays flat.","tokens_in":1738,"feed_emoji":"⚖️","tokens_out":1740,"duration_ms":88661,"temperature":0.7,"pith_summary":"The paper sets out to bring Dominant Resource Fairness (DRF), the standard scheme for fairly sharing several resource types among users with heterogeneous demands, onto a blockchain, where the block gas limit makes DRF's iterative allocation loop impractical. Its algorithm, Autonomous Dominant Resource Fairness (ADRF), replaces the loop with three smart-contract functions, demand, claim, and update state, which users themselves execute in alternating block epochs, so gas cost depends on the number of resource types but not on the number of users. Measured on a local Ethereum test bed, all three functions scale linearly in the number of resources with near-perfect $R^2$ values, and the regressions stay far below the block gas limit even at hundreds of resources. What a sympathetic reader gains is a concrete recipe for loop-free, multi-resource fair allocation on-chain, and a measured case that the block gas limit need not confine blockchain resource management to fixed bundles.","feed_headline":"Fair multi-resource sharing on-chain, at gas cost flat in users","feed_subtitle":"Adapts Dominant Resource Fairness to smart contracts: gas grows with resource types, never with user count.","key_machinery":"The load-bearing object is the precomputed cycle count of PDRF, $k' = \\min_r \\left( ds'^*_r\\, r\\, p \\big/ \\sum_i ds'_i\\, d_{ir} \\right)$, computed once per epoch by a single loop over the resource vector in the update state function, with the per-user share $\\lfloor k\\, ds^*/ds_i \\rfloor \\cdot d_i$ evaluated client-side in the claim function. Two further mechanisms carry the 'autonomous' part: a $2 \\times m$ cyclic buffer holding reserves for two alternating resource pools so that demand registration (reading one pool) and claims (draining the other) run in parallel epochs, and an epoch-synchronisation scheme in which users are expected to call demand and claim in consecutive fixed-length block windows. The rearrangement of the $k'$ formula into precomputable aggregates, namely $p/ds^*$ and $\\sum_i p\\, d_{ir}/ds_i$, stored incrementally by each demand call, is what lets the contract avoid floating point arithmetic while keeping rounding error to at most one task.","core_discovery":"On the paper's own terms, the discovery is that DRF-style multi-resource fair allocation can run on a blockchain without any central allocation loop: ADRF is 'an implementation of pure PDRF' in which each user registers a demand vector in one epoch and claims a precomputed share in the next. The contract computes the number $k'$ of DRF cycles available before some resource is depleted, using fixed-point arithmetic with precision $p = 10^6$ to emulate floating point, and each user receives $\\lfloor k \\cdot ds^*/ds_i \\rfloor \\cdot d_i$ units, where $ds_i$ is her dominant share, the largest fraction of any single resource she requests, and $d_i$ her demand vector. The attached measurements give linear gas models $g_c = 15,130\\,m + 36,486$ for claim, $g_d = 13,616\\,m + 47,245$ for demand, and $g_u = 11,295\\,m + 23,539$ for update state, all with $R^2$ close to 1 up to $m = 100$ resource types, with user count absent from the cost. The fairness claim is inherited from PDRF: under discrete uniform demands, on average 47% of users get one task fewer than DRF would give and 0.06% are overallocated, and ADRF simply hands unallocated excess reserves to the next epoch.","pith_inferences":["A natural next test the paper leaves open is to measure ADRF's actual distance from DRF on-chain, including the effect of rolling excess reserves between epochs, rather than inheriting PDRF's off-chain statistics; if the one-task error bound survives rollover, the fairness claim is on firmer ground.","The same demand/claim division of labour could be applied to other iterative allocation schemes, such as variants of DRF with indivisibilities or dynamic weights, since the technique replaces any central loop with precomputed client-side shares.","The paper's aside that DRF is an implicit pricing mechanism, valuing most-demanded and least-supplied resources through the dominant share, suggests ADRF could double as an on-chain price-discovery rule for token ecosystems, a use case the paper mentions but does not develop."],"forward_implications":["Within Ethereum's 32,000,000 block gas limit, the measured regressions imply ADRF can manage well over 1,000 resource types, and hundreds comfortably.","Because per-call gas does not grow with the user set, the contract can serve arbitrarily many users without re-deployment or user-count-dependent gas costs.","Multi-resource allocation with heterogeneous demands becomes practical on-chain, replacing the fixed virtual-machine bundle abstraction that dominates the blockchain resource-allocation literature.","A weighted variant of ADRF costs only one extra division per function, so user- or resource-level weights can be added without approaching the gas limit.","Unused reserves from one epoch are not wasted; they roll into the next epoch's pool, which the paper argues fits blockchain tokens that users may save and spend later."],"supporting_citations":[{"why":"Defines Dominant Resource Fairness, the multi-resource fairness target that ADRF approximates on-chain.","marker":"[9]"},{"why":"Supplies the precomputation formula for DRF and the approximation statistics (47% short by one task, 0.06% overallocated) that ADRF inherits.","marker":"[11]"},{"why":"Establishes the autonomous demand/claim execution pattern for blockchain allocation that ADRF adapts from the single-resource setting.","marker":"[10]"},{"why":"Defines smart contracts and the block gas limit whose exhaustion risk motivates a loop-free design.","marker":"[2]"},{"why":"Documents that prior blockchain resource allocation assumes fixed virtual-machine bundles, the gap ADRF addresses.","marker":"[8]"},{"why":"Provides the published contract code and test data behind the reported gas measurements.","marker":"[44]"},{"why":"Supplies the testing framework used to collect the gas-cost measurements.","marker":"[46]"}],"fun_headline_variants":["Blockchain fair sharing: gas cost flat in users","ADRF: multi-resource fairness with user-count-free gas","On-chain DRF fairness with gas independent of users","Autonomous fair resource allocation without user gas growth"],"cache_read_input_tokens":17792,"weakest_assumption_plain":"The paper's fairness claim is borrowed, not measured: ADRF inherits PDRF's approximation of DRF, on average one task short for 47% of users and 0.06% overallocated, figures reported only for discrete uniform demands in a companion preprint, and the paper does not measure ADRF's own deviation from DRF.","fun_headline_variants_meta":{"raw":{"variants":["Blockchain fair sharing: gas cost flat in users","ADRF: multi-resource fairness with user-count-free gas","On-chain DRF fairness with gas independent of users","Autonomous fair resource allocation without user gas growth"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000519,"raw_usage":{"total_tokens":2619,"prompt_tokens":1151,"completion_tokens":1468,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":767,"completion_tokens_details":{"reasoning_tokens":1404}},"tokens_in":767,"tokens_out":1468,"duration_ms":12719,"temperature":1.0,"reasoning_tokens":1404,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:12:35.929647+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Deploy the published contract, draw the same discrete uniform demands as the paper, and run a reference DRF implementation on the same demand vectors: if noticeably more than 47% of users end up more than one task short of their DRF allocation, or if overallocation exceeds 0.06%, the inherited approximation bound fails for ADRF. The 'unlimited users' claim is separately checkable by holding resources fixed, doubling the number of registered users, and recording claim gas, which the paper claims stays flat.","supporting_citations":[{"cited_title":"Ghodsi, M","cited_arxiv_id":null,"evidence_quote":"Defines Dominant Resource Fairness, the multi-resource fairness target that ADRF approximates on-chain."},{"cited_title":"Metin, Precomputed dominant resource fairness, arXiv preprint arXiv:2507.08846 (2025)","cited_arxiv_id":null,"evidence_quote":"Supplies the precomputation formula for DRF and the approximation statistics (47% short by one task, 0.06% overallocated) that ADRF inherits."},{"cited_title":"Metin, C","cited_arxiv_id":null,"evidence_quote":"Establishes the autonomous demand/claim execution pattern for blockchain allocation that ADRF adapts from the single-resource setting."},{"cited_title":"Buterin, et al., Ethereum white paper, GitHub repository 1 (22-23) (2013) 5–7","cited_arxiv_id":null,"evidence_quote":"Defines smart contracts and the block gas limit whose exhaustion risk motivates a loop-free design."},{"cited_title":"Baranwal, D","cited_arxiv_id":null,"evidence_quote":"Documents that prior blockchain resource allocation assumes fixed virtual-machine bundles, the gap ADRF addresses."},{"cited_title":"Metin, blockchainFaucet (Oct","cited_arxiv_id":null,"evidence_quote":"Provides the published contract code and test data behind the reported gas measurements."},{"cited_title":"Hauser, Brownie documentation, accessed at 08.07.2025 (2020)","cited_arxiv_id":null,"evidence_quote":"Supplies the testing framework used to collect the gas-cost measurements."}],"review_version":1}