Pith. sign in

REVIEW 3 major objections 5 minor 9 references

Transaction Proximity: A Graph-Based Approach to Blockchain Fraud Prevention

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper claims that a wallet's hop distance to a regulated exchange separates legitimate stablecoin users from known exploiters, and that a one-hop EAI access rule could have blocked 83% of studied exploits.

desk verdict Promising compliance idea, but a contradictory definition of EAI distance makes the headline percentages uninterpretable until the authors say which direction they actually used. read the letter →

arxiv 2505.24284 v1 pith:5EWNCXGY submitted 2025-05-30 cs.CR cs.CEecon.GNq-fin.EC

Gordon Y. Liao , Ziming Zeng , Mira Belenkiy , Jacob Hirshman This is my paper
classification cs.CRcs.CEecon.GNq-fin.EC
keywords transactionproximityEasilyAttainableIdentitiesgraphanalysisblockchainfraudpreventionEthereumstablecoinsaccesscontrolDeFicompliance
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proposes a middle-ground fraud-deterrence tool for public blockchains: measure how close a wallet sits to centralized exchanges in the transaction graph, treat addresses within one hop as Easily Attainable Identities, and let protocols use that label to gate, monitor, or risk-score transactions. The authors aim to show that this one-hop test separates legitimate activity from known attackers because 91% of USDC wallet-to-wallet transfers above $2,000 involve at least one EAI, while 83% of known exploiter addresses are not EAIs. They also show that the label can be maintained and verified onchain at modest cost through three implementations with different privacy and gas tradeoffs. If the empirical numbers hold, the approach would give DeFi protocols a preventative, pseudonym-preserving alternative to reactive blocklisting and universal KYC.

What carries the argument

The central object is the directed transaction graph of Ethereum, with nodes as wallet addresses and an edge when one wallet sends at least $10 worth of ETH or a major stablecoin to another. Distance is computed by breadth-first search outward from a seed set of known centralized exchange addresses; an Easily Attainable Identity is an address at distance 0 (an exchange wallet itself) or distance 1 (a wallet that has directly received funds from such an exchange). This one-hop notion makes identity attainable in the sense that an exchange that knows its own customers can, when legally required, connect the pseudonymous address to a person. The same distance scale, extended to five hops, is what separates normal wallets from the known exploiter sample, and it also powers the proposed onchain implementations.

What would settle it

Recompute the distance distribution with an independently compiled, continuously updated exchange-address set and independently verified labels for the 431 exploiter addresses; if a material share of those exploiters falls within one hop after correcting missing or wrong exchange addresses, the 83% deterrence claim fails. A prospective test would deploy an EAI-gated token and observe whether exploiters route funds through an exchange hop to regain access.

Watch

Extended reading notes

Core claim

The paper's central claim is that on Ethereum, legitimate stablecoin activity stays physically close to centralized exchanges while known attackers do not. The empirical basis is a directed graph analysis as of May 31, 2024: 56% of USDC wallets with lifetime maximum balances exceeding $10,000 qualify as EAIs, 88% are within one hop of an EAI, and 91% of USDC wallet-to-wallet transfers above $2,000 involve at least one EAI. In contrast, 83% of 431 known exploiter addresses from the post-May 2022 period are not EAIs, and 21% are more than five hops from any regulated exchange. The authors conclude from this separation that a protocol-level rule requiring a transaction to involve at least one EAI would have deterred most past exploits while disturbing only a small share of ordinary large transfers.

Load-bearing premise

The whole result rests on the list of known exchange wallets being complete and correct, since every EAI distance and every reported percentage is measured from that list, and the paper concedes the list comes from public sources and may not be fully accurate.

Editorial extensions

If this is right

  • Protocols that require at least one party of a transaction to be an EAI would exclude a large majority of the 431 studied exploiter addresses, while allowing most ordinary large USDC transfers to proceed.
  • Because 88% of large USDC wallets are within one hop of an EAI, an EAI-based access rule would not force most existing users to change wallets or complete new identity procedures.
  • The three implementation approaches (on-chain registry with bit flags, off-chain registry with signatures, and Merkle-tree registry) put per-transaction verification costs in the $2.69 to $4.25 range at the stated gas prices, making the control embeddable in ERC-20 transfers.
  • EAI distance can function as a continuous risk signal beyond a binary gate, so lending, reputation, and governance protocols could tune their own thresholds rather than adopt a single global rule.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper, an obvious adversarial response is for an attacker to route funds through a compromised or low-KYC exchange account to become an EAI, so the framework's long-run value depends on exchange KYC quality and on detecting freshly created one-hop addresses.
  • Beyond the paper, the 91% traceability figure is a snapshot of current user behavior; if EAI gating were widely adopted, legitimate users might migrate to farther-from-exchange wallets, and the separation the framework relies on could shrink.
  • Beyond the paper, the approach could be tested prospectively by deploying EAI-gated tokens on a test network and measuring whether simulated exploiters launder proximity through exchanges, which would let protocols adjust hop thresholds before mainnet use.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes a graph-based compliance and fraud-deterrence framework for public blockchains. It constructs a directed transaction graph from Ethereum data, defines an 'Easily Attainable Identity' (EAI) as a wallet within one hop of a known centralized exchange address, and reports that most large USDC wallets and transactions are near EAIs while most known exploiter addresses are not. It then evaluates three on-chain/off-chain registry implementations with gas-cost benchmarks and argues that EAI-based access control can deter a large share of past exploits.

Significance. If the empirical claims withstand scrutiny, the paper offers a potentially useful middle-ground compliance primitive: a passive, privacy-preserving alternative to universal KYC that leverages existing centralized-exchange relationships. The contribution is strengthened by the large-scale dataset (206M nodes, 442M edges), the public GitHub repository for replication, and the concrete gas-cost comparison of registry implementations. The central idea is falsifiable and the paper candidly lists several limitations, including incomplete exchange coverage and the dynamic nature of transaction graphs. However, the current manuscript has two load-bearing problems: an unresolved directional inconsistency in the formal definition of EAI distance, and a logical gap between address-level exploiter statistics and the claim that transaction-level access control would block most exploits.

major comments (3)
  1. [Section 2.1, Eqs. (4)-(6); Section 4 code] The formal definition of d(v) is inconsistent with the prose and with the implementation sketch. Eq. (5) defines d(v) as the shortest directed path from v to an exchange e, so d(v)≤1 means v has sent funds directly to e (edge v→e). However, the text immediately after Eq. (6) says EAIs are wallets 'that have directly received funds from these exchange wallets' (edge e→v), and the dynamic flagging code in §4 marks the recipient when msg.sender is an exchange. In a directed graph these are different sets, and all headline percentages in §3.1–§3.4 (56%, 88%, 91%, 83%, 21%) depend on which orientation was actually used in the BFS. The authors must state the traversal direction used, align Eq. (5) with the actual computation, and rerun or reinterpret the statistics if the direction changes.
  2. [Section 3.4 and Section 5, conclusion item 3] The deterrence claim does not follow from the address-level statistic. The proposed access-control rule in Section 2.3 (Eq. (8)) classifies a transaction as involving an EAI if either the sender or the receiver is an EAI. An exploiter address being non-EAI does not imply that the exploiter's transactions fail this test, because the counterparty could be an EAI. The paper reports the EAI status of 431 exploiter addresses but not the EAI status of the actual exploit transactions. To support 'prevent most of these incidents,' the authors need to analyze the transactions associated with these exploits and report what fraction would have been blocked under each candidate rule.
  3. [Section 2.4 and Section 3.4] The completeness and accuracy of the exchange seed list and the Dune Analytics exploit labels are load-bearing for all measured percentages, but the paper provides no sensitivity analysis or external validation. Missing exchange addresses would shift measured distances upward and could inflate the apparent separation between exploiters and ordinary wallets, so the acknowledged 'may not be fully accurate' seed list is more than a minor caveat. Similarly, the paper does not describe how the 431 exploit addresses were vetted or how the exchange-hack cases mentioned in footnote 2 would be excluded. The authors should report the size and provenance of the exchange list, test robustness to subsets of exchange seeds, and describe the curation and exclusion procedure for exploit labels.
minor comments (5)
  1. [Section 2.2, step 3] The BFS description should specify whether the traversal follows outgoing or incoming edges; this is related to the directional ambiguity in Eq. (5) and should be made explicit even after the formal definition is corrected.
  2. [Table 1 caption] The table uses 'EAI distance' with values 0 through 5 and a '5+' column; please define whether 5+ means 'no directed path found within 5 hops' or 'distance at least 5'.
  3. [Footnote 2] The footnote states that exchange-hack addresses should be excluded from EAI assignment, but the paper does not describe how such exclusions are implemented; please provide the exact rule and how it was applied in the empirical analysis.
  4. [Section 4 code listings] The code samples contain formatting artifacts (for example, spaces within identifiers such as 'e a i R e g i s t r y' and 'b a l a n c e A n d S t a t u s F l a g s') that should be cleaned up for readability.
  5. [Table 4] The first row of the 'Whitelist Size' column is missing; use '0' or 'N/A' for the on-chain registry row.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the EAI classification is an external graph measurement and the exploiter labels are independent; the only self-citation is a non-load-bearing implementation reference, though an unresolved direction inconsistency in Eq. (5) is a correctness risk.

full rationale

The paper's empirical claims are measurements rather than fitted derivations. EAI status is defined by graph distance from a seed set of exchange addresses (Eqs. 5-6), and exploiter status comes from independent Dune Analytics labels; neither set is fitted to the other. The 83% non-EAI exploiter fraction and the 91% EAI-involved transaction fraction are computed statistics, not quantities solved from the model, so there is no equation in which the input reappears as the output. The deterrence claim is an extrapolation from those statistics, not a prediction forced by construction. The only self-citation is the reference to Circle's FiatTokenV2_2.sol contract [Financial, 2023] as an implementation precedent for bit-flag blocklisting; this is an implementation detail and does not carry the central argument. The paper itself discloses that the exchange seed list comes from public sources and 'may not be fully accurate,' which is a data-quality limitation, not circularity. A more serious issue is a definitional inconsistency: Eq. (5) defines d(v) as the shortest directed path from wallet v to an exchange (v to e), while Section 2.1's prose and Section 4's code define an EAI as an address that received funds from an exchange (e to v). These are opposite orientations in the directed graph, so the reported percentages are attached to an ambiguous predicate. This undermines reproducibility but is not a circular reduction of the claim to its inputs. Overall, the derivation chain is self-contained against external benchmarks, and the circularity score is low.

Assumptions & free parameters 4 free parameters · 4 assumptions · 1 invented entities

The framework rests on the accuracy of external anchor data (exchange addresses and exploit labels), the interpretability of on-chain proximity as identity attainability, and a set of hand-chosen thresholds. No external data or machine-checked proofs are provided, so the reader must take these on faith.

free parameters (4)
  • Edge threshold = $10
    Eq. (4) creates an edge only for transfers of at least $10; no sensitivity analysis is shown.
  • Large wallet threshold = $10,000 max lifetime balance
    Section 3.1 defines 'large' as max lifetime USDC balance > $10,000 to match BSA reporting; the 56% and 88% claims depend on this cutoff.
  • Transaction size threshold = $2,000
    Section 3.3 reports 91% for USDC transfers exceeding $2,000; the percentage changes with the cutoff.
  • BFS hop limit = 5 hops
    Section 2.2 truncates BFS at 5 hops; addresses beyond that are aggregated as 5+, and the 98.2% coverage statement conflicts with Table 1's 5+ counts.
assumptions (4)
  • domain assumption The curated list of centralized exchange addresses is complete and correct.
    Used as BFS seeds in Section 2.2; the paper notes it may be incomplete, which would systematically change all EAI distances.
  • domain assumption A wallet within one hop of an exchange can be identified through that exchange's KYC records.
    This is the definitional basis of EAI (Eq. 6) and the policy value of the framework; the paper flags dependence on exchange KYC effectiveness.
  • domain assumption The exploiter labels from Dune Analytics are accurate.
    Section 3.4 uses 431 labeled exploiter addresses; incorrect labels would bias the 83% and 21% claims.
  • standard math Standard graph algorithms and Ethereum data extraction are reliable.
    BFS complexity and data extraction are unremarkable, but no code is shipped to confirm the implementation.
invented entities (1)
  • Easily Attainable Identity (EAI) status independent evidence
    purpose: A binary wallet classification used as a privacy-preserving compliance signal for access control.
    EAI status is derived entirely from public on-chain data, so any third party can recompute it; however the paper does not publish the exact algorithm or exchange list used.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Transaction Proximity: A Graph-Based Approach to Blockchain Fraud Prevention." pith.science (2026). https://pith.science/paper/5EWNCXGY

@misc{pith2026250524284,
  author       = {Pith},
  title        = {Pith review of: Transaction Proximity: A Graph-Based Approach to Blockchain Fraud Prevention},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5EWNCXGY}},
  note         = {Machine review of arXiv:2505.24284}
}
abstract

This paper introduces a fraud-deterrent access validation system for public blockchains, leveraging two complementary concepts: "Transaction Proximity", which measures the distance between wallets in the transaction graph, and "Easily Attainable Identities (EAIs)", wallets with direct transaction connections to centralized exchanges. Recognizing the limitations of traditional approaches like blocklisting (reactive, slow) and strict allow listing (privacy-invasive, adoption barriers), we propose a system that analyzes transaction patterns to identify wallets with close connections to centralized exchanges. Our directed graph analysis of the Ethereum blockchain reveals that 56% of large USDC wallets (with a lifetime maximum balance greater than \$10,000) are EAI and 88% are within one transaction hop of an EAI. For transactions exceeding \$2,000, 91% involve at least one EAI. Crucially, an analysis of past exploits shows that 83% of the known exploiter addresses are not EAIs, with 21% being more than five hops away from any regulated exchange. We present three implementation approaches with varying gas cost and privacy tradeoffs, demonstrating that EAI-based access control can potentially prevent most of these incidents while preserving blockchain openness. Importantly, our approach does not restrict access or share personally identifiable information, but it provides information for protocols to implement their own validation or risk scoring systems based on specific needs. This middle-ground solution enables programmatic compliance while maintaining the core values of open blockchain.

Figures

Figures reproduced from arXiv: 2505.24284 by the authors.

Figure 1
Figure 1. Transaction Proximity and Easily Attainable Identities (EAIs). This diagram illustrates [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Distance to EAIs for USDC wallets with maximum lifetime balances greater than $10,000. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Distance to EAIs for USDC wallet-to-wallet transactions greater than $2,000. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Distance to EAIs for Known Exploiter Addresses versus Typical Wallets. The distribution [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

9 extracted references · 7 canonical work pages

  1. [1]

    Disclosure, dapps and defi

    Chris Brummer. Disclosure, dapps and defi. Stanford Journal of Blockchain Law and Policy, March 2022. URL https://ssrn.com/abstract=4065143. Forthcoming

    work page 2022

  2. [2]

    Blockchain resource pricing

    Vitalik Buterin, Zoë Hitzig, and E Glen Weyl. Blockchain resource pricing. ACM Transactions on Economics and Computation, 10 0 (4): 0 1--37, 2022

    work page 2022

  3. [3]

    Blockchain privacy and regulatory compliance: Towards a practical equilibrium

    Vitalik Buterin, Jacob Illum, Matthias Nadler, Fabian Sch \"a r, and Ameen Soleimani. Blockchain privacy and regulatory compliance: Towards a practical equilibrium. Blockchain: Research and Applications, 5 0 (1): 0 100176, 2024. doi:10.1016/j.bcra.2023.100176

    arXiv 2024

  4. [4]

    The chainalysis 2024 crypto crime report

    Chainalysis . The chainalysis 2024 crypto crime report. Report, Chainalysis, 2024. URL https://go.chainalysis.com/crypto-crime-2024.html

    work page 2024

  5. [5]

    A note on privacy and compliance for stablecoins

    Darrell Duffie, Odunayo Olowookere, and Andreas Veneris. A note on privacy and compliance for stablecoins. Technical report, Stanford University, York University, University of Toronto, March 2025. Preliminary Draft

    work page 2025

  6. [6]

    Fiattokenv2\_2.sol, 2023

    Circle Internet Financial. Fiattokenv2\_2.sol, 2023. URL https://github.com/circlefin/stablecoin-evm/blob/master/contracts/v2/FiatTokenV2_2.sol

    work page 2023

  7. [7]

    How to design a compliant, privacy-preserving fiat stablecoin via zero-knowledge proofs

    Jonas Gross, Johannes Sedlmeir, and Simon Seiter. How to design a compliant, privacy-preserving fiat stablecoin via zero-knowledge proofs. Technical report, etonec GmbH, University of Luxembourg, Hauck Aufhäuser Lampe, 2022. Technical Report

    work page 2022

  8. [8]

    Quantitative analysis of the full bitcoin transaction graph

    Dorit Ron and Adi Shamir. Quantitative analysis of the full bitcoin transaction graph. In Financial Cryptography and Data Security, pages 6--24, Berlin, Heidelberg, 2013. Springer. Analysis conducted on Bitcoin blockchain data through May 13, 2012

    work page 2013

Show all 9 references
  1. [9]

    Taher, Siddeeq Y

    Shimal Sh. Taher, Siddeeq Y. Ameen, and Jihan A. Ahmed. Advanced fraud detection in blockchain transactions: An ensemble learning and explainable ai approach. Engineering, Technology & Applied Science Research, 14 0 (1): 0 12822--12830, 2024. doi:10.48084/etasr.6641

    2024 doi

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.