pith. sign in

arxiv: 2606.26292 · v1 · pith:242VBK4Unew · submitted 2026-06-24 · 💻 cs.DC

FinWhale: An Optimally Resilient Two-Round Terminating DAG Protocol

Razya Ladelsky , Roy Friedman This is my paper

Pith reviewed 2026-06-26 01:27 UTC · model grok-4.3

classification 💻 cs.DC
keywords DAG consensusByzantine Fault Tolerancefast pathtwo-round terminationpartial synchronyevidence blocksuncertified DAG
0
0 comments X

The pith

FinWhale adds a two-message-delay fast path to DAG-based Byzantine consensus while preserving safety across local views.

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

The paper introduces FinWhale as the first DAG-based Byzantine Fault Tolerant protocol that can commit decisions after only two message delays when conditions allow. It does this by extending an existing uncertified DAG protocol with a new fast-path commit rule that runs alongside the original slower rules. New structures called fast path evidence blocks let validators reconcile the two paths even when their local DAG views differ. The design works in the partially synchronous model with the minimal validator count known to support fast consensus. If correct, the result shows that optimal-latency fast paths can be added to DAG protocols without sacrificing their throughput advantages.

Core claim

FinWhale is the first DAG based Byzantine Fault Tolerant protocol with a two message delay fast path. It extends Mysticeti with a novel fast path commit mechanism that safely coexists with the protocol's original slow path rules. To preserve safety across different local DAG views the protocol introduces new commit structures based on fast path evidence blocks enabling validators to combine fast path and slow path reasoning consistently. The protocol uses n equals three f plus two p minus one validators, tolerates up to f Byzantine faults, and achieves fast termination whenever at most p validators fail during the fast path.

What carries the argument

The novel fast path commit mechanism based on fast path evidence blocks, which lets validators merge fast-path and slow-path decisions consistently.

If this is right

  • The protocol reaches optimal two-message-delay termination under favorable conditions while retaining the high throughput of uncertified DAGs.
  • It matches the known lower bound on the number of validators required for fast Byzantine consensus.
  • Validators tolerate up to f total faults yet still obtain fast termination when at most p faults occur on the fast path.
  • Optimal-latency fast paths become compatible with uncertified DAG consensus protocols in the partial-synchrony model.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same evidence-block technique could be tested on other uncertified DAG protocols to check whether the fast-path addition remains safe.
  • Implementations would need to verify that storing and checking evidence blocks does not materially increase per-round communication cost.
  • The coexistence rule might allow future DAG protocols to offer both fast and slow paths without requiring separate voting structures.

Load-bearing premise

The fast-path commit mechanism can safely coexist with the original slow-path rules without producing conflicting decisions under differing local DAG views.

What would settle it

An execution trace in which one validator commits a value via the fast path while another validator, seeing the same evidence blocks, later commits a conflicting value via the slow path.

read the original abstract

DAG based Byzantine Fault Tolerant protocols provide high throughput consensus under partial synchrony but existing DAG protocols still require at least three message delays to commit decisions. In contrast fast path Byzantine Fault Tolerant protocols can achieve optimal two message delay termination under favorable conditions though they do not naturally extend to DAGs. We present FinWhale the first DAG based Byzantine Fault Tolerant protocol with a two message delay fast path. FinWhale extends Mysticeti with a novel fast path commit mechanism that safely coexists with the protocol's original slow path rules. To preserve safety across different local DAG views we introduce new commit structures based on fast path evidence blocks enabling validators to combine fast path and slow path reasoning consistently. FinWhale operates in the partially synchronous model with n equals three f plus two p minus one validators matching the known lower bound for fast Byzantine consensus. The protocol tolerates up to f Byzantine faults and achieves fast termination whenever at most p validators fail during the fast path where p is between one and f. Our results show that optimal latency fast paths can be integrated into uncertified DAG consensus protocols.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper presents FinWhale, the first DAG-based BFT protocol achieving two message delay fast-path termination. It extends Mysticeti with a novel fast-path commit mechanism using fast path evidence blocks that is claimed to safely coexist with the original slow-path rules. The protocol runs in partial synchrony with n=3f+2p-1 validators (matching the known lower bound), tolerates f Byzantine faults, and achieves fast termination when at most p validators fail on the fast path (1≤p≤f).

Significance. If the safety and liveness arguments hold, the result would be significant: it is the first integration of an optimal-latency fast path into an uncertified DAG consensus protocol while preserving the DAG's high-throughput properties and matching the resilience lower bound.

major comments (2)
  1. The abstract asserts that fast-path evidence blocks enable consistent combination of fast- and slow-path reasoning across differing local DAG views, yet provides no proof sketch, invariant, or argument establishing that the new commit structures preserve safety when a validator observes only a subset of the evidence blocks. This coexistence claim is load-bearing for the central contribution.
  2. The model parameter n=3f+2p-1 is stated to match the known lower bound for fast Byzantine consensus, but the manuscript supplies neither a reference to the bound nor a derivation showing that the FinWhale construction meets it exactly when p>1; without this, the optimality claim cannot be verified.
minor comments (1)
  1. The abstract mentions 'our results show' optimal latency integration but contains no experimental data, latency measurements, or throughput figures; these should be added or the claim qualified.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading and constructive comments. We address each major comment below and will revise the manuscript to strengthen the presentation of the safety argument and optimality claim.

read point-by-point responses

  1. Referee: The abstract asserts that fast-path evidence blocks enable consistent combination of fast- and slow-path reasoning across differing local DAG views, yet provides no proof sketch, invariant, or argument establishing that the new commit structures preserve safety when a validator observes only a subset of the evidence blocks. This coexistence claim is load-bearing for the central contribution.

    Authors: The abstract is a concise summary; the full safety proof, including the invariants ensuring consistent fast- and slow-path reasoning even with partial observation of evidence blocks, appears in the body of the manuscript. To address the concern, we will revise the abstract to include a one-sentence high-level sketch of the key invariant that guarantees safety of the combined commit rules. This makes the central claim more self-contained while preserving the abstract's brevity. revision: yes

  2. Referee: The model parameter n=3f+2p-1 is stated to match the known lower bound for fast Byzantine consensus, but the manuscript supplies neither a reference to the bound nor a derivation showing that the FinWhale construction meets it exactly when p>1; without this, the optimality claim cannot be verified.

    Authors: We agree that an explicit reference and short derivation are needed for verifiability. We will add a citation to the established lower-bound result for fast-path Byzantine consensus and include a brief derivation in the model section showing why n=3f+2p-1 is tight and how the protocol meets the bound for any 1≤p≤f by tolerating up to p fast-path failures while remaining resilient to f faults overall. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper presents a protocol construction (FinWhale) that extends Mysticeti with a fast-path commit mechanism using evidence blocks. No derivation chain, predictions, fitted parameters, or self-referential definitions appear in the abstract or described claims. The work is a design and safety argument for a new BFT protocol under partial synchrony, not a reduction of results to inputs by construction. Self-citations (if any) are not load-bearing for a central mathematical claim. This matches the default expectation for a construction paper.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The protocol rests on standard domain assumptions from BFT literature and introduces one new entity to support the fast path.

axioms (1)
  • domain assumption Partially synchronous communication model with n = 3f + 2p - 1 validators
    Invoked in the abstract as the setting that matches the known lower bound for fast Byzantine consensus.
invented entities (1)
  • fast path evidence blocks no independent evidence
    purpose: To enable consistent combination of fast-path and slow-path reasoning across local DAG views
    New commit structures introduced to support the fast path commit mechanism while preserving safety.

pith-pipeline@v0.9.1-grok · 5723 in / 1258 out tokens · 26044 ms · 2026-06-26T01:27:41.530807+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

19 extracted references · 13 canonical work pages

  1. [1]

    Shoal++: High throughput dag bft can be fast!ArXiv, abs/2405.20488,

    2 Balaji Arun, Zekun Li, Florian Suri-Payer, Sourav Das, and Alexander Spiegelman. Shoal++: High throughput dag bft can be fast!ArXiv, abs/2405.20488,

    arXiv

  2. [2]

    4 Leemon Baird

    arXiv:2310.14821. 4 Leemon Baird. The swirlds hashgraph consensus algorithm: Fair, fast, byzantine fault tolerance. Technical Report SWIRLDS-TR-2016-01, Swirlds, Inc.,

    arXiv 2016

  3. [3]

    Internet computer consensus

    5 Jan Camenisch, Manu Drijvers, Timo Hanke, Yvonne-Anne Pignolet, Victor Shoup, and Dominic Williams. Internet computer consensus. InProceedings of the 2022 ACM Symposium on Principles of Distributed Computing, PODC’22, page 81–91, New York, NY, USA,

    2022

  4. [4]

    6 George Danezis, Lefteris Kokoris-Kogias, Alberto Sonnino, and Alexander Spiegelman

    Association for Computing Machinery.doi:10.1145/3519270.3538430. 6 George Danezis, Lefteris Kokoris-Kogias, Alberto Sonnino, and Alexander Spiegelman. Narwhal and tusk: a dag-based mempool and efficient BFT consensus. In Yérom-David Bromberg, Anne-Marie Kermarrec, and Christos Kozyrakis, editors,EuroSys ’22: Seventeenth European Conference on Computer Sys...

    work page doi:10.1145/3519270.3538430 2022

  5. [5]

    Journal of the ACM35(2), 288–323 (1988).https://doi.org/10 .1145/42282.42283

    7 Cynthia Dwork, Nancy Lynch, and Larry Stockmeyer. Consensus in the presence of partial synchrony.J. ACM, 35(2):288–323, April 1988.doi:10.1145/42282.42283. R. Ladelsky and R. Friedman 25 8 Roy Friedman, Achour Mostefaoui, and Michel Raynal. Simple and Efficient Oracle-Based Con- sensus Protocols for Asynchronous Byzantine Systems . InReliable Distribute...

    work page doi:10.1145/42282.42283 1988

  6. [6]

    URL: https://doi.ieeecomputersociety.org/10.1109/RELDIS.2004.1353024, doi: 10.1109/RELDIS.2004.1353024

    IEEE Computer So- ciety. URL: https://doi.ieeecomputersociety.org/10.1109/RELDIS.2004.1353024, doi: 10.1109/RELDIS.2004.1353024. 9 Adam Gągol, Damian Leśniak, Damian Straszak, and Michał Świętek. Aleph: Efficient atomic broadcast in asynchronous networks with byzantine nodes. InProceedings of the 1st ACM Conference on Advances in Financial Technologies, A...

    work page doi:10.1109/reldis.2004.1353024 2004

  7. [7]

    10 Philipp Jovanovic, Lefteris Kokoris-Kogias, Bryan Kumara, Alberto Sonnino, Pasindu Tennage, and Igor Zablotchi

    Association for Computing Machinery.doi:10.1145/3318041.3355467. 10 Philipp Jovanovic, Lefteris Kokoris-Kogias, Bryan Kumara, Alberto Sonnino, Pasindu Tennage, and Igor Zablotchi. Mahi-mahi: Low-latency asynchronous bft dag-based consensus. In2025 IEEE 45th International Conference on Distributed Computing Systems (ICDCS), pages 549–559, 2025.doi:10.1109/...

    work page doi:10.1145/3318041.3355467 2025

  8. [8]

    Cordial miners: Fast and efficient consensus for every eventuality

    12 Idit Keidar, Oded Naor, Ouri Poupko, and Ehud Shapiro. Cordial miners: Fast and efficient consensus for every eventuality. In Rotem Oshman, editor,37th International Symposium on Distributed Computing, DISC 2023, October 10-12, 2023, L’Aquila, Italy, volume 281 of LIPIcs, pages 26:1–26:22. Schloss Dagstuhl - Leibniz-Zentrum für Informatik,

    2023

  9. [9]

    14 Petr Kuznetsov, Andrei Tonkikh, and Yan X Zhang

    Proceedings., pages 262–267, 2002.doi:10.1109/RELDIS.2002.1180196. 14 Petr Kuznetsov, Andrei Tonkikh, and Yan X Zhang. Revisiting optimal resilience of fast byzantine consensus. InProceedings of the 2021 ACM Symposium on Principles of Distributed Computing, PODC’21, page 343–353, New York, NY, USA,

    work page doi:10.1109/reldis.2002.1180196 2002

  10. [10]

    2021 , isbn =

    Association for Computing Machinery.doi:10.1145/3465084.3467924. 15 Dahlia Malkhi, Chrysoula Stathakopoulou, and Maofan Yin. Bbca-chain: One-message, low latency bft consensus on a dag.arXiv preprint arXiv:2310.06335,

    work page doi:10.1145/3465084.3467924

  11. [11]

    Fast byzantine consensus.IEEE Trans

    16 Jean-Philippe Martin and Lorenzo Alvisi. Fast byzantine consensus.IEEE Trans. Dependable Secur. Comput., 3(3):202–215, July 2006.doi:10.1109/TDSC.2006.35. 17 Nikita Polyanskii, Sebastian Mueller, and Ilya Vorobyev. Making uncertified DAG BFT provably live with linear payload and quadratic metadata communication. Cryptology ePrint Archive, Paper 2025/567,

    work page doi:10.1109/tdsc.2006.35 2006

  12. [12]

    18 Longfei Qiu, Jingqi Xiao, and Zhong Shao

    URL:https://eprint.iacr.org/2025/567. 18 Longfei Qiu, Jingqi Xiao, and Zhong Shao. Mechanized Safety and Liveness Proofs for the Mysticeti Consensus Protocol under the LiDO-DAG Framework . In2026 IEEE Symposium on Security and Privacy (SP), pages 1522–1541, Los Alamitos, CA, USA, May

    2025

  13. [13]

    doi: 10.1109/SP63933.2026.00001

    IEEE Computer Society. URL: https://doi.ieeecomputersociety.org/10.1109/SP63933.2026. 00009,doi:10.1109/SP63933.2026.00009. 19 Ehud Shapiro. Brief Announcement: Grassroots Distributed Systems: Concept, Examples, Implementation and Applications. In Rotem Oshman, editor,37th International Symposium on Distributed Computing (DISC 2023), volume 281 ofLeibniz ...

    work page doi:10.1109/sp63933.2026 2026

  14. [14]

    URL:https://drops.dagstuhl.de/entities/document/10.4230/ LIPIcs.DISC.2023.47,doi:10.4230/LIPIcs.DISC.2023.47

    Schloss Dagstuhl – Leibniz- Zentrum für Informatik. URL:https://drops.dagstuhl.de/entities/document/10.4230/ LIPIcs.DISC.2023.47,doi:10.4230/LIPIcs.DISC.2023.47. 20 Victor Shoup. Sing a Song of Simplex. In Dan Alistarh, editor,38th International Symposium on Distributed Computing (DISC 2024), volume 319 ofLeibniz International Proceedings in Informatics (...

    work page doi:10.4230/lipics.disc.2023.47 2023

  15. [15]

    URL:https://drops.dagstuhl.de/entities/document/10.4230/ LIPIcs.DISC.2024.37,doi:10.4230/LIPIcs.DISC.2024.37

    Schloss Dagstuhl – Leibniz- Zentrum für Informatik. URL:https://drops.dagstuhl.de/entities/document/10.4230/ LIPIcs.DISC.2024.37,doi:10.4230/LIPIcs.DISC.2024.37. 21 Victor Shoup, Jakub Sliwinski, and Yann Vonlanthen. Kudzu: Fast and Simple High- Throughput BFT. In Dariusz R. Kowalski, editor,39th International Symposium on Distributed 26 FinWhale: An Opti...

    work page doi:10.4230/lipics.disc.2024.37 2024

  16. [16]

    URL: https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.DISC

    Schloss Dagstuhl – Leibniz-Zentrum für Informatik. URL: https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.DISC. 2025.42,doi:10.4230/LIPIcs.DISC.2025.42. 22 Nibesh Shrestha, Rohan Shrothrium, Aniket Kate, and Kartik Nayak. Sailfish: Towards improving latency of dag-based bft.Cryptology ePrint Archive,

    work page doi:10.4230/lipics.disc 2025

  17. [17]

    24 Alexander Spiegelman, Neil Giridharan andF Alberto Sonnino, and Lefteris Kokoris-Kogias

    Springer-Verlag.doi:10.1007/978-3-540-87779-0_30. 24 Alexander Spiegelman, Neil Giridharan andF Alberto Sonnino, and Lefteris Kokoris-Kogias. Bullshark: DAG BFT protocols made practical. In Heng Yin, Angelos Stavrou, Cas Cremers, and Elaine Shi, editors,Proceedings of the 2022 ACM SIGSAC Conference on Computer and Communications Security, CCS 2022, Los An...

    work page doi:10.1007/978-3-540-87779-0_30 2022

  18. [18]

    Shoal: Improving DAG- BFT latency and robustness.CoRR, abs/2306.03058,

    25 Alexander Spiegelman, Balaji Aurn, Rati Gelashvili, and Zekun Li. Shoal: Improving DAG- BFT latency and robustness.CoRR, abs/2306.03058,

    arXiv

  19. [19]

    Association for Computing Machinery.doi:10.1145/3652892.3700788. R. Ladelsky and R. Friedman 27 Algorithm 1DAG Construction for validatorvi, Part 1 1:Global variables: 2:buffer←∅ 3:blocksToPropose←∅▷Valid blocks of transactions from clients 4:δLT←2∆ 5:n←3f+ 2p−1 6:r decided←0▷The most recent round in the sequence of decided leaders 7:r highest←0▷The highe...

    work page doi:10.1145/3652892.3700788