DAG consensus
Inclusion of Orphan Blocks
Inclusive
Yoad Lewenberg, Yonatan Sompolinsky, and Aviv Zohar (The Hebrew University of Jerusalem)
FC'15
Game theoretic model of the competition for fees between the nodes under the new protocol.
Show that in each case nodes randomize transaction selection from a wider range of transactions.
SPECTRE paper argues "does not avoid the security-scalability trade-off".
Prism paper argues "it does not achieve optimal throughput".
Conflux
Chenxing Li, Peilun Li, and Dong Zhou, Tsinghua University; Zhe Yang, Ming Wu, and Guang Yang, Conflux Foundation; Wei Xu, Tsinghua University; Fan Long, University of Toronto and Conflux Foundation; Andrew Chi-Chih Yao, Tsinghua University
Total ordering
1. Select pivot chain (kinda main chain) by GHOST, which defines the epoch
2. Sort blocks in each epoch in such a way that off-chain blocks referenced from the pivot chain block comes first
Therefore, finalization of a block in the pivot chain is finalization of the ordering
The analysis of this paper is light, just saying that safety and liveness is the same with GHOST
Explain liveness attack on PHANTOM
From Deconstructing the Blockchain to Approach Physical Limits,
Conflux’s reference links are not used to determine where to mine blocks or how to confirm them; they are only used to include side-chain blocks into the main chain to improve throughput. The main chain itself is selected by the GHOST rule
As a result, the Conflux’s security is limited by GHOST
Block has a deferred_state_root field: a commitment of the state a few epochs ago. May be reverted if the block ends up off the pivot chain Jute
SPECTRE, PHANTOM
https://gyazo.com/c12149442791c7db8179f8e813772cbd
Yonatan Sompolinsky, Yoad Lewenberg, and Aviv Zohar (The Hebrew University of Jerusalem, Israel)
Confirmation Times in SPECTRE Medium Transaction Selection Games in BlockDAGs Medium No liveness for conflicting txs
This is OK for payment but not OK for smart contract
Updated with proofs in 2020.2
To find k-cluster is NP-hard -> GHOSTDAG
Georgios Birmpas, Elias Koutsoupias (University of Oxford), Philip Lazos (Sapienza University of Rome), Francisco J. Marmolejo Cossío (University of Oxford)
FC'20
Contribution: Theoretical framework that captures a large family of DAG-based protocols (Bitcoin and SPECTRE)
Fairness: Miners earn block reward proportional to the computational resources they spend
Efficiency: Higher throughput of transactions (original goal of DAG-based protocols)
PoW efficiency: is the fraction of globally valid transactions that are present within the valid sub DAG of the block DAG, over all published transactions.
Fantômette
Sarah Azouvi, Patrick McCorry, and Sarah Meiklejohn (University College London)
PoS with incentive design: considers both rational and Byzantine adversaries (w/ punishment)
Caucus: VRF for private leader election (and use VDF as fallback)
Betting and fork-choice
Checkpointing with two quorum votes (justified/finalized) for long-range attacks (no proof of this?)
However, from this answer, it seems like normal subjective checkpoint Avalanche
Others
Emmanuelle Anceaume (IRISA) et al.
Kyle Butt, Derek Sorensen, and Michael Stay
Serguei Popov, Olivia Saa Paulo Finardi
Tweet about game theoritic issue by Alex (NEAR) Sharding and DAG
Design of sharding protocols is design of dependency DAGs