# Researchers map trade-offs across Tendermint consensus designs

Published 2026-11-13 · Crypto.world Team · Crypto.world

> A new survey classifies research that changes Tendermint’s protocol choices and the properties those changes affect. It groups the work around performance, validator accountability and protection of exploitable information.

## A map of protocol trade-offs

A new survey organizes research on Tendermint-style consensus by the protocol choices each study changes and the properties those changes affect. In the preprint submitted to [arXiv](https://arxiv.org/abs/2611.15847) on November 11, 2026, Sanaz Taheri-Boshrooyeh and Yahya Hassanzadeh-Nazarabadi say the work brings together studies that had not previously been surveyed as one Tendermint research family.

The survey treats Tendermint as a baseline rather than focusing on a particular implementation. It traces how the same underlying protocol is implemented first by Tendermint Core and now by CometBFT, and describes how the Application Blockchain Interface separates consensus from application execution.

The baseline is organized around seven choices: vote dissemination, signatures, locking and round progression, proposer rotation, the mempool, bonding and slashing, and reward distribution. The classification also records whether a result holds under the baseline’s assumptions or needs a stronger fault or timing model.

## Performance and accountability

The authors group the surveyed work’s aims into three areas: performance, validator accountability and protection of information that could be exploited. Performance proposals address vote traffic and the sequential process of deciding blocks; accountability proposals seek verifiable evidence of harmful actions or mechanisms that prevent them.

One example is Semantic Gossip, which changes how consensus votes are relayed. The survey describes its filtering of votes already processed or no longer needed for a quorum, and its bundling of votes into fewer messages; the votes remain individually signed when unpacked.

For a Go prototype evaluated against classic gossip, the survey reports throughput gains ranging from 1.24 times at 32 validators to 3.42 times at 128 validators. At 128 validators, it reports median latency 2.67 times better than classic gossip. The experiment ran multiple consensus instances concurrently over the same validators, a setup the paper notes was used to raise message load.

Other approaches in the classification include pipelining consecutive block heights, which the survey says can raise throughput while deferring each block’s commit. It also covers proposals for verifiable-random-function-based proposer selection, encrypted transactions intended to limit front-running, and external-chain checkpoints used in work on long-range attacks and liveness.

## Assumptions and remaining questions

The paper emphasizes that proposed gains may rely on assumptions beyond the baseline. For example, some surveyed results require synchrony, trusted hardware or an external chain; the authors mark such dependencies in their classification rather than treating them as properties available under the baseline model.

That model assumes fewer than one-third of total voting power is controlled by a Byzantine adversary, along with partial synchrony and a gossip communication property. The paper describes these as conditions for the baseline’s safety and liveness guarantees, not as findings that every modification preserves those guarantees.

The survey closes with open questions on attributing a proposer’s omission of a precommit, detecting stalls when blocks have a size limit, making the “amnesia attack” attributable, and improving proposer availability within a height. Those questions identify gaps in the surveyed approaches under their stated fault and timing models.

## Sources

1. Original source · Preprint [SoK: Design Choices in Tendermint BFT Consensus](https://arxiv.org/abs/2611.15847) arXiv:2611.15847

## Related stories

- [Researchers propose DEX approvals before consensus](https://crypto.world/news/researchers-propose-dex-approvals-before-consensus)
- [Researchers update Angelfish consensus preprint](https://crypto.world/news/researchers-update-angelfish-consensus-preprint)
- [Isotaxis Targets MEV With Asynchronous Transaction Ordering](https://crypto.world/news/isotaxis-mev-resistant-asynchronous-consensus)
- [Ethereum Draft Targets a Fourfold Cut in Finality Time](https://crypto.world/news/ethereum-eip-8453-decoupled-consensus-finality)
- [Solana Sets Alpenglow Mainnet Switch for Oct. 21](https://crypto.world/news/solana-alpenglow-mainnet-switch-october-21)

## Explore

[More articles](https://crypto.world/news)

---

Canonical: https://crypto.world/news/researchers-map-tendermint-consensus-design-tradeoffs
