Author: Marco González
Editor: Randall Roland
The EOS Network and underlying Antelope.IO protocol continue to push the envelope of what it means to lead and innovate in the blockchain space. A recent Fireside Chat with a notable Antelope developer (Gnome) illustrates the point.
Instant finality (IF) and inter-blockchain communication (IBC) is at the heart of the discussion about the soon-to-be-released Antelope Leap 5.0. Behind IF is HotStuff. IF and IBC profoundly impact the Antelope ecosystem and the high-performance EOS EVM.
About the HotStuff Consensus Algorithm
HotStuff is a product of VMware (2018). Gnome talked about a new, major upgrade for Antelope. He’s quoted throughout this article. Gnome works with both EOS and UX teams. He introduced the benefits of HotStuff with these words:
“we're talking faster finality better crypto intrinsics new consensus possibilities”
A HotStuff public overview is currently underway. It will cover both, what HotStuff is, and how it fits into the big picture for AntelopeIO and EOS.
Original Concept
VMware’s HotStuff protocol solves state replication across different servers, environments, and networks. HotStuff is touted as:
“...first partially synchronous BFT replication protocol exhibiting these combined properties.”
Early Development and Notable Implementations
HotStuff started as a building block algorithm. After a year or two of development, the protocol caught the attention of Facebook’s Libra team. What’s intriguing is that a company as large as FB experimented with HotStuff.
Other companies also saw promise with HotStuff. Aptos and SUI run implementations.
Applications for EOS
HotStuff is a proven solution with decidedly more promise than is in use today. HotStuff offers benefits for both technical and governance applications. For example, Gnome said:
“...increases the range of different …consensus mechanisms that can be built now …with an algorithm…”
Around the 8-minute mark in the Fireside Chat, things got interesting. Gnome describes EOS’ current model- an implementation that already leads the industry in speed, performance, and security. He describes how the rounds of block production take about 3 minutes. In the current system, doubling the number of block producers would double the round times to 6 minutes. The reference here is a new block’s irreversibility on the chain, not transaction speeds.
Instant Finality and IBC
With HotStuff, the approach can be different:
“...instead of waiting for …an implicit confirmation by a [single] block producer [to] agree on a specific block …on top of the chain when it's their turn …it [HotStuff] essentially works by broadcasting the block to everyone and then having all … block producers…broadcast a message [in agreement]...”
The result is a constant, split-second, and irreversible addition to the chain. HotStuff’s consensus model holds up regardless of the number of block producers. Hence, a potentially new consensus model expanded from just 21 active BPs.
Implementing HotStuff was never the objective. Solving the instant finality (IF) problem was. After investigating methods “to reach a faster time to finality” it became clear that HotStuff was the best solution.
HotStuff benefits for inter-blockchain communication (IBC) are immense. As stated, current IBC transfers and transactions take about three minutes. Gnome confirms the IBC advantage on the Fireside Chat:
“...one of the key benefits of adding …Instant Finality and HotStuff consensus is that now… IBC can also benefit…”
He goes on to say about user experience:
“... a Web 2.0 experience.. now we get the best of both worlds…”
Gnome describes how Antelope’s IF combines ownership, censorship resistance, and immutability with the advancements made by Web 2.0 in fast, centralized systems. It’s web-3.0 at Web-2.0 speed and convenience.
Aggregate Signatures and Realizing Blockchain’s Potential
Gnome briefly mentions aggregate signatures before fielding a question. As early as 2013 and a few years before HotSuff, a discussion about Boneh–Lynn–Shacham (BLS) signatures took place on Bitcoin Forum (Bitcointalk.org):
“...an Anonymous Author wrote a brief paper on using BLS signatures to achieve a… non-interactive coinjoin… it also could have useful anti-censorship properties… Another application… is using them to reduce blocksize… transaction sizes… Aggregate BLS signatures… compatible with 'efficient' fraud proofs…”
A decade ago, long before Antelope, obstacles presented themselves:
“...though the cost of doing probably makes it not a win…downside of this… the BLS signatures involve Pairing Cryptography-- less mature security assumptions than plain ECC signatures, and more importantly they're slow to verify…”
But what if, in 2023, those obstacles were no longer an issue?
Looking a bit deeper into the process:
“...a BLS signature one can take a collection of {message, pubkey, signature} tuples and merge their signatures… which is infeasible to unmerge them without knowing the components, and still be able to verify the result. Via a clever trick …the authorship of inputs and outputs can be unlinked…”
See the decade-old discussion on CoinJoin: Bitcoin privacy for the real world (August 22, 2013). There’s also a more recent paper, Fast-HotStuff: A Fast and Robust BFT Protocol for Blockchains (arXiv:2010.11454v10 [cs.DC] 3 Nov 2022), that refers to aggregate signatures (and HotStuff) in its introduction.
Block Producer Response
All BPs seem to approve of HotStuff. There's much excitement over the possibilities. HotStuff introduces new ‘mechanisms’ into the Antelope environment, like a new governance model. Systems of governance (running Antelope) may:
“...ultimately choose whatever …makes the most sense…”
OUTLOOK
The anticipated release date for HotStuff is Q4 (following Leap 5.0). However, the development team is hopeful of an earlier release. Gnome added:
“...it's obviously like a very big uh upgrade to the system… the software…”
Sources & References