Ouroboros Leios (Sneak Peak)

Opens May 1st 2024 to preview a major research paper on the Ouroboros Leios agenda, also known as input endorsers and parallel chains, aimed at solving the blockchain trilemma for good.

The paper is submitted and being cleaned for publication, with Aggelos doing a long-form scalability video series later in the month, so this is only a whiteboard sneak peek.

A normal blockchain's rate of change is a block committed by a leader, selected either Nakamoto-style from proof of work or BFT-style.

Calls the general family proof of X consensus, where X can be work or stake, but notes these run on a Delta-bound security model that caps delivery time and leaves most network capacity unused at low throughput.

Leios optimizes the protocol, work or stake, to use as much network resource as possible, expressed as one minus Delta times the honest-actor bound, ideally approaching full utilization.

The paper solves two problems, the first being protocol burst, where a leader withholds legitimate history then broadcasts it all at once to flood the network and starve everyone else.

The second problem is message equivocation, where a leader submits multiple indistinguishable but conflicting valid messages to spam and consume all network resources.

Leios uses five complementary mechanisms to kill both attacks, letting the network use nearly all its resources toward an alpha of zero, uniquely leveraging extended UTXO and Mithril.

Introduces proofs of equivocation, proofs of data availability, asynchronous block delivery, and a cloud of input blocks stored by reference with associated endorser blocks.

Contrasts optimized BFT protocols like Red Belly, fast at hundreds of thousands of TPS but collapsing under Byzantine behavior, since burst and equivocation dramatically slow performance.

Calls Leios a meta protocol, not a replacement but a reconfiguration of Ouroboros, allowing parallel block processing that still looks like a single sharded environment running at optimal speed.

Promises a dramatic increase in TPS and transactions per throughput plus a larger computation-unit budget, which is very hard in an accounts model but easy thanks to extended UTXO and Mithril.

An active team is prototyping Ouroboros Praos over the summer, and they plan to scale it up and start prototyping Leios and the Peras fast-finality protocol, both meta protocols closely related to Praos.

Wants both protocols into an RFP state with vendors picked through Intersect on fixed-cost fixed-time contracts, expecting huge scalability demand from crypto adoption in 2025.

Credits the elegant result to years of in-house work built on industry papers from Pramod Viswanath and David Tse, calling it the closest the industry has come to solving the trilemma while keeping 50 percent Byzantine resistance.

Opening the ledger for Leios adds primitives and hooks for recursion, Mithril certificates, Plutus enhancements, core bridges, and Hydra, so everything works better together.

Tweag continues work on Genesis alongside the Chang bookend hard fork through the summer, with Peras and Leios next in line, so Cardano runs without shutting down every 15 minutes.

Making the Roger Ver video the day before while taking notes on the paper, and recaps community events, a Catalyst gathering of 100-plus in Barcelona and a hackathon in Toulouse.

Teases Hydra game demos at Rare Evo, notes Plutus V3 unlocks the ZK world, and says growing Mithril adoption pulls it into Cardano's core for the required data-availability proofs.

Urges builders to join Intersect so that while the ledger is opened for surgery for a meta protocol like Leios, needed changes to interfaces, block headers, and data structures get discussed and turned into proposals.

Intersect as a development organization like the Linux Foundation, not a bureaucracy, holding the repos and product backlog for the Haskell node that 15-plus companies build.

Calls this the end of an era, the completion of the original Ouroboros agenda: Kronos for a decentralized clock, Peras for fast finality, Leios for max throughput, Genesis for trustless bootstrapping, and Praos for adaptive security.

Reflects on eight years since the community started in 2016, thousands of citations and millions of lines of code, calling it bittersweet as work turns to the next-generation Ouroboros Omega.

Recalls the scientists first hammering Leios out at a vineyard for a week after the Crypto conference in Santa Barbara, pulling together parallel chains and other pieces.

Floats a post-quantum VRF for the freshness proofs that would also enable post-quantum Mithril certificates as a quantum-immune internal checkpointing system, praising Algorand's lead and easy collaboration.

The extended UTXO model keeps paying dividends, from isomorphic state channels and easier rollups to easier scalability, and that diffusion pipelining from 2022 needs an overhaul for Leios.

The multi-client environment means porting the hard-fork combinator so upgrades trigger together on the Haskell, Pragma, and Harmonic Lab nodes, raising the question of what makes a certified client.

168 scientists have published over 210 papers with Input Output, this being roughly the 210th, and that this citation-rich collaborative model keeps the technology best in class; partner chains add a parallel stack with access to innovations like JAM.

Closes reflecting on over 10 years since this was the original Ethereum roadmap he had as CEO, noting Ethereum later copied liquid non-custodial staking and UTXO without credit, but you build because the world needs it.