Itai Abraham

The a16z Show | Itai Abraham & Tim Ruffgarten

1. Key Themes

Byzantine Fault Tolerance Is the Hidden Core of Every Major Blockchain

What Satoshi Nakamoto built was not a novel invention from nowhere — it was, knowingly, a solution to a 40-year-old academic problem. Itai Abraham makes this explicit: "In early emails from Satoshi Nakamoto, he kind of realized that. He said the core technical aspect of Bitcoin is solving Byzantine agreements." [00:04:07] Today, this framing has become consensus itself: "I would say de facto all the major chains that we know are running some version of Byzantine fault tolerance." [00:00:22]

Two Parallel Research Threads Have Converged Into One Field

Distributed computing research (dating back to the late 1970s) and blockchain research (post-2009) ran largely independently for years before merging. Itai Abraham identifies the inflection point: "Around 2017 or 16 was when this changed. I think the first protocols were Tendermint, as an example. These were the protocols that kind of were using the classical Byzantine fault-tolerant protocols, but applying them on proof of stake." [00:09:02] Tim Ruffgarten characterizes this as an ongoing convergence: "Those two threads have been coming together over the past maybe five years or so." [00:08:28]

Proof of Stake Unlocked Classical Consensus Techniques That Proof of Work Structurally Cannot Use

This is a subtle but critical technical insight. Proof of work's anonymous, permissionless nature — its Sybil resistance mechanism — is actually incompatible with classical Byzantine agreement protocols. Tim Ruffgarten explains: "If you don't know who people are and you can't do one vote per person because you don't know who the people are, you have to do one vote per some scarce resource... In some sense, that's incompatible with the types of techniques that Drs. Lamport and Liskov will be talking about. Whereas proof of stake Sybil resistance... actually unlocks those techniques." [00:10:03] Itai Abraham adds context to how non-obvious this was: "I remember 2015, people were saying, oh, there's this idea of proof of stake, but how do we do it? It seems impossible." [00:10:51]

Throughput and Latency Are the Central Engineering Battles of Modern Consensus

The gap between the earliest blockchains and what large-scale economic systems require is enormous. Itai Abraham frames the stakes: "If you're thinking about serving billions of people or systems that really manage large economies, then that wouldn't be enough for some types of use cases." [00:00:22] Two families of innovation are driving progress: DAG-based protocols for throughput (seen in protocols like Mysticeti) and dual-mode protocols for latency — "A fast path that has just two message delays... the smallest latency that you could possibly imagine." [00:15:59]

The "Wartime / Peacetime" Design Philosophy Is the Right Mental Model for Consensus Protocols

Rather than designing for worst-case all the time — which kills performance — modern protocols optimize the common (peacetime) case while preserving the ability to switch modes under attack. Itai Abraham articulates this: "You want to have kind of a wartime mode and a peacetime mode. So in peacetime, there's no failures. Everything's good. You want to be super fast and super efficient... And the thing is that you do want to be able to switch to wartime. So if you are under attack, then you do have a way to kind of overcome a massive attempt to corrupt your system." [00:00:22]

Blockchain Has Supercharged Dormant Areas of Theoretical Computer Science

Fields like Byzantine fault tolerance and zero-knowledge proofs (SNARKs), once considered purely academic or impractical, have been jolted to life by the economic value of blockchain applications. Tim Ruffgarten notes: "SNARKs, which for many decades was viewed as a purely theoretical construct... now we're really seeing very concretely efficient SNARKs come into production... consensus protocols have been supercharged as well by having blockchain technology as a sort of extremely high-value application." [00:12:43] He adds context on the pre-blockchain view of Byzantine fault tolerance: "In 2007... there were two big complaints. One is that maybe nobody needs it and the other was that the performance was horrible." [00:00:00]

Theory and Practice Are in Genuine Two-Way Dialogue — Not a One-Way Transfer

The standard assumption is that theory trickles down into practice. In this field, the relationship runs both directions. Itai Abraham: "It's a two-way street, right? So it's not that there's somebody sitting in an ivory tower and writing theorems. It's actually, you see what's happening in practice and you analyze that and then you realize you can... It was very fruitful, I think, for both sides." [00:19:58]

2. Contrarian Perspectives

Byzantine Fault Tolerance Was Considered Practically Useless as Recently as 2007

The mainstream computer science community, as late as 2007, believed Byzantine fault tolerance was both unnecessary and too slow for real systems. This was not fringe opinion — it was the prevailing view at a dedicated academic workshop. Itai Abraham: "The goal of the workshop is to kind of see whether Byzantine fault tolerance is practical or not... there were two big complaints. One is that maybe nobody needs it and the other was that the performance was horrible." [00:00:00] Bitcoin then proved within two years that adversarial fault tolerance was not only needed but could be the foundation of a trillion-dollar system.

Bitcoin Was Recognized by Its Own Creator as a Computer Science Solution, Not Just a Financial One

The popular narrative presents Bitcoin purely as a monetary or libertarian project. In fact, Nakamoto explicitly framed it in academic distributed-systems terms. Itai Abraham: "If you look at early emails from Satoshi Nakamoto, he kind of realized that. He said the core technical aspect of Bitcoin is solving Byzantine agreements." [00:04:07] This reframes Bitcoin's origin story — it was a deliberate engineering solution to a specific, named, decades-old problem in computer science.

Proof of Stake Was Considered Impossible by Many Serious Researchers in 2015

Even as Ethereum publicly committed to transitioning to proof of stake before its 2015 launch, many technically sophisticated people doubted it could be done. Itai Abraham: "I remember 2015, people were saying, oh, there's this idea of proof of stake, but how do we do it? It seems impossible. There were a lot of people that said there's no way to kind of do something similar." [00:10:51] The eventual seven-year gap between Ethereum's launch and its 2022 proof-of-stake transition confirms this was not an easy problem.

"Optimal Fault Tolerance in Partial Synchrony" Has Become Table Stakes, Not a Research Luxury

What was once a rarefied academic specification has become an expected baseline for any serious new blockchain protocol. Tim Ruffgarten: "Even if it's just meant to be a purely practical blockchain protocol, still is an expectation that you would have, for example, optimal fault tolerance in partial synchrony... that has almost become table stakes for new generations of blockchain protocols." [00:17:41] This implies that protocol teams that cannot speak this language or meet this bar are already behind.

3. Companies Identified

VMware

Large enterprise technology company. Mentioned because Itai Abraham was a founding member of VMware's blockchain project, establishing a direct link between enterprise distributed systems research and blockchain protocol design.

"He's a founding member of VMware's blockchain project." [00:02:59] — Tim Ruffgarten

Ethereum

Leading smart contract blockchain. Mentioned for the sophistication of its consensus architecture — a two-layer design combining a longest-chain lower layer with the Casper finality gadget, which directly applies classical Byzantine fault-tolerant techniques. Also noted for the complexity of its proof-of-stake transition (2015 intent, 2022 execution).

"Today's Ethereum has sort of two layers. It has like a longest chain, sort of lower layer, and then a finality gadget based on top — Casper." [00:11:32] — Tim Ruffgarten

Solana (via Alpenglow)

High-performance Layer 1 blockchain. Mentioned specifically for Alpenglow, a proposed new consensus protocol described as a leading in-production implementation of dual-mode (fast path / regular path) Byzantine agreement.

"Alpenglow, which is the proposed new version of Solana's consensus protocol, which I think will be rolled out in 2026. That's really kind of an in-production implementation of this idea." [00:16:27] — Tim Ruffgarten

Cosmos (via Tendermint)

Blockchain ecosystem. Mentioned because Tendermint — the consensus engine underlying Cosmos — was among the first protocols to apply classical Byzantine fault-tolerant techniques in a proof-of-stake setting, representing a pivotal moment in the convergence of academic and blockchain research.

"You mentioned Tendermint, which may be familiar to listeners from the Cosmos ecosystem... these were the protocols that kind of were using the classical Byzantine fault-tolerant protocols, but applying them on proof of stake." [00:11:04] — Tim Ruffgarten

4. People Identified

Satoshi Nakamoto

Pseudonymous creator of Bitcoin. Mentioned for deliberately framing Bitcoin as a solution to Byzantine agreement — a known academic problem — demonstrating a level of technical depth often absent from popular retellings of Bitcoin's origin.

"In early emails from Satoshi Nakamoto, he kind of realized that. He said the core technical aspect of Bitcoin is solving Byzantine agreements." [00:04:07] — Itai Abraham

Leslie Lamport

Pioneer of distributed systems; inventor of Paxos and state machine replication. Highlighted as one of the two foundational figures whose theoretical work underpins all modern blockchain consensus protocols. Will be a direct interview subject in the series.

"The very powerful abstraction — this is an abstraction that we'll hear about from Leslie Lamport — is what's called state machine replication." [00:07:04] — Itai Abraham

Barbara Liskov

Turing Award-winning computer scientist; creator of foundational work in Byzantine fault-tolerant protocols (PBFT). Highlighted alongside Lamport as a direct intellectual ancestor of modern blockchain consensus design. Will be a direct interview subject in the series.

"Together, we unpack the work of two pioneers, Barbara Liskov and Leslie Lamport, whom you'll hear directly from in the episodes ahead." [00:02:59] — Tim Ruffgarten

Dahlia Malkhi

Prominent distributed systems researcher. Mentioned as co-author with Itai Abraham of a survey paper that was among the first to formally articulate the connection between Bitcoin and classical Byzantine agreement — a document that appears to have been genuinely influential in reshaping how researchers understood the field.

"I think I first became aware of this interesting interplay... it was a survey you wrote, I think, with Dahlia Malkhi, that sort of first pointed that out. So that was maybe 2017 or something like that. That made a big impression on me at the time." [00:11:32] — Tim Ruffgarten

Research partner at a16z Crypto; leading academic researcher in Byzantine agreement and consensus protocols; founder of Decentralized Thoughts blog; founding member of VMware's blockchain project. One of the key figures bridging academic consensus theory and practical blockchain protocol design.

"Itai is one of the world's leading researchers in Byzantine agreement and consensus protocols. He's a founding member of VMware's blockchain project. He's also the founder of Decentralized Thoughts, one of the field's most respected and long-running technical blogs." [00:02:59] — Tim Ruffgarten

Tim Ruffgarten

Head of research at a16z Crypto; professor of computer science at Columbia University. Co-author (with Itai Abraham) of the influential 2017 survey connecting Bitcoin to classical Byzantine agreement. One of the architects of the intellectual bridge between academia and the blockchain industry.

"I'm Tim Ruffgarten, head of research at A16Z Crypto and professor of computer science at Columbia University." [00:01:42] — Tim Ruffgarten

5. Operating Insights

Design Systems With Explicit Peacetime and Wartime Modes

For anyone building distributed systems, consensus infrastructure, or even organizational decision-making processes: optimize hard for the common case, but preserve the ability to switch to a more robust (slower, fault-tolerant) mode under adversarial conditions. The blockchain community has empirically validated that the "adversarial case" is genuinely rare, making this design philosophy not a compromise but a strict improvement. Itai Abraham: "You want to have kind of a wartime mode and a peacetime mode. So in peacetime, there's no failures. Everything's good. You want to be super fast and super efficient... That's going to be 99% of the time or even more... And the thing is that you do want to be able to switch to wartime." [00:00:22]

Theoretical Rigor Is Now a Competitive Moat in Protocol Design

Teams building blockchain protocols that cannot articulate formal guarantees — optimal fault tolerance, partial synchrony — are now behind the baseline expectation. Investing in researchers who can speak both languages (theory and engineering) is not an academic luxury but a product requirement. Tim Ruffgarten: "Even if it's just meant to be a purely practical blockchain protocol, still is an expectation that you would have, for example, optimal fault tolerance in partial synchrony... that has almost become table stakes for new generations of blockchain protocols." [00:17:41]

Watch the Two-Message-Delay Latency Threshold as a Benchmark for Protocol Quality

A concrete technical bar has emerged: the theoretical minimum latency for a Byzantine agreement fast path is two message delays (client → replicas → commit). Any protocol claiming state-of-the-art latency should be evaluated against this bound. Itai Abraham: "These new protocols... have two different modes. A regular mode that maybe has three message delays and a fast path that has just two message delays. And this is the optimal thing that you could expect... really, you get the smallest latency that you could possibly imagine." [00:15:59]

6. Overlooked Insights

DAG-Based Consensus Protocols Are a Distinct, Underappreciated Architectural Shift — Not Merely an Optimization

In a single sentence, Itai Abraham names DAG-based protocols as a separate family of innovation from latency-focused dual-mode protocols, citing Sui and Mysticeti as examples. This is easy to pass over, but it is architecturally significant: DAG (Directed Acyclic Graph) based consensus separates data dissemination from ordering, enabling parallelism that linear-chain architectures cannot match. This represents a fundamental rethinking of how throughput scales, not just a parameter tweak. The mention is brief and uncommented upon, but it flags a distinct design space that is already in production.

"One type of innovation is these DAG-based protocols. These are protocols that have two different layers. And they really push the throughput of these systems quite a bit. We've seen this, for example, in Sui and protocols like Mysticeti." [00:15:32] — Itai Abraham

Alpenglow's 2026 Rollout Is a Named, Dated Production Milestone Worth Tracking

Tim Ruffgarten states a specific, named, dated prediction with no elaboration: Alpenglow, Solana's new consensus protocol incorporating dual-mode Byzantine agreement, is slated to roll out in 2026. This is the most concrete forward-looking claim in the episode and is dropped without fanfare. If it ships on schedule, it will be the highest-scale real-world validation of academic fast-path consensus techniques to date — a meaningful signal for both protocol researchers and anyone assessing Solana's technical trajectory.

"Alpenglow, which is the proposed new version of Solana's consensus protocol, which I think will be rolled out in 2026. That's really kind of an in-production implementation of this idea." [00:16:27] — Tim Ruffgarten