Dominic Williams has stirred the blockchain conversation pot with a detailed critique of “instant finality,” a buzzword often championed by blockchain projects to assert speed and efficiency. His Twitter thread cuts through the claims with a mix of technical dissection and dry humour, challenging the validity of these promises in light of the basic laws of distributed computing and the internet’s physical limitations.
The central premise of “instant finality” suggests that a transaction becomes irrevocable as soon as it’s included in a block. Williams dismantles this notion by addressing the fundamental requirement for consensus, the process by which multiple parties—nodes in a blockchain—agree on the validity of a transaction or block. Reaching consensus, he notes, isn’t magic but a result of communication between nodes. Messages need to traverse networks, and that takes time.
At the heart of his argument lies the unavoidable constraint of physics. Data transmission relies on fibre-optic cables, where light travels at two-thirds of its speed in a vacuum—about 200,000 kilometres per second. This might sound lightning-fast, but real-world internet connections introduce delays. Williams uses the example of a hypothetical blockchain node in London communicating with one in Tokyo, a journey that would take about 100 milliseconds for a round trip in an ideal world. The reality, however, involves network routing, where multiple links and devices increase latency. He estimates real-world communication between well-connected nodes might range from 250 to 400 milliseconds, illustrating that even under optimal conditions, the idea of instantaneous agreement is far-fetched.
Williams dives deeper into how block finalisation actually works. For a block to be deemed final, the block-producing node must first distribute the block to other nodes. These nodes then vote on whether the block should be accepted, and the block is only finalised when a supermajority—typically two-thirds plus one—agrees. The process involves not just transmitting the block, which can take nearly a second due to its size, but also collecting votes, which, though faster, still require time. By his calculation, the Internet Computer, a blockchain he’s closely associated with, achieves finalisation in about one second. This isn’t instant, but it’s remarkably fast within the constraints of current technology.
To underscore the limits of “instant finality,” Williams examines ways blockchains might attempt to reduce latency. One approach is geographically concentrating nodes, which might speed up communication but compromises decentralisation—a key pillar of blockchain technology. Another common tactic is relying on cloud services like Amazon Web Services (AWS) to host nodes, a practice that raises significant centralisation risks. Williams points out the irony in this approach: while projects tout their speed, they simultaneously expose themselves to potential interference or shutdowns by the very entities they rely on for infrastructure.
Williams also anticipates counterarguments. Some might claim that their protocols use methods like random beacons to designate leaders, allowing blocks to be deemed final as soon as a leader produces them. Others might argue that advanced cryptographic proofs, such as zero-knowledge proofs, could validate blocks instantaneously. However, as Williams points out, these claims ignore the essential need for redundancy and failover mechanisms in distributed systems. If a designated leader fails to produce a block, other nodes must step in to maintain the system’s integrity, which inevitably takes time. Without the supermajority of approvals mandated by mathematical principles in distributed computing, finality cannot be assured.
His critique isn’t just theoretical. He highlights additional delays that arise in real-world applications. Beyond network latency, the time taken to process a transaction, execute smart contract functions, and retrieve results can significantly impact the overall speed. Williams argues that “end-to-end latency”—the total time from submitting a transaction to finalisation—is the metric that truly matters. By focusing on this, he suggests the Internet Computer offers a more honest and practical approach, avoiding the pitfalls of overpromising and underdelivering.
The thread is punctuated with Williams’ characteristic wit, dubbing “instant finality” as a product of “Snakeoil Central.” This playful branding reflects his broader scepticism about overhyped claims in the blockchain space, urging users and developers to question the practicality and transparency of such promises. His call to action is clear: scrutinise technical assertions, demand evidence, and avoid getting swept up in flashy marketing.
Williams’ debunking serves as a timely reminder of the trade-offs inherent in blockchain technology. While innovation continues to push the boundaries of what’s possible, some claims, like “instant finality,” demand a reality check. The quest for speed and efficiency must balance the principles of decentralisation and security. Without these, the very essence of blockchain risks being compromised.
This isn’t the first time bold claims in the tech world have faced scrutiny, and it won’t be the last. What sets Williams apart is his ability to blend technical depth with a sense of humour that resonates beyond the usual audience of distributed computing enthusiasts. His thread not only educates but also entertains, making the complexities of blockchain accessible to a wider audience.