DL Research Content

A Conversation with Sam Snyder, Head of Blockchain Research of Improbable

A Conversation with Sam Snyder, Head of Blockchain Research of Improbable
Illustration: Andrés Tapia; Source: Improbable

Sam Snyder is a distributed systems engineer with seven years of experience in high-performance computing and distributed systems across the gaming and blockchain industries. He is particularly focused on software efficiency and building systems that optimise computational performance. As Head of Blockchain Research at Improbable, Sam oversees the technical development of Somnia, a blockchain network designed to support interconnected virtual environments. Somnia aims to streamline user and developer experiences in the metaverse, offering tools for content creation, collaboration, and decentralised governance.

What specific optimisations have you implemented in Somnia, and how has your background in high-performance computing informed them?

Performance hinges on resource management and minimising system variance. My focus has been on aligning worst-case scenarios with average cases, ensuring consistent performance regardless of input. This principle, central to all high-performance systems I’ve worked on, underpins Somnia’s architecture.

To achieve this, we prioritised accelerating single-core speed over parallel execution (optimistic concurrency). Parallelism often introduces performance variability due to inter-thread communication and dependency on transaction relations, which undermines efficiency. Our single-threaded approach, commonly used in low-latency systems, avoids these pitfalls.

A cornerstone of this approach is IceDB, our custom-built database designed for deterministic performance. By allowing gas charges to reflect the exact time required for database reads, IceDB significantly reduces performance variance, enabling steady and predictable system behaviour.

What are the key innovations behind the Somnia blockchain, and how do they address web3 challenges?

Somnia addresses four critical bottlenecks essential for optimal performance:

1. Execution speed: Enhancing the speed of program operations.

2. Database speed: Improving state read/write performance.

Join the community to get our latest stories and updates

3. Bandwidth: Expanding the capacity to share transaction data.

4. Signature speed: Increasing the rate of signature verifications.

Each bottleneck required a unique solution, as resolving only some would limit the overall system’s effectiveness. By tackling all four, Somnia achieves a balance that delivers tangible performance benefits.

What are the goals of the DevNet phase, and how will it prepare for Somnia’s full rollout?

The DevNet phase aims to demonstrate the potential of high-performance blockchains by letting developers experiment with a live Somnia instance. Removing traditional blockchain performance constraints opens opportunities for a new class of applications.

This phase also enables rigorous testing of our technology in real-world conditions. By “dogfooding” the system—using it internally—we can identify and resolve issues before advancing to the TestNet phase, scheduled for early next year, where the chain will undergo extensive audits and stress testing.