Equilibrium Infra Bulletin #12: Benefits of Light Clients, Filecoin's Long-Term Vision and Coprocessors 101
Equilibrium Labs builds the state-of-the-art of decentralized infrastructure. We are a global team of ~30 people who tackle challenges around security, privacy, and scaling.
This newsletter allows us to share more about what we read, what excites us, and what we think is relevant to the space. In addition, you will get a glimpse into the organization and our culture. You can also find us on Telegram.
We recently launched a referral campaign and successful referrals will be rewarded! The three-tiered referral works the following way:
Refer 1 person - Get an invite to a private Telegram chat.
Refer 10 people - Get a shoutout in the next newsletter.
Refer 25 people - Get an Equilibrium beanie to keep you warm throughout the winter.
Use the box above to get a unique referral link that you can share with friends (either directly or on social networks like X/Twitter), and earn rewards as they sign up. We’ll reveal more rewards for successful referrals in the near future, so stay tuned… 👀
Research, Articles and Industry News:
📚 Light Nodes and Their Benefits - Recommended by Joakim:
Blockchain nodes come in three main categories - validator, full, and light nodes:
Validator nodes are responsible for creating & proposing blocks when it’s their turn and voting on blocks proposed by other validators.
Full nodes act as “guardians” and verify that validators follow the rules by downloading all the data. Full nodes have similar hardware requirements as being a validator but don’t require putting capital at stake (in PoS networks). In other words, all validators are also full nodes, but not all full nodes need to be validators.
Light nodes (a.k.a light clients) use cryptographic techniques to lower the cost of verifying blockchain information and hence have lower hardware requirements. Today, they can be used to verify that a transaction was included in a block by downloading the block header (a small unique cryptographic fingerprint). However, they need to trust that the majority of block producers are honest since they don’t download all the data.
The long-term vision for light clients is to offer full node security while keeping hardware requirements low (for example enabling users to run light nodes on their phones in the background). While we still lack formal definitions around the different types of light nodes, Dino proposes the following categorization:
Consensus verifiers - Check that a transaction is included in the canonical chain by verifying Merkle proofs. Still trusting the validators, however, which could include invalid transactions or state transitions.
State verifiers - Enable checking that state transitions are valid without processing transactions by using either validity proofs (ZKPs) or fraud proofs. Validity proofs are more appealing, as they can be used to directly verify the validity of a block, without requiring an honest full node to tell them fraud occurred (as is the case with fraud proofs). Even with validity proofs, however, validators can withhold publishing data, which makes it difficult to fetch information about the new blockchain state from full nodes.
Data availability (DA) verifiers - Using data availability sampling to check that data (inputs to the computation) is stored correctly and available to download if needed. Rather than downloading all the data which is costly, each light node only samples a random piece of the data. The confidence grows with the number of light nodes participating, i.e. more nodes sampling = more security.
Full verifiers - the combination of the three above. Enable almost the same level of trustless properties as full nodes, without having to download blocks.
Key Takeaway: Full nodes are important for ensuring that validators behave correctly, but running a full node requires high hardware requirements and is hence restrictive to most users. Light nodes aim to reduce this friction while keeping trust assumptions to a minimum, but no solution has yet managed to let us both have our cake and eat it.
🎧 The Technical Ambitions of Filecoin - Recommended by Hannes:
Filecoin is a decentralized cloud storage marketplace, protocol, and incentive layer. It leverages IPFS (a P2P network for sharing and distributing data) and the goal of Filecoin is to allow anyone to store, retrieve, and compute over data in a distributed way. In this recent fireside chat, Jonathan Victor (Ecosystem Lead at Protocol Labs) gives an introduction to the network, followed by an update on utilization and adoption.
Filecoin often gets bucketed within decentralized infrastructure networks - either DePIN (physical network) or DeREN (resource networks). We don’t think this classification is fair given that it puts Filecoin in the same category as the Hivemappers and Heliums of the world. Instead, we believe it’s more useful to think about what Filecoin is trying to achieve in the long term, namely decentralized cloud storage and a platform for general-purpose compute. This makes Ethereum and other permissionless blockchains better comparables.
The recent launch of the Filecoin Virtual Machine (FVM) brings programmability to Filecoin - both between services built on top of Filecoin and the ability to add instructions to data. We covered the larger topic of “Compute over Data” in Infra Bulletin #9. After having spent the last years building out the storage network (supply-side) and working on onboarding customers to store data on Filecoin (demand-side), the focus is now on the last step of the “master plan” - namely enabling computation, retrieval, and other related services on that data directly on the Filecoin network.
Key Takeaway: Filecoin is building out the infrastructure for a decentralized cloud storage network. We think the current categorization of Filecoin under DePIN or DeREN is misaligned with the long-term trajectory of the protocol, particularly following the recent addition of programmability on the network through the FVM. Instead, it’s more useful to compare Filecoin to other permissionless blockchains. That said, while the vision of Filecoin is appealing, a lot of work is still required to make that vision into reality.
📚 A Brief Intro to Coprocessors - Recommended by Joakim:
Most blockchains have limited capacity for on-chain computation, which has led to protocols moving certain components off-chain (such as order books or risk-management systems). The simplest reason for this is the difficulty of maintaining these performance-centric systems on-chain efficiently and in a scalable manner, but it also .
Coprocessors in blockchains combine customized and performant computational environments tailored to a specific task (similar to traditional computer architecture) with additional verifiability that the off-chain computation is done correctly. This removes the need to trust that the off-chain system works as it says it does, while still getting the benefits of cost-savings and scale/performance.
ZK-coprocessors allow smart contracts to offload heavy computation off-chain while providing a zero-knowledge proof that can be verified to ensure that the code was executed correctly. Companies in the space include Axiom and RiscZero:
Axiom enables Ethereum smart contracts to access more historical on-chain data and perform complex computations on it, without requiring all nodes to store the full archive (this maintains decentralization). The longer-term vision includes more advanced ZK computations beyond EVM capabilities.
Risc Zero’s general-purpose zkVM allows proving arbitrary programs written in languages such as Rust, C/C++, and Go, without needing to design custom circuits. Bonsai is the off-chain proving service that leverages ZKPs to provide a bridge between on-chain smart contracts and off-chain computations in zkVM.
Alternative solutions to achieve the same end goal (i.e. verifiable off-chain computation) include Multi-party computation (MPC) and Secure Enclaves/TEE. MPC allows for joint computing on sensitive data (no entity sees the entire picture), while TEE is a hardware-based solution. Both come with tradeoffs.
Key Takeaways: Most blockchains today have limited performance, which has led to dApps performing more intense computation off-chain to improve performance and lower cost. However, this requires trusting the off-chain system to work as promised. Coprocessors aim to change this with verifiable off-chain computation. ZK-coprocessors leveraging ZKPs have stronger guarantees, but alternative solutions such as MPC and TEE might be more efficient for now. Ultimately, the choice of coprocessor depends on the specific use case.
Personal Recommendations From Our Team:
📚Reading: There Is No Antimemetics Division: An antimeme is an idea with self-censoring properties; an idea which, by its intrinsic nature, discourages or prevents people from spreading it. qntm (real name Sam Hughes) brings us a connected series of short stories set in the SCP Foundation.
🎧Listening: Paramore - Big Man, Little Dignity (Re: DOMi & JD BECK): This song is part of the larger cross-over album “Re: This is Why” by Paramore, which the band itself describes as “almost a remix album”. It features artists that Paramore feels have influenced them as a band and/or who have cited Paramore as an influence.
💡Other: 'I've Got Nothing to Hide' and Other Misunderstandings of Privacy: Professor Daniel Solove examines the nothing to hide argument in this short essay. The argument ultimately boils down to questions about what privacy is, how we can assess its value, and how to weigh privacy against countervailing values (such as national security). An interesting read all around!