Original Author: Momir, IOSG Ventures
Smart contracts have limitations because they lack the ability to interact with the environment, which limits the development potential of decentralized applications (dApps). To achieve more complex functionality, DeFi protocols have two options: they can adopt flexible designs where users can personalize various scenarios, or they can introduce external dependencies - relying on off-chain infrastructures such as oracles, keepers, or off-chain computations - to maintain a simple user experience.
In a recent thought-provoking article titled "Why DeFi is Broken and How to Fix It - Part 1: No Oracle Protocols", Dan Elitzer advocates for the use of zero external dependency DeFi primitives to minimize attack vectors. The idea is to eliminate the need for trust in third-party institutions. However, a zero dependency DeFi ecosystem will inevitably require higher specialization. Most users lack the time, expertise, or resources to become liquidity providers on Uniswap v3 or evaluate collateral quality in protocols without external dependencies. They have to rely on trusted intermediaries to participate.
Therefore, the pursuit of zero dependencies may bring us back to square one, or worse, force non-professional users to trust complex entities or deposit funds into transitional smart contracts, adding security risks. Instead of striving for complete elimination of external dependencies, it is better to consider practical approaches such as stricter scrutiny of external dependencies and limiting potential black swan scenarios. We must recognize that some degree of dependency is inevitable and crucial for the industry's development.
Among well-known DeFi projects, earlier versions of Uniswap came closest to achieving zero dependencies. However, the recent introduction of Uniswap v4 indicates a shift in trend, pushing this field forward through a highly modular approach ("Hooks").
Data Primitives
Discussions about external dependencies mainly revolve around the ability of smart contracts to interact with external data. Nowadays, data interaction typically relies on oracles to access off-chain information, although the scope is limited (mainly including the prices of major cryptocurrencies).
As more and more activities migrate to the blockchain, a large amount of valuable on-chain data can be used to enhance mechanism design in an algorithmic and transparent manner. However, despite the transparency of on-chain data, integrating it with smart contracts is not an easy task. Reading, processing, and delivering meaningful data requires the establishment of a complex and trusted infrastructure. Therefore, developers often rely on existing tools to meet their data needs. However, most existing data solutions are rooted in Web 2.0 frameworks, and even more Web 3.0 native protocols cannot guarantee the accuracy of the data they provide.
Discussion on Sushiswap's inaccurate data transmission regarding Polygon Sushi-Matic subgraph
Considering that smart contracts can even manage billions of dollars in deposits, it is neither acceptable nor practical for them to directly connect to a trusted API source because such a dependency would undermine the decentralized nature of the blockchain ecosystem.
Building Anti-Tampering Data Solutions
Our investment philosophy revolves around a fundamental belief that anti-tampering data will become the cornerstone of the next generation of DeFi protocols. However, achieving data anti-tampering is not a simple task as it requires complex infrastructure and extensive optimization to make it economically viable.
In this context, Space and Time has emerged as a pioneer in building anti-tampering data infrastructure. A key component is its SQL proof, which is an improvement over SNARK proofs specifically designed for querying data from relational databases. It provides guarantees to ensure that queries and their underlying data are not tampered with. Moreover, it provides data integrity guarantees when retrieving data through RPC calls from archival nodes.
Other notable projects in the realm of trustless data primitives include but are not limited to Nil Foundation, Axiom, Brevis, Herodotus, etc.
Anti-tampering data opens up new horizons for DeFi protocols, enabling them to break the boundaries of functionality and drive further growth and innovation in the industry.
Below, we will discuss data-driven protocol design optimizations in the following scenarios:
1. Personalized user experience
2. Self-parametrizing protocols
3. Protocol economics
4. Qualified access
1. Personalized User Experience
In the field of technical business, providing customized services to users is commonplace. However, smart contracts (essentially code strings representing certain business logics) often standardize user experiences, which usually results in poor user experiences. For example, in some lending platforms, User A is a beginner, User B is a long-term agreement user, and User C is a seasoned trader. This lack of differentiation fails to explain user behaviors and misses opportunities to enhance user stickiness, incentivize positive behaviors, and optimize capital utilization.
Protocols have vested interests in identifying user behaviors and making corresponding adjustments. For instance, by leveraging credit ratings, offering cheaper credit or lower loan-to-value ratios to well-performing customers. A project like this would naturally attract users away from platforms with standardized terms. Additionally, this approach provides implicit incentives to users, encouraging them to exhibit good behaviors in order to gain more favorable conditions.
Drawing from the fintech domain, companies like SoFi have gained market share by refusing to standardize, and DeFi dApps can learn from this. For example, SoFi identified the inefficiency in the student loan market, where Stanford graduates were charged the same interest rates as other borrowers despite their higher chances of securing high-paying jobs after graduation. By adjusting rates to better reflect users' risk profile, SoFi achieved significant success.
Similarly, in the DeFi field, we envision an opportunity for an innovative protocol that incorporates user risk into interest rates and collateral factors. However, caution must be exercised not to heavily rely on insufficient collateral-backed loans based solely on existing historical data, as historical data becomes irrelevant when game theory changes.
It is worth mentioning that projects like Spectral and Cred Protocol are attempting to build credit scoring models from on-chain data. However, these projects operate on centralized databases, and thus, major DeFi protocols are unlikely to connect to their APIs as long as the data and model they rely on are sourced from centralized data and easily tampered with. On the other hand, if these projects adopt tamper-proof solutions, they have the potential to become ubiquitous DeFi credit oracles, driving a range of innovative applications.
2. Self-Parametrized Protocols (Minimized Governance Intervention)
Many DeFi protocols still rely on manual governance processes, often guided by off-chain consultancy firms, to adjust their parameters. For example, AAVE pays a hefty sum to external consultancy firms to monitor and guide the risk parameters of the protocol.
However, this approach presents several issues:
1. Lack of real-time support: The system lacks the ability to respond to constantly changing market conditions or newly emerging risks.
2. Manual systems: Dependence on manual intervention introduces issues of delay and potential inefficiency when adjusting protocol parameters.
3. Trust in off-chain entities: Relying on external consultancy firms raises concerns about transparency and the methods used when making recommendations.
This static approach was exposed in an attack on AAVE, resulting in the occurrence of bad debt that could have been avoided with appropriate lending parameters that better reflect the liquidity borrowed. Furthermore, the risk associated with using circulating tokens as collateral in lending protocols has yet to be fully addressed.
To address these limitations, projects should transition towards real-time, automated, transparent, and trustless designs. For example, lending protocols can utilize infrastructures similar to "Space and Time" to monitor data in real-time. This would enable them to dynamically adjust collateral, lending parameters, and other crucial parameters.
Similarly, exchanges can introduce dynamic fee structures based on volatility or impermanent loss. Many liquidity pools on top of Uniswap v3 struggle to achieve sustainable operations mainly due to the inability to charge LPs dynamically. With Uniswap v4's Hooks or Valantis modules, dynamic fees become possible.
Furthermore, aggregators can adapt to the constantly changing risks and rewards of underlying protocols without interference from manual or fixed fees. The collaboration between Spool and Solity is a step towards this direction, with Solity utilizing big data methods to analyze the risk and return of pools.
3. Protocol Economy
Data-driven approaches have the potential to enhance the protocol economy and token economic models in DeFi, where projects can share incentives with users who meet certain conditions.
For example, a DEX aggregator that seeks user stickiness and loyalty can distribute slippage profits to users who meet certain conditions, such as executing a specified number of trades and reaching a minimum trading volume.
Such incentives greatly motivate early users, build loyalty within the user community, and directly provide incentives to existing users, promoting the usage of the protocol within their own group.
4. Qualified Access
Although blockchain has a permissionless nature, it also allows for selective freedom. In multiple cases, permissioned access at the application layer can ensure that protocols are not used for malicious purposes or effectively interact with the target user group.
For example, privacy protocols like Tornado Cash are currently under scrutiny by regulatory bodies because they could potentially be used for money laundering or other illegal activities. To prevent money laundering, protocol developers can take measures to prevent bad actors from interacting with their platform.
Additionally, for market makers, understanding counterparties is invaluable, but decentralized exchanges (DEXs) typically do not have access to such information. Assuming it is possible to utilize data to construct proof of personhood, DEXs can only allow non-bot addresses to interact, thus addressing such issues.
Requirements for Verifiable Computation
By integrating with trusted data primitives, the above discussed content can be fully realized. However, additional resources will be needed to perform statistical calculations or machine learning. For example, a credit scoring project can utilize tamper-proof data but still requires machine learning algorithms to generate credit scores.
Alternatively, in the context of Risk Oracle, obtaining data such as circulating supply, quantity, transaction count, holder count, and time since TGE for specific tokens is crucial for determining appropriate collateral and lending factors. However, machine learning techniques need to perform precise calculations based on this data.
source: https://chainml.substack.com/p/web3-needs-ai-to-realize-its-potential
Other areas in DeFi that require more complex computations include but are not limited to:
Yield Aggregator: estimate the yield and risk of underlying protocols and find the optimal allocation.
Portfolio Optimization: calculate the allocation of the target portfolio based on predetermined criteria, adjust directional exposures based on technical indicators, etc.
Decentralized Derivatives Exchange: systematic risk management, funding cost adjustments, derivatives pricing, etc.
Advanced Trading Execution Algorithms
Liquidity Pool Market Making Logic
Clearing House
Projects like ChainML meet this need by providing a verifiable off-chain computing layer supported by dedicated consensus mechanisms. Other projects that build distributed machine learning computing layers include but are not limited to GenSyn, Together.xyz, Akash, etc.
Similarly, ZKML presents an interesting opportunity where ZK proofs can compress computations into concise proofs that can be verified on the chain or demonstrate the use of specific models without revealing their attributes. Examples of ZK projects include Modulus Labs, Giza, etc.
However, implementing machine learning in ZK is currently very expensive, which increases the challenges of practical implementation. While hardware acceleration and circuit optimization may improve performance in the future, the computational demands of artificial intelligence are expected to grow at an even faster pace, making ZKML limited to niche computing approaches and unable to accommodate state-of-the-art AI models. Therefore, methods like ChainML's consensus-based pessimistic approach or fraud-proof-based optimistic approach may offer the best opportunity to integrate the latest AI algorithms into Web 3.0.
Summary
The convergence of tamper-proof data, advanced computing power, and data-driven decision-making has the potential to unlock new innovations, improve efficiency, and enhance user satisfaction in the DeFi ecosystem. While this article focuses on optimizations that can be done on-chain data primitives, we also see opportunities in integrating various off-chain data through zk proofs. We believe that data will enhance interoperability between on-chain and off-chain, facilitating the integration of decentralized finance with traditional financial systems.
As the industry continues to evolve, protocols must embrace emerging technologies, collaborate with leading projects, and prioritize transparency and trustlessness. This can not only establish a strong and sustainable future for DeFi but also realize the vision of DeFi to have a profound impact on the global financial landscape.
Disclaimer: Space and Time, ChainML, Nil Foundation, and Solity are portfolios of IOSG.
References:
Crypto x AI: https://messari.io/report/growing-synergies-in-ai-and-crypto?referrer=all-research
ecosystem: https://twitter.com/Louissongyz/status/165808735477367193
IOS
Android