Automated Governance: DeFi’s scientific evolution 🧪 | by Tarun Chitra | Gauntlet | Medium

Barbarians in the Boardroom

We are in the middle of a Cambrian explosion in the use of governance tokens to incentivize user participation and ownership of decentralized financial (DeFi) protocols.

Throughout history, financial instruments have evolved to provide asset holders with increasing control. However, the current financial system has a disconnect between governance rights and execution, where individuals’ preferences are represented and intermediated by institutions.

The core of DeFi is a rejection of this paradigm, allowing individuals to exercise governance and risk rights over complex financial assets directly. With this power, token-holders are responsible for monitoring risk, analyzing protocol proposals, and voting for optimal functionality.

Unlike traditional capital markets, DeFi does not require trusted counter-parties with predictable behavior and as a result, the value of a user’s assets depend on all other users’ expected behavior. This makes proper risk assessment in DeFi inherently challenging, requiring a model of user behavior using on-chain, market, and social data.

Gauntlet helps protocols and their stakeholders make the right governance choices by continuously simulating adversarial behavior. Our mission is to enable a safer, more efficient DeFi ecosystem that’s resilient to attacks and rewards honest participants fairly.

Agent-Based Simulation

The core of Gauntlet’s tooling is agent-based simulation, a robust framework used for risk mitigation in systems with complex user incentives, such as algorithmic trading, AI (Go, Starcraft), and self-driving cars.

We’ve pioneered the use of rigorous agent-based simulations to stress test crypto protocols, including Aave, Ampleforth, Celo, Compound, THORChain, and Uniswap.

At a high-level, agent-based simulation:

  1. Codifies the rules that market participants must follow.
  2. Defines the profit functions that different types of users (agents) have.
  3. Simulates combinations of agents interacting with each other, under the assumption that they are profit-maximizing.

When analyzing DeFi protocols, Gauntlet runs millions of simulations that synthesize on-chain, market, and social data into simple, actionable metrics like the probability that a borrower in an on-chain lending protocol defaults. Unlike simple regression techniques and “snapshot” estimates, Gauntlet can estimate the uncertainty in these estimates, e.g., 5% +/- 0.25%, and measure them continuously, giving users confidence in prediction accuracy.

There’s rapidly increasing interconnectedness in DeFi. Gauntlet’s simulations don’t just analyze protocols in isolation; they assess how risk composes when using assets minted from one protocol within another protocol. Estimating composable risk is necessary when facilitating governance, as seen clearly in recent MakerDAO and Yearn Finance changes.

The early days of DeFi have been filled with discretionary “finger in the air” style parameter selection. As DeFi matures, communities are becoming increasingly aware of the need for scientific parameter choices and informed governance decisions. With Compound and MakerDAO growing to nine figures, there’s been significant community demand for scientific governance. Aave’s community has spent resources codifying its risk framework, with a sharp eye on adding insurance and reinsurance pools for their new token and liquidation economics.

While many prominent DeFi protocols have already used Gauntlet’s simulation systems, we are advancing the state-of-the-art in DeFi by going one step further with Automated Governance.

What is Automated Governance?

Imagine a system that could:

  1. Continuously estimate the risks of catastrophic and expected events to a protocol from governance, markets, and protocol-to-protocol interactions on a minute basis.
  2. Determine if there is a need for a governance vote to adjust a risk parameter to a scientifically chosen value, reducing the likelihood of a catastrophic event.
  3. Generate smart contract code to execute the change.
  4. Automatically submit a proposal to the blockchain, subject to a governance vote.

This system would allow holders to confidently vote on the change with a level of precision and confidence not known in the stone age of DeFi, backed by statistical evidence from the system.

What we’re describing is Automated Governance, powered by Gauntlet. This is how it works:


Automated Governance acts as an immune system for a DeFi contract, providing the community with an off-chain risk monitoring system. Investors can rest assured, knowing that their assets are monitored for safety even in the rockiest conditions, much like a hedge fund’s risk management system or application performance monitoring.

By automating data collection, analysis, and proposal generation, investors making voting decisions only need to look at dashboards that are automatically generated and shared on community forums. Gauntlet can adjust and improve models as conditions change, ensuring that predictions and recommendations are as up-to-date and as accurate as possible.

We believe that the best way to empower participants to make informed decisions while preserving investors’ rights is to use best-in-class, fully automated risk measurements. But note that we strongly believe that automated suggestions should only exist for quantitative questions — what collateral factor should we use? — and not qualitative questions (“should we pay Uniswap aggregators?”).

By combining lessons from traditional finance and artificial intelligence with crypto-native infrastructure, we believe that the time is right for Automated Governance.

Right now, we have built out automated infrastructure that can refit models and rerun simulations at high frequency (e.g. every minute). We are currently fine-tuning the outputs of these simulations into actionable predictions, such as our risk score with DeFi Pulse that will be out within the next few weeks. These live, public predictions will help us improve our accuracy and quality assurance around the complex risk measurements needed to evaluate DeFi protocols. We hope to have our first case studies of Automated Governance focused on the Compound protocol live in the coming months. We’ve focused on Compound for our first launch for a number of reasons:

  1. Compound’s governance contract has become a Schelling point for numerous DeFi protocols including Uniswap and YAM
  2. Compound’s community is one of the first in DeFi to pride itself on requiring quantitative and rigorous risk management for parameter change
  3. We’ve spent a lot of resources on optimizing our Compound agent models to be best-in-class to represent the novel risks posed by the unique liquidation conditions in DeFi credit

So what can you do to have Gauntlet’s Automated Governance as a part of your protocol?

  1. Research case studies such as our Compound report or proposal 21 that illustrate which parameters are best suited for Automated Governance
  2. Work closely with us to integrate and audit the economic nuances of your community’s protocol, akin to those in our previous reports
  3. Have the community decide on how they are going to spend their budget and ensure that there is ample funding for Automated Governance, for example:
  • Compound’s community has decided to set aside 20% of its budget for non-liquidity-related spending, such as security audits, insurance, governance incentives, and developer compensation
  • Aave’s Ecosystem Fund represents roughly 50% of the non-liquid portion of the token and will be used for spending on security and insurance
  • Synthetix’s DAOs will provide funding for development, security, and governance improvements

We’re excited for DeFi to slowly take over traditional finance by optimizing capital efficiency and security and hope that we can dramatically improve governance processes across the DeFi landscape with our tooling.