How Does Compound V2 Work?

Single shared lending pool where all supplied assets serve as collateral for all borrowing — cross-asset risk contagion by design

V2's shared pool model means a bad debt event in any single asset can cascade to affect all lenders. This is the architecture that V3 specifically moved away from with isolated Comet markets.

Dynamic interest rates based on utilization ratio with a sharp kink above the optimal utilization threshold

Standard kinked utilization curve. At extreme utilization, rates spike but may not be sufficient to prevent liquidity crises where lenders cannot withdraw.

When health factor drops below 1.0, up to 50% of a borrower's position can be instantly liquidated by any liquidator at a fixed discount

V2's instant liquidation model creates cascade risk during market crashes as liquidated collateral floods the market. V3 improved this with gradual absorption.

Suppliers receive cTokens (e.g., cDAI, cUSDC) representing their deposit plus accrued interest, with an increasing exchange rate

cToken model was innovative at launch but is now standard. Exchange rate manipulation in empty markets has been a recurring vulnerability in V2 forks.

COMP token governance with Governor Bravo, timelock, and on-chain binding proposals

Battle-tested governance but proven vulnerable to whale coordination. The $24M Golden Boys treasury extraction demonstrated structural weakness of token-weighted voting with low participation.

Chainlink price feeds for all collateral and borrowable assets determining liquidation thresholds

Standard Chainlink dependency. Oracle staleness or manipulation could delay or prevent necessary liquidations, especially for long-tail assets.

Protocol reserves accumulate from interest rate spread between supply and borrow rates, governed by COMP holders

Treasury has been a governance attack target. DAO treasury management remains contentious with ongoing proposals for fund allocation.

In a market crash, the shared pool model means cascading liquidations across multiple asset types simultaneously. As liquidated collateral floods DEXs, slippage increases, reducing liquidation efficiency and potentially creating bad debt that is socialized across all lenders in the pool.

Low voter participation (often <10% of circulating supply) combined with concentrated COMP holdings enables governance attacks. The 2024 Golden Boys extraction of $24M demonstrated this is not theoretical — the structural vulnerability persists for future whale-coordinated proposals.

The empty pool attack — exploiting cToken exchange rate rounding in newly initialized markets — originated in V2's architecture and has caused over $10M in losses across V2 forks (Hundred Finance $7M, Onyx Protocol $2.1M). While V2 mainnet markets are established, governance proposals for new markets could reintroduce this vector.

In the shared pool model, oracle delays for any single asset can cause systemic mispricing of liquidation thresholds. During network congestion, Chainlink updates may lag, allowing undercollateralized positions to persist longer than intended.

V2's aged codebase is complex and difficult to modify safely, as demonstrated by the $147M COMP distribution bug introduced via Proposal 62. Any governance-approved upgrade carries elevated risk of introducing new bugs into the legacy system.