Curve Finance’s current near-death expertise (and its averted propagation) could appear to be a blur in Web3’s rear-view mirror, but it surely’s truly one thing that retains occurring within the business. It’s not the primary time {that a} decentralized finance protocol — or any decentralized app for that matter — has been affected by an assault that’s completely authorized inside its personal code. Extra so, the disaster might’ve been prevented if on-chain danger administration existed.
All of this factors to a broader downside in Web3. That’s the downside of restricted expressivity and assets that exist in its improvement environments and the way it impacts safety general.
Hack or exploit?
When the Curve Finance attacker was capable of retrieve US$61.7 million in belongings from Curve Finance’s good contracts, many media shops and commentators referred to as the occasion a “hack.” However this was not a hack — it was an exploit. The distinction right here is essential.
On this context, a hack would’ve taken place if the attacker had in some way bypassed or damaged an current safety measure. However the assault on Curve was an exploit. Nothing that occurred that was out of the extraordinary when it comes to what the protocol’s Vyper code allowed for. The looter merely took benefit of how the protocol’s design labored.
Who’s responsible for this? Nobody. Curve’s Vyper code, like a lot of the (Solidity) code that’s utilized in Web3 functions, is severely restricted in its capacity to specific complexity past comparatively easy transaction logic.
This makes it laborious for anybody to design safety measures that may stop this or another assaults. Extra worryingly, it additionally makes it laborious for anybody to correctly design instruments to stop their unfold throughout DeFi’s huge and composable liquidity panorama.
On-chain danger evaluation
But it surely doesn’t imply there was nothing Curve might do to stop this assault and its unfold throughout DeFi. A easy instance of an answer could be on-chain danger evaluation.
The generalized model of a problematic sample that could possibly be solved may be summarized in a hypothetical scenario like this one:
- Unhealthy actor Bob buys $5 million value of the extremely unstable $RISKY token through a flashloan.
- The worth of $RISKY token is successfully pumped by Bob after the acquisition.
- Bob takes out a $100 million mortgage on Naive Finance backed by $RISKY.
- Naive Finance checks the value of $RISKY and confirms that Bob is “good” for the cash.
- Bob runs.
- When Naive Finance liquidates $RISKY it is just value $5 million.
(One other instance of this basic sample may be discovered within the Euler hack from March.)
Historically, this downside is solved by danger evaluation options that decide how good of a assure an asset may be. In the event that they existed on-chain, Naive Finance might examine statistical estimations primarily based on the token’s historic value earlier than approving the mortgage. The protocol would’ve seen by way of the pump and denied Bob the $100 million.
DeFi is missing this sort of on-chain danger evaluation and administration.
Going again to Curve Finance, a variety might’ve been prevented if Aave and Frax had an automatic, on-chain restrict on mortgage approvals once they cross a proportion of the collateral token’s circulating provide. This could’ve been a safer and fewer stress-inducing scenario for everyone.
Restricted expressivity and assets
The true downside right here is that present Web3 ecosystems can’t assist one thing like this on-chain danger evaluation answer. They’re restricted by the sort of libraries and frameworks which are out there in digital machines just like the Ethereum Digital Machine. They’re additionally restricted when it comes to the assets at their disposal.
To be able to develop one thing like this danger evaluation and administration answer, a decentralized app would wish to rely on coding libraries which have features for at the least fundamental mathematical ideas like logarithms and others.
This isn’t the case in Web3 as a result of dApps don’t have entry to NumPy, the maths module in Python, for instance. The standard toolbox isn’t there and builders should reinvent the wheel as a substitute.
Then we have now one other downside. Even when that they had these libraries, they might be too costly to code. Actually costly. The Ethereum Digital Machine is designed in order that there’s a value for each computation.
Whereas there are legitimate causes for this, akin to stopping infinite loops and such, it additionally creates a useful resource limitation for dApps which may have to scale computationally with out incurring unreasonable prices. One might simply see how a danger administration answer would value extra to run than what it’s capable of save in funds.
Specializing in the correct issues
At a localized stage, the unfold of the Curve Finance deadlock might’ve been prevented with on-chain danger administration. At a basic stage, this complete class of assaults could possibly be prevented with extra expressivity and assets in Web3.
These are two elements of blockchain scalability which have lengthy been ignored as a result of they transcend affording extra shared block area for dApps. They really contain the creation of improvement environments in Web3 that emulate these of Web2. They’re about computational scalability and programmability, not simply scaling the quantity of knowledge that’s out there on-chain.
Maybe if protocol builders at Curve, Aave or Frax had the power to rely on a greater toolbox and extra assets, these and future exploits could possibly be averted altogether. Possibly we might begin with on-chain danger administration.