Two months ago, I spoke with someone from Fidelity to talk about using DeFi products at institutional scale. The conversation kept circling back to one issue: how can a CTO, responsible for billions in assets, trust code running on a public blockchain when even the best audits fail to catch everything? And even when they do, new patches, upgrades, and integrations introduce fresh risks that require retesting. There's no final state. Only a moving target.

This week's $40 million exploit on GMX illustrated how even mature protocols remain vulnerable. A vulnerability, believed to be a reentrancy bug, allowed an attacker to manipulate the minting of liquidity provider tokens—an accounting token representing a proportional claim on pooled assets, allowing redemption. GMX, often seen as among the simpler derivatives platforms, nonetheless suffered a critical failure. There was some reliance on third-party infrastructure, including Circle's USDC, but the response mechanisms weren't fast or robust enough to prevent loss. The lack of immediate on-chain recovery tools once again undermined trust in DeFi's operational security.

‍

The False Comfort of Time and Audits

Time in the market is not a guarantee of security. Neither is the number of audits. Both are necessary, but neither is sufficient. Especially when you're dealing with composable protocols, real-time pricing, margin calculations, and especially complex collateral management systems.

More sophisticated on-chain systems like clearing engines need more than static audits or simple proofs. We're not talking about a lending pool or a bridge where you can wrap logic and verify interactions through formal proofs, fuzzing, or even emerging efforts in runtime verification. On-chain clearing needs to solve QUBOs (Quadratic Unconstrained Binary Optimization problems) and manage state-dependent risk. This isn't static verification territory. It's dynamic, non-linear and conditional.

‍

There Is Only One Path Forward: Supervision

That leaves us with one viable path. If you want complex systems like clearinghouses or margin engines to operate safely on-chain, you need supervisory tooling. That means establishing a privileged access point. An oversight mechanism that can intervene under specific conditions.

Yes, that might sound like it breaks decentralization. But the answer is: it depends.

Normal trading flows? Trustless. Automated pricing? Trustless. Collateral movement? Trustless.

But if a wallet shows clear signs of exploiting a pricing bug or triggering an anomalous P&L spike, there must be a way to block withdrawals, unwind the transaction, or quarantine the funds. Whether that's triggered by transaction thresholds, abnormal margin behavior, or time-based withdrawal patterns, this isn't inconsistent with decentralization. It's about operational resilience. A circuit breaker, not a gatekeeper.

In fact, this is closer to how TradFi operates today. The reason traditional firms are comfortable with clearinghouses isn't because of how regulated they are. It is what this represents. It's because they trust that there are draconian, boring, and very manual processes—often dictated in regulatory licenses—that are in place that will stop a disaster when it starts. Translating that confidence to DeFi requires something similar. Not more hope, but more control.

‍

Own Chain Action vs. Complex Contract Logic

There are two ways to enforce supervisory actions:

Own Chain Access Control: Limiting execution permissions, gas use, or validator access for key contract interactions. This ensures only trusted parties can commit transactions under exceptional circumstances.

Smart Contract Complexity: Embedding fallback mechanisms, triggers, and administrative checks within the contract logic itself. Much harder to scale, and far riskier if bugs creep in.

Our view? Own chain access is cleaner, faster, and more reliable for supervisory functions. It doesn't interfere with trustless trading flows but ensures that if something goes wrong, there's an unambiguous route to intervene without a broad set of commitment capability. Multiple people can run validators but not everyone. We have trusted partners.

‍

Pascal and the Supervision Spectrum

This is exactly why Pascal, our on-chain clearing protocol, was built with supervisory logic baked in. It supports everything from:

• Full autonomy (no intervention) for internal treasury hedging and purest DeFi native use cases
• Supervised environments where market actors can define strict controls—enough to satisfy regulators if needed

For instance, if a group wanting to launch a regulated FX swap market could implement the necessary oversight. A protocol wanting internal delta hedging could opt out. The protocol flexes, depending on the market.

‍

Our Implementation Jetstream: Exploit Doesn't Mean Escape

Jetstream, our venue, already deploys basic supervisory measures. Exploits, if they occur or are attempted, do not mean immediate loss of funds. Transactions are flagged, withdrawals delayed, and abnormal behavior gets caught in flight.

That doesn't eliminate risk. The trading book would not look good. But it stops the worst outcome: an attacker walking away with all the cash. And that's already a meaningful step up from how most DeFi protocols operate. It is the first clear step in making sure there is no reward for expensive attempts to manipulate markets. Draconian yes, but safety wins here.

‍

The Road Ahead

Adoption in DeFi doesn't hinge on more TVL, liquidity mining, or meme-driven growth. It hinges on making smart contract systems that institutions can trust. That means:

• Accepting that audits and time in market are not enough
• Building circuit breakers and oversight tools
• Enabling layered supervision without sacrificing the benefits of decentralization
• A high-quality and supportive incentive scheme that engages productive development that understands users, not just casinos

If we want to keep the composability, transparency, and permissionless innovation that defines DeFi, we need to add one thing: accountable intervention.

That's how we make this thing safe enough to scale.

‍

About the Author

James Davies is the CEO of Pascal Protocol and Jetstream, the first platform launched on Pascal. He began writing high-frequency trading software in the 1990s and has worked at Trayport and BGC Partners, then chaired and controlled by Howard Lutnick, now U.S. Secretary of Commerce. His experience in traditional market infrastructure informs his current work: building the clearing layer DeFi needs—transparent, deterministic, and accessible.