AL #83: The 4 Rollup Architectures and the Future of Gnosis Chain
How apps inform diverse rollup architectures.
This post is taking part in the Devconflict x Kiwi writing contest.
“To be or not to be an L2”.
While this is a question that too many people are asking, few can answer it well.
I can't think of a much better combo than Vitalik and Martin Köppelmann from Gnosis to try.
This is exactly what happened this Saturday at DevConflict where Martin defended the choice for Gnosis to exist as an alternative L1.
Without repeating it, the debate covered some really fun ideas:
The idea that Gnosis could become a semi-based rollup with 6 second block times, half of which settle on Ethereum (Martin)
The possibility of Ethereum introducing 100 interoperable native based rollups (Martin)
Introducing ZK precompiles to support heterogeneous based rollups (Vitalik)
The idea that rollups will bifurcate between based rollups and centralized rollups with low latency pre-confirmations (Vitalik). As an aside, I think Unichain is an interesting counterpoint because it has low latency pre-confirmations but also strives to be less centralized than most rollups. I'm biased, however.
Ultimately Martin/Vitalik chose to focus on the security axis (it's an Ethereum event after all).
Martin essentially argued that Gnosis chain has less bridge risk due to limited reliance on code itself.
Vitalik countered by saying that the entire smart contract movement is predicated on the idea that code can become safe enough. He further explained that code gets better with time, trust doesn't.
Nevertheless, Vitalik agreed that the use of multisigs to diversify risk is pragmatic in the short term and he expects chains to move closer to idealized L2s as they progress along the security axis.
What I found fascinating is that the debate didn't touch on apps nearly enough.
Martin briefly mentioned that Gnosis chain isolates Gnosis Pay (a payment solution) from blob market volatility, but that was it.
If you’re not familiar with the Gnosis product suite, we've covered it before:
Martin also stressed that Gnosis Chain would always be looking to occupy a unique niche.
E.g., it is occupying a unique niche as an L1 and if it were to become an L2 it would look to differentiate itself against the “127” other L2s.
But isn't the opposite approach more natural?
Design L2s not to be differentiated from other L2s but to be the best possible rollup for a cluster of apps.
It's all about the apps
Apps have clear performance requirements (security, latency, cost) and interop requirements.
A standalone game may not need to interoperate with anything but a base chain for its token.
A platform protocol like Uniswap v4 needs to be highly interoperable, however.
These requirements necessitate entirely different kinds of blockspace.
The four types of apps
We place apps along 2 axes:
The Hardness to Latency axis, covering the trade-offs between based rollups that inherit Ethereum security and centralized rollups that ensure low latency confirmations
The Fairness to Cost axis, covering the trade-offs between rollups that support a fair transaction priority market as opposed to chains that optimize for transaction costs and their stability.
Different apps are expressed differently across these axes.
Use cases like trading require low latency and a fair & transparent priority fee market. These require centralized or rotating sequencers with transparent ordering to achieve low transaction latency.
Micro-payments (in gaming, agentic web, etc.) should prioritize low cost and low latency. These benefit from alt-DA often as part of an L3 package.
Transaction banking use cases (institutional payments) are cost optimizing but may not be as time sensitive, relying on corporate payment terms.
Finally large auctions (NFTs, large token sales) need to have fair auctions and strong security guarantees for settlement. These may be the kinds of properties that based rollups can provide.
While this is a very simplified take, it hopefully illustrates that we should think about apps and their requirements first and then rollups.
In a Gnosis context, their choice of architecture may depend on answering what sets of apps are most important to serve well and where do they fit on the spectrum.
For now they are able to achieve similar or better levels of security than rollups at a low & stable cost for transactions which serves their payment networks well.
But as L2s develop, the best architectures will evolve with them and I would expect alt-L1s may not be able to match Ethereum L2 level hardness going forward.
It’s possible that the top left corner gets replaced by validiums or another architecture entirely.
That’s when Gnosis should reevaluate this choice.