One of the critical features of rollups are that they offer trustless bridges to a highly secure layer — something that is nigh impossible to achieve in a monolithic world. Or so we thought! =nil; Foundation is doing groundbreaking work to enable actually trustless bridges.
Consider the Ethereum-Mina bridge that -nil is building, commissioned by both Mina Foundation and Ethereum Foundation. Now, I have been highly critical of monolithic blockchains, but zk-monolithic chains are by far the most interesting in the space, particularly their final form of zk-sharded networks. With trustless bridges now possible, I’d even go as far as call them semi-modular, rather than monolithic! There are still some disadvantages versus rollups, which I’ll get into later.
To recap, just like zk rollups, Mina generates a validity proof — Pickles SNARK. (Leaving some technicalities aside — it’s actually a proof-of-a-proof that’s verified, and not Mina’s Pickles SNARKs directly, to optimize for gas costs.) Mina’s full state can thus be verified in Ethereum - again, just like zkRs. Given the assumption that Mina is healthy, the Ethereum-Mina bridge has similar properties to rollup bridges — an incredible achievement!
Of course, rollups still have a crucial advantage, because you don’t need to make the above assumption. Even if the rollup is compromised, the Ethereum-rollup bridge will continue to be operational. Thus, rollups effectively inherit Ethereum’s high security, while Mina does not. So, you’re still bound by Mina’s much less secure layer (which doesn’t yet exist — and as I’ve discusssed, there’s no shortcut to building network effects for a security layer).
Interestingly, if I understand correctly, Mina is now just one step away from becoming a validium. All it needs to do is respond to verification on Ethereum — a much more secure layer. What this means is if Mina is compromised, instead of thefts, reorgs, corruptions, and nasty outcomes, it can simply freeze and prevent invalid state transitions. Again, I’m definitely missing technical details here, but if my intuition is correct, this could give Mina exponentially high security by inheriting Ethereum’s security partially at no cost (as the bridge is verifying anyway). Take it one step further from there — commit compressed calldata on Ethereum — and you become a rollup. Now, I don’t expect Mina, or other zk-monolithic/semi-modular chains like Aleo or Dusk to actually do this, but the path is there if we enter a rollup-centric world where StarkNet or zkSync 2.0 are overwhelmingly popular like I expect. I will note that there may be some features that may not work as a rollup, but I don’t know what they are.
=nil; doesn’t just stop there. Their next project is intriguing — SNARKifying Solana’s light client for a more trust-minimized Ethereum-Solana bridge! (Side note: is this the first production use of RedShift?) Of course, this is just verifying light client — not quite as exciting as full state verification like Mina and zkRs. But still, a step forward on the trustlessness spectrum. I do wonder — what would a full state validity proof of Solana look like?