Snark: On-chain data scaling

On-chain data is growing. Given the deterministic nature of a blockchain, this will keep growing. One strategy is checkpoints. Points in time we all agree to, ignoring historic data. We trust a specific point in time, destroying the nature of a blockchain.

What if we only needed to store states and balances. In “On-Chain data, off-chain execution”, we spoke towards achieving verified computing results which give us a merkle path and a zero knowledge proof. Applying this same logic on-chain, would mean that at any given point, we only needed to store the full merkle, and the current zero knowledge proof.

Let’s consider a block, a block is our proof of transactions in it. A block is proof of a state transition. State S1 when applied with block B1 gives us state S2. So B1 is a state transition. In “on-chain proof, off-chain execution” we showed how a state transition can be represented as a zero knowledge proof. So instead of a block, we could represent this state transition with a zero knowledge proof.

Ethereum block size is currently around 25k bytes, a zk-SNARK proof, would be around 288 bytes. At time of writing Ethereum is 1 099 511 627 776 bytes. By simply replacing the blocks with zero knowledge blocks, you would decrease the system to 12 195 622 907 bytes, 1% of the current size.

The above also considers you want to keep each state transition and merkle from the beginning of time, which isn’t required. This is a different form of verified computing, a full deterministic transaction history isn’t required. All that is required is the most recent merkle and most recent zk proof. So actual size will be less than a megabyte.