This is essay 1 of 4 for the BanklessDAO Writers Cohort.
The longer I spend writing about and otherwise inhabiting the cryptoverse, the more interested I am in the architecture behind blockchains, and the infrastructure we use to access these distributed ledgers. I’m not really technical in the developer sense, but I’ve always been interested in the Ethereum Improvement Proposals process, which governs how upgrades make their way from idea to adoption on the Ethereum network. For example, EIP-3675 precipitated the Merge, and before that, EIP-1559 kickstarted the ETH burning mechanism.
The most recent EIP to catch my eye is EIP-4844, adorably titled Shared Blob Transactions. If that isn’t a great title, its nickname is even better - proto-danksharding. Before we get into danksharding, it’s important to highlight an underreported benefit of the Merge. While all the attention was on the 99.95% energy reduction, the transition to the proof-of-stake consensus mechanism also set the stage for the remaining half of the Ethereum Roadmap. The Ethereum Roadmap is long and complex, but in terms of EIP-4844, all we have to know is that the Ethereum network is now scalable.
As Ethereum users know, the network needs to scale, and the proliferation of rollup and side chain scaling solutions are testament to that fact. But with proof-of-stake, Ethereum doesn’t need to only rely on Optimism, Arbitrum, and Polygon to scale; now the stage is set for scaling at the network level. And this Ethereum network scaling solution is called danksharding. No, danksharding is not a cheeky reference to really great sharding, but it is a reference to Dankrad Feist, the Ethereum researcher who initially proposed this scaling architecture.
As you probably know, a ‘shard’ is a sharply edged piece of ceramic, metal, glass, or rock. In terms of blockchain architecture, sharding is the idea that the network data can be split up into pieces, shards. Each shard would contain its own set of independent blockchain data, including token balances and smart contracts.
In practice, sharding helps to solve two problems: centralization and scaling. Right now, the Ethereum blockchain contains about 1TB of data, which makes it expensive to buy the storage capacity required to run a node. The more expensive it is to run a node, the less likely it is to be decentralized (and FYI, you can run a node without validating the network. Relatedly, the Ethereum blockchain requires that each transaction be processed by every node, which makes network-based scaling nearly impossible - as scaling is achieved when each node does less work. Sharding addresses both of these concerns by breaking up the blockchain data into independent parts, enabling lower storage requirements and nodes to process independent sets of blockchain data.
Danksharding is a one type of sharding architecture. While previous sharding frameworks sought to create more space for transactions, danksharding is a rollup-centric solution that creates more space for data-hungry rollups by using ‘blobs’ to store data. These data blobs are large, but are relatively cheap to transact with because the blob data is stored on the consensus layer rather than the computation-heavy execution layer. With data blobs, the execution layer need only be able to know what the commitments of the blob data are, not the details of that data.
While Danksharding is super complex, for the purposes of this article, you just need to keep in mind the main innovation behind it, which is the merged fee market. The merged fee market, despite its name, actually breaks out the job of building blocks and proposing that they be included on chain. The big change is that only the block builder needs to process all the data, while network validators verify the blocks through data availability sampling, which allows nodes to verify large amounts of data by ‘sampling’ some of it. Because nodes only need to sample some of the data to know it’s valid, the Ethererum network will be able to process way more data than it could before, providing a faster, cheaper network that is primed for both network-based scaling and rollup optimization.
All that sounds both confusing and amazing, and it is, but getting the Ethereum network to the point where it’s ready for sharding is going to take years. And this is where EIP-4844 comes in, as this proposal introduces proto-danksharding.
Proto-danksharding is the next step on the path to danksharding. EIP-4844 introduces a new transaction format that essentially adds space for the data ‘blob’ to accompany ordinary transactions. Full danksharing will add 16 MB of space to each block for rollup data; Proto-danksharding, however, will only add about 1MB of space. But even this small measure will make a big impact, as transacting on rollup-based scaling solutions will immediately be less expensive. Right now, blocks can carry between 50-100 kb of data, so the 1 MB of data availability targeted for rollups will reduce rollup transaction costs, as gas fees will not need to be paid to access this data.
In addition to introducing a new transaction format, EIP-4844 will also include the verification rules, logic, and gas fee adjustments to prepare for full-on danksharding. Notably, this proposal also includes a provision that historical network blob data will be deleted after 30 days, long enough for people and protocols who need the historical data to retrieve and store it.
It’s also important to note that while the proto-danksharding upgrade will introduce major changes to the Ethereum network, each node will still need to process all data. Nonetheless, proto-danksharding is a huge upgrade that will make transacting on rollups cheaper, while also preparing the Ethereum network for the rest of its roadmap.
Blockchain architecture is complex - and the Ethereum Roadmap even more so. If you take anything away form this article, just know proto-danksharding will make transacting on rollup scaling solutions cheaper in the short term, while providing the infrastructure needed to ensure that Ethereum is scalable, secure, and decentralized. While danksharding may still be a ways off, even just a 1/16 of the eventual blob size is enough to really expand our minds to see the possibilities of what interacting with a fully scaled Ethereum may look like.
Hiro Kennelly is a writer, editor, and coordinator at BanklessDAO and the Editor-in-Chief at Good Morning News. He is also helping to build a grants-focused organization at DAOpunks.