Bankless - Dive into Danksharding | Vitalik, Dankrad, Protolambda, Moderated by Tim Beiko
Primer: What is Proto-Danksharding and Danksharding and how does it relate to the broader Ethereum 2 network? In this Bankless episode, Tim Beiko moderates a conversation with panelists Vitalik Buterin, Dankrad Feist, and Protolambda on the topic.
The Panel
Tim Beiko
Manages the Core Dev Calls
Coordinating the Merge
Vitalik Buterin
Co-Founder of Ethereum
Co-Founder of Bitcoin Magazine
Dankrad Feist
Ethereum researcher since 2019
Protolambda
Used to work at Ethereum Foundation as part of their ETH 2 Research Team
Currently working at Optimism doing the same thing
What Is Danksharding?
Danksharding and proto-danksharding are iterations over Ethereum’s sharding design
Sharding is a way for Ethereum to have more data pass through the network
“All these flavors of sharding all have the same end goal, which is to create a cheap place for layer two solutions to post data on Ethereum.”
- Tim Beiko
EIP-4844 is proto-danksharding
Proto-danksharding is the first step towards sharding
Danksharding is the simplification that they had on the previous roadmap
Shift From Full Execution Sharding To Only Data Sharding
There have been an ongoing simplification of the sharding roadmap since its release in 2016
Their first step was the decision to not bother with anything beyond quadratic sharding
Their second step was to have every shard block to get directly included in the beacon chain
Instead of shard blocks containing transactions that will be executed at the Ethereum layer, they shifted to shard blocks would just contain big blobs of data
Layer 2 rollup protocols would be responsible to use that data space to create secure and scalable experiences for their users
Next is Danksharding, where there’s only one proposer that chooses all of the shard blocks and shards that appear within a particular beacon block
Currently halfway along the timeline to get full sharding out
Why Is It Possible To Implement Danksharding Without Sacrificing The Decentralization Of The Chain?
Proof-of-Stake has proposer-builder separation
Traditionally, the block proposer builds the block
Proposing is simple and cheap whereas building is a complex process
Building is more suited towards capitalized entities — people with larger machines that can handle a large amount of data
Anyone can verify the blocks are correct
How Does EIP-4844 Help Layer 2s?
Danksharding introduces data availability sampling. It attempts to better distribute the job across the network
They can optimize for the security properties that L2s need
How it helps Layer 2s:
Publishing the data and making sure that the honest minority that protects the L2 is able to get the data in the first place
Ensures that actors are able to get the data even with downtime, censorship, unforeseen circumstances, etc.
What Is Data Availability And How Is It Different From Data That Is Stored On Ethereum
Data availability: Did the data go through a publication process where it got broadcast on a public network and anyone who wanted to download the data had the time to do so?
When Ethereum is compared to IPFS, Ethereum is able to provide consensus on data availability
Layer 2 protocols depend on data being available out there
Example of ZK Rollup:
Sequencer accepts transaction —> sequencer publishes blocks that contain state-delta
The sequencer manages the internal state
Difference between ZK Rollup and Validium:
In ZK Rollup, state-delta are on-chain
In Validium, only the proofs are on-chain. Everything else is off-chain
Only difference is what happens when the sequencer disappears. In Validium, this is a problem as the money gets stuck there forever if the sequencer disappears
The beacon chain would only contain hashes of data
The shards do not need to store the data forever. The point is to give enough time for people who want to download the data to download it
How Does Data Availability Sampling Work?
We need a scalable way to ensure that some amount of data is available to be downloaded when people want to download it
Data availability sampling involves picking random parts of the data and trying to get access to them
If you could get access to the majority of the data, that’s when you consider the data to be available
Reed-Solomon encoding enables you to reconstruct the entire data set if any 50% of the data is available
This enables scaling as it does not require 100% of the data to be available
Attackers have to reduce data availability beyond 50% in order to be able to attack the network
Downloading every 10% of the data decreases attacks by a factor of 10. This is a scalable way to ensure data availability
The EIP-4844 World
Continue to scale the consensus layer
Some of the data will be retained. After which, they will be pruned
The data will be available for a sufficient amount of time for it to secure the network
The amount of data that is available in proto-danksharding is less than the amount that is available in full danksharding
How Much Does It Lower The Cost Of Storing Data?
Currently, Ethereum blocks are between 50-100 kbs
Can grow it from 50 kbs to a mb per block
Rollups will benefit from reduced costs
In full danksharding, it will be an order of magnitude increase
Difference Between People Who Pay A Lot To Execute Computations VS People Who Pay A Lot To Store Data
Added additional fee parameters to create different markets
Vitalik has wrote a post on Multidimensional EIP-1559
KZG Commitments
When the data is sampled, one is not able to tell whether the encoding is correct
KZG commitment is a way to hash a polynomial function and guarantees that the encoding is correct
What Is A Trusted Setup?
Have to set up elliptic curve points that have a certain relation. This is one of the fundamental inputs of the KZG commitment scheme
Nobody is allowed to know the actual relation between them
In a trusted setup, even if only one person did it properly and everyone else colluded, then that setup is completely safe
When Will EIP-4844 Get Deployed?
Earlier this year, they wrote a proposal for it
In the ETHDenver hackathon, they implemented it and have been testing it ever since
Two different branches:
Further develop the client software and to create a testnet
The development of the trusted setup
Once it is complete, they would include it as an EIP
Proto-danksharding is a stepping stone towards the full implementation
Trying to get it done in the Shanghai Hard Fork
If People Want To Contribute From A Research/Engineering Standpoint, What Are Some Of The Open Questions In Sharding?
Figuring out the networking of data availability sampling
Would be good if there’s more people with networking expertise
Economic challenges:
How do you make a good proposer-builder separation protocol?
How do you add censorship resistance to bypass censoring builders?
What would a better Proof-of-Stake design look like over the long-term?
The switch to Layer 2s and ways to increase L2 decentralization
Anything Else That They Would Like To Share?
Protolambda
If you feel down about the bear market, try to read a post and get involved with projects
Visit eip4844.com to learn more
Vitalik
“Stay optimistic. But in the long term, hopefully you'll stay zero knowledge.”
- Vitalik Buterin
Dankrad
The bear market has shown that you need solid designs and to build things that are able to last
All information presented above is for educational purposes only and should not be taken as investment advice. Summaries are prepared by The Reading Ape. While reasonable efforts are made to provide accurate content, any errors in interpreting and summarizing the source material are ours alone. We disclaim any liability associated with the use of our content.