Subscribe to Terminally Onchain by YB
>9.8K subscribers
Subscribe to Terminally Onchain by YB
>9.8K subscribers
Share Dialog
Share Dialog
Welcome to the 5 new people who joined the onchain letters community since Sunday!
I'm thankful for all 74 of you and I hope everyone is having a great start to the week 🔥
If you're enjoying my writing, please share Onchain Letters with your friends in Crypto 🤝
After combing through a variety of resources on all things layer 2, I decided to write this letter to articulate my thinking on what clicked for me - if you're confused and want to simply understand what's happening with Rollups/4844/Blobs/Proto-Danksharding then hopefully this helps :)
Last week Ethereum Goerli upgraded to EIP-4844! The ecosystem is one step closer to having proto-danksharding on mainnet.
To be honest, if you asked me what that sentence meant a few days ago, I would have no idea what to say other than "good for L2s". So I decided to go down the rabbithole in order to form my own opinion on 4844 and create a framework on how to view L2s this upcoming year.
The letter has 4 sections:
Even Ethereum Iterates
The Rise of DSL
Ethereum's Broadband Moment
Looking ahead
Let's dive in 🚀
Ethereum is constantly evolving. Even the gigabrains doing core protocol research are always iterating on what the best path forward is.
It's worth noting that before 2020, the plan to scale Ethereum had always been focused on sharding.
However, the Ethereum Foundation realized that sharding is a lot more difficult to implement in the L1 architecture and would take significantly more time and money than everyone initially thought.
On the other hand, by pivoting to a rollup centric roadmap, it would be possible to offload the scaling experimentation to the markets. Rollups could raise money, try new mechanisms, compete with each other, and let the onchain activity prove what works best.
For context, Loopring went live in December 2019 and that marked the first DEX on a ZK rollup. And soon after, ZK Sync and StarkEx launched as well.
By October 2020, the Ethereum core team decided that the best path forward for the near future was to focus on rollups. Here's a post from Vitalik on the Ethereum Magicians forum:
And now, more than 4 years after Loopring launched, we have a thriving L2 ecosystem and it's possible to transact onchain reliably for under $1. In the past year, I've found myself consistently using L2s such as Optimism, Base, Zora, etc. At times, looking back at the last bull market makes me cringe...I have memories of paying upwards of $300 in gas for certain NFTs 😭
Today, there's $21.67 billion locked up in L2s and L2 Beat is tracking 36 active projects. Take a look at this graph - there's clearly something special happening here. The Ethereum ecosystem is evolving and the mindset around transacting onchain has actually changed since the last bull.
In fact, the L2 ecosystem has grown so much, that there is a whole spectrum of solutions and many of these projects have large, dedicated communities. For example, Optimism recently ran a public goods funding round where they rewarded 500 builders with 30 million $OP. That comes out to approximately $90 million at OP's current price 🤯
My point is that rollups are here to stay. In the past 4 years they've made a mark on crypto and we'll look back at fall 2020 as one of the watershed moments of Ethereum scaling.
So what's next?
Though the L2 ecosystem has made a ton of progress, we're still in the first innings.
The most important thing happening that's disrupting L2s right now is EIP-4844. This proposal is known as "Proto-Danksharding", named after two of the core authors Protolambda and Dankrad Feist.
To better understand the proposal, it's worth expanding on this sentence in Coinbase's Q3 earnings report:
The transition from L1 networks to L2 networks can be likened to the transition from dial-up Internet to broadband.
I think this example is directionally correct but can be improved. To understand what I'd change, I'll explain the rise of DSL in this section and cover the parallels of 4844 in the next.
In the first era of the internet, people went online by using dial-up modems. Meaning, they accessed the internet through an active phone line. So at any given time, you could be either using the phone or the internet, but not both.
The first commercial modem, the Bell 103, was released by AT&T in 1962 and ran at 300 bps (bits per second)! Over the next 30 years, there were constant updates to the modem and the speeds increased from 300 bps to 56k bps.
However, due to constraints of phone lines (see Shannon's limit) and FCC regulation, 56kbs was the maximum speed people could access the internet using dial up modems.
So...imagine a world where all of us were still stuck with that kind of internet speed. It would have been a lot tougher for the internet to go mainstream and provide the ubiquitous utility it does today.
Fortunately, two engineers at Bell Labs, Richard Gitlin and Jean Werner filed a patent for a new technology called digital subscriber lines, or DSL.
The incredible part about DSL is that it used the same underlying phone lines but operated at a higher frequency for data transmission. This not only bypassed the bandwidth on traditional modems but also allowed for the internet to always "be on". Now, users didn't have to pick which one to use at any given time, they ran in parallel.
Dial up connections use the voice circuit for data transfer so the bandwidth is limited to bandwidth of voice channel, whereas DSL uses a separate frequency range for data which is much broader than voice bandwidth (hence the term Broadband). -Superuser
It's safe to say that without Broadband, the internet would be nothing like it is today. Gaming, video streaming, etc. would be effectively impossible at scale. It was broadband that finally increased internet bandwidth and decreased costs so that the average user could simply plugin and get started.
Okay, now that we have context on how broadband changed internet bandwidth, let's dive into 4844.
Ethereum has an execution layer where transactions are processed and also has a consensus layer that updates & broadcasts the state of the blockchain.
Rollups are a scaling solution that act as blockchains on top of Ethereum. They effectively mimic how Ethereum operates and still prioritize decentralization and consensus.
The only new component are sequencers which are responsible for batching transactions and including them on the main Ethereum blockchain.
After the rollup finishes executing a set of transactions, the sequencers post the transaction data onto Ethereum via calldata. Note: calldata is data passed as arguments in a transaction. This is way more efficient than if every transaction was executed on the L1 itself. And clearly, it works. Today, it's cheaper to transact on any L2 compared to Ethereum.
But the truth is that even these fees are unacceptable for the mainstream audience. They may seem attractive to those of us in the ecosystem who are seeing the dramatic change from the previous 3-digit fees. However, the rest of the world doesn't care and it's our job to continue bringing the prices down.
So how do we do this?
Calldata is great but can still be expensive for rollups.
Is it necessary to have the sequencers post there?
Well, turns out the answer is no. Ethereum is a billboard, not a database. Meaning that the goal is not to have the transaction data of every single interaction since the genesis block. Rather, the promise of Ethereum is to be a flashy billboard that anyone can look at and know that the current state of the blockchain is correct and has been verified.
If that's the case, then if we can have the data available on the Ethereum blockchain for a long enough time to prove that it's valid, then we don't need the data after that point.
That's where 4844 comes in.
The simplest way to explain this proposal is to use our DSL example from above. We discussed that the key innovation was the fact that DSL enabled internet connection to run on a different frequency than the phone connection.
Similarly, 4844 is creating a new frequency on the Ethereum blockchain known as blobspace which will allow Rollups to run in parallel of the main L1.
So now, instead of having the sequencers post the transaction data into "calldata", they'll instead post in the new blobspace. And there will be a link of each blob to the blockchain. This will make it 5-10x cheaper for rollups to update their status on Ethereum.
Just as it didn't make sense to have the phone and internet on the same frequency, there's no need to have core Ethereum transactions and rollup transactions running together.
If you partition to different "frequencies", there's a huge bandwidth unlock.
Here are the key numbers you need to know:
18 days: how long the blob data is stored and can be checked before Ethereum validators will delete
50 gb: how much validators will have to increase their storage by to hold blob data
128 kb: size of each blob
Each block can hold up to 6 blobs max but the average will probably 3 blobs
Now, given these numbers, you start to realize that there is a huge unlock in transaction bandwidth for the Ethereum ecosystem:
In any case, a block with six full blobs attached does not represent an inconsequential increase in a block’s size. If a block today can be as large as ~1.875MB, and a full set of blobs can add as much as ~0.75MB, we should note the possibility of increasing the block size by as much as ~40%. -Consensys
Increasing block size by as much as 40% 🤯
And this is just with proto-danksharding. In a few years, once danksharding is implemented, that number will moon.
I'll cover the technicals around danksharding in a future post but for right now just know that danksharding will make it possible to have way more than 6 blobs. This is because nodes will not have to download every single blob to verify but will instead use new mechanisms known as data availability sampling and proposer builder separation.
So, how would I change Coinbase's initial statement?
The transition from Rollups using blockspace to blobspace can be likened to the transition from dial-up modems to broadband.
It wasn't the rise of L2s but rather having these L2s run on a separate frequency that will define Ethereum's broadband moment. We're just getting started.
Blob Services
It's fair to say that some people won't be fans of the fact that there's no access to blob data after 18 days. So a low hanging fruit will be for another protocol in the ecosystem to provide a subscription type service that will store all the user's relavant blob data.
There are also other ideas such as blob sharing protocols that you can look into.
Induced Demand
Dan Romero made a point that 4844 might not make the rollup fees as cheap as we think because of induced demand. Basically, when more bandwidth opens up, demand increases and prices go back to what they were.
However, Jesse Pollak made a counter that even today the demand is not getting filled up though we have cheaper fees.
My guess is that initially the effect of lower prices will be there but as the bull market comes along and we see a flood of activity, then the bandwidth will eventually get filled up and prices will rise.
It's just like how a new lane on a highway feels fresh and empty at first, but over time it also suffers the same congestion problem as the rest of the highway.
To me, either way it's a win. One leads to cheaper prices and the other means that there's more things happening onchain!
If you're interested, you can also see what the community is betting on Polymarket as well.
Not all L2s are Equal
Today, I personally use all of the following: iMessage, WhatsApp, Messenger, Telegram, Signal, Instagram DMs, Twitter DMs, Discord, etc.
And each one serves a different purpose. iMessage for my friends and daily use, WhatsApp to communicate with my family in India, Telegram for crypto friends, Discord for DAOs, etc.
They're all messaging services but come in different flavors. Of course, there will naturally be a few dominant players (i.e. iMessage, WhatsApp) but that shouldn't take away from the functionality of niche solutions (i.e. Signal, Telegram).
Similarly, a lot of these L2s are taking different approaches and are aiming to serve different purposes.
It'll be interesting to see how teams think about using blobs. Those focused on security will charge a bit higher and purchase blob real estate every block. While others that are focused on eliminating transaction fees will be price efficient and upload to blobs when it's cheaper.
Bookmark L2 Beat
Over the next year, I'll be watching L2 beat closely. Where is the money flowing? What projects are popping up? Is there a long tail of rollups that are forming?
L2 beat tracks the decentralization stages of a rollup as well. All the solutions today are operating on centralized sequencers. Over the next year, it will be essential that the rollups work their way to stage 2 (the final one).
The focus can't be just about the # of users or TVL - if we don't make progress on decentralizing scaling solutions, then it defeats the whole purpose of Ethereum in the first place.
Don't forget about L1s
DCinvestor makes an important point that the ecosystem can't be only focusing on L2s. The core L1 still plays a crucial role as the liquidity/settlement layer. People who are going to be eventually transacting in the millions and billions will depend on the L1 to work just as securely and efficiently.
I know today's post was long so if you're reading this, thanks for sticking with me!
Just remember, the name of the game is bandwidth. The internet only became popular after broadband technology started spreading. Like DSL, blobspace is providing us a new frequency for rollups every single block! It's like the Ethereum mayor approved a project to build a 6 lane highway to help shorten your morning commute.
That's all for today's letter!
- YB
Welcome to the 5 new people who joined the onchain letters community since Sunday!
I'm thankful for all 74 of you and I hope everyone is having a great start to the week 🔥
If you're enjoying my writing, please share Onchain Letters with your friends in Crypto 🤝
After combing through a variety of resources on all things layer 2, I decided to write this letter to articulate my thinking on what clicked for me - if you're confused and want to simply understand what's happening with Rollups/4844/Blobs/Proto-Danksharding then hopefully this helps :)
Last week Ethereum Goerli upgraded to EIP-4844! The ecosystem is one step closer to having proto-danksharding on mainnet.
To be honest, if you asked me what that sentence meant a few days ago, I would have no idea what to say other than "good for L2s". So I decided to go down the rabbithole in order to form my own opinion on 4844 and create a framework on how to view L2s this upcoming year.
The letter has 4 sections:
Even Ethereum Iterates
The Rise of DSL
Ethereum's Broadband Moment
Looking ahead
Let's dive in 🚀
Ethereum is constantly evolving. Even the gigabrains doing core protocol research are always iterating on what the best path forward is.
It's worth noting that before 2020, the plan to scale Ethereum had always been focused on sharding.
However, the Ethereum Foundation realized that sharding is a lot more difficult to implement in the L1 architecture and would take significantly more time and money than everyone initially thought.
On the other hand, by pivoting to a rollup centric roadmap, it would be possible to offload the scaling experimentation to the markets. Rollups could raise money, try new mechanisms, compete with each other, and let the onchain activity prove what works best.
For context, Loopring went live in December 2019 and that marked the first DEX on a ZK rollup. And soon after, ZK Sync and StarkEx launched as well.
By October 2020, the Ethereum core team decided that the best path forward for the near future was to focus on rollups. Here's a post from Vitalik on the Ethereum Magicians forum:
And now, more than 4 years after Loopring launched, we have a thriving L2 ecosystem and it's possible to transact onchain reliably for under $1. In the past year, I've found myself consistently using L2s such as Optimism, Base, Zora, etc. At times, looking back at the last bull market makes me cringe...I have memories of paying upwards of $300 in gas for certain NFTs 😭
Today, there's $21.67 billion locked up in L2s and L2 Beat is tracking 36 active projects. Take a look at this graph - there's clearly something special happening here. The Ethereum ecosystem is evolving and the mindset around transacting onchain has actually changed since the last bull.
In fact, the L2 ecosystem has grown so much, that there is a whole spectrum of solutions and many of these projects have large, dedicated communities. For example, Optimism recently ran a public goods funding round where they rewarded 500 builders with 30 million $OP. That comes out to approximately $90 million at OP's current price 🤯
My point is that rollups are here to stay. In the past 4 years they've made a mark on crypto and we'll look back at fall 2020 as one of the watershed moments of Ethereum scaling.
So what's next?
Though the L2 ecosystem has made a ton of progress, we're still in the first innings.
The most important thing happening that's disrupting L2s right now is EIP-4844. This proposal is known as "Proto-Danksharding", named after two of the core authors Protolambda and Dankrad Feist.
To better understand the proposal, it's worth expanding on this sentence in Coinbase's Q3 earnings report:
The transition from L1 networks to L2 networks can be likened to the transition from dial-up Internet to broadband.
I think this example is directionally correct but can be improved. To understand what I'd change, I'll explain the rise of DSL in this section and cover the parallels of 4844 in the next.
In the first era of the internet, people went online by using dial-up modems. Meaning, they accessed the internet through an active phone line. So at any given time, you could be either using the phone or the internet, but not both.
The first commercial modem, the Bell 103, was released by AT&T in 1962 and ran at 300 bps (bits per second)! Over the next 30 years, there were constant updates to the modem and the speeds increased from 300 bps to 56k bps.
However, due to constraints of phone lines (see Shannon's limit) and FCC regulation, 56kbs was the maximum speed people could access the internet using dial up modems.
So...imagine a world where all of us were still stuck with that kind of internet speed. It would have been a lot tougher for the internet to go mainstream and provide the ubiquitous utility it does today.
Fortunately, two engineers at Bell Labs, Richard Gitlin and Jean Werner filed a patent for a new technology called digital subscriber lines, or DSL.
The incredible part about DSL is that it used the same underlying phone lines but operated at a higher frequency for data transmission. This not only bypassed the bandwidth on traditional modems but also allowed for the internet to always "be on". Now, users didn't have to pick which one to use at any given time, they ran in parallel.
Dial up connections use the voice circuit for data transfer so the bandwidth is limited to bandwidth of voice channel, whereas DSL uses a separate frequency range for data which is much broader than voice bandwidth (hence the term Broadband). -Superuser
It's safe to say that without Broadband, the internet would be nothing like it is today. Gaming, video streaming, etc. would be effectively impossible at scale. It was broadband that finally increased internet bandwidth and decreased costs so that the average user could simply plugin and get started.
Okay, now that we have context on how broadband changed internet bandwidth, let's dive into 4844.
Ethereum has an execution layer where transactions are processed and also has a consensus layer that updates & broadcasts the state of the blockchain.
Rollups are a scaling solution that act as blockchains on top of Ethereum. They effectively mimic how Ethereum operates and still prioritize decentralization and consensus.
The only new component are sequencers which are responsible for batching transactions and including them on the main Ethereum blockchain.
After the rollup finishes executing a set of transactions, the sequencers post the transaction data onto Ethereum via calldata. Note: calldata is data passed as arguments in a transaction. This is way more efficient than if every transaction was executed on the L1 itself. And clearly, it works. Today, it's cheaper to transact on any L2 compared to Ethereum.
But the truth is that even these fees are unacceptable for the mainstream audience. They may seem attractive to those of us in the ecosystem who are seeing the dramatic change from the previous 3-digit fees. However, the rest of the world doesn't care and it's our job to continue bringing the prices down.
So how do we do this?
Calldata is great but can still be expensive for rollups.
Is it necessary to have the sequencers post there?
Well, turns out the answer is no. Ethereum is a billboard, not a database. Meaning that the goal is not to have the transaction data of every single interaction since the genesis block. Rather, the promise of Ethereum is to be a flashy billboard that anyone can look at and know that the current state of the blockchain is correct and has been verified.
If that's the case, then if we can have the data available on the Ethereum blockchain for a long enough time to prove that it's valid, then we don't need the data after that point.
That's where 4844 comes in.
The simplest way to explain this proposal is to use our DSL example from above. We discussed that the key innovation was the fact that DSL enabled internet connection to run on a different frequency than the phone connection.
Similarly, 4844 is creating a new frequency on the Ethereum blockchain known as blobspace which will allow Rollups to run in parallel of the main L1.
So now, instead of having the sequencers post the transaction data into "calldata", they'll instead post in the new blobspace. And there will be a link of each blob to the blockchain. This will make it 5-10x cheaper for rollups to update their status on Ethereum.
Just as it didn't make sense to have the phone and internet on the same frequency, there's no need to have core Ethereum transactions and rollup transactions running together.
If you partition to different "frequencies", there's a huge bandwidth unlock.
Here are the key numbers you need to know:
18 days: how long the blob data is stored and can be checked before Ethereum validators will delete
50 gb: how much validators will have to increase their storage by to hold blob data
128 kb: size of each blob
Each block can hold up to 6 blobs max but the average will probably 3 blobs
Now, given these numbers, you start to realize that there is a huge unlock in transaction bandwidth for the Ethereum ecosystem:
In any case, a block with six full blobs attached does not represent an inconsequential increase in a block’s size. If a block today can be as large as ~1.875MB, and a full set of blobs can add as much as ~0.75MB, we should note the possibility of increasing the block size by as much as ~40%. -Consensys
Increasing block size by as much as 40% 🤯
And this is just with proto-danksharding. In a few years, once danksharding is implemented, that number will moon.
I'll cover the technicals around danksharding in a future post but for right now just know that danksharding will make it possible to have way more than 6 blobs. This is because nodes will not have to download every single blob to verify but will instead use new mechanisms known as data availability sampling and proposer builder separation.
So, how would I change Coinbase's initial statement?
The transition from Rollups using blockspace to blobspace can be likened to the transition from dial-up modems to broadband.
It wasn't the rise of L2s but rather having these L2s run on a separate frequency that will define Ethereum's broadband moment. We're just getting started.
Blob Services
It's fair to say that some people won't be fans of the fact that there's no access to blob data after 18 days. So a low hanging fruit will be for another protocol in the ecosystem to provide a subscription type service that will store all the user's relavant blob data.
There are also other ideas such as blob sharing protocols that you can look into.
Induced Demand
Dan Romero made a point that 4844 might not make the rollup fees as cheap as we think because of induced demand. Basically, when more bandwidth opens up, demand increases and prices go back to what they were.
However, Jesse Pollak made a counter that even today the demand is not getting filled up though we have cheaper fees.
My guess is that initially the effect of lower prices will be there but as the bull market comes along and we see a flood of activity, then the bandwidth will eventually get filled up and prices will rise.
It's just like how a new lane on a highway feels fresh and empty at first, but over time it also suffers the same congestion problem as the rest of the highway.
To me, either way it's a win. One leads to cheaper prices and the other means that there's more things happening onchain!
If you're interested, you can also see what the community is betting on Polymarket as well.
Not all L2s are Equal
Today, I personally use all of the following: iMessage, WhatsApp, Messenger, Telegram, Signal, Instagram DMs, Twitter DMs, Discord, etc.
And each one serves a different purpose. iMessage for my friends and daily use, WhatsApp to communicate with my family in India, Telegram for crypto friends, Discord for DAOs, etc.
They're all messaging services but come in different flavors. Of course, there will naturally be a few dominant players (i.e. iMessage, WhatsApp) but that shouldn't take away from the functionality of niche solutions (i.e. Signal, Telegram).
Similarly, a lot of these L2s are taking different approaches and are aiming to serve different purposes.
It'll be interesting to see how teams think about using blobs. Those focused on security will charge a bit higher and purchase blob real estate every block. While others that are focused on eliminating transaction fees will be price efficient and upload to blobs when it's cheaper.
Bookmark L2 Beat
Over the next year, I'll be watching L2 beat closely. Where is the money flowing? What projects are popping up? Is there a long tail of rollups that are forming?
L2 beat tracks the decentralization stages of a rollup as well. All the solutions today are operating on centralized sequencers. Over the next year, it will be essential that the rollups work their way to stage 2 (the final one).
The focus can't be just about the # of users or TVL - if we don't make progress on decentralizing scaling solutions, then it defeats the whole purpose of Ethereum in the first place.
Don't forget about L1s
DCinvestor makes an important point that the ecosystem can't be only focusing on L2s. The core L1 still plays a crucial role as the liquidity/settlement layer. People who are going to be eventually transacting in the millions and billions will depend on the L1 to work just as securely and efficiently.
I know today's post was long so if you're reading this, thanks for sticking with me!
Just remember, the name of the game is bandwidth. The internet only became popular after broadband technology started spreading. Like DSL, blobspace is providing us a new frequency for rollups every single block! It's like the Ethereum mayor approved a project to build a 6 lane highway to help shorten your morning commute.
That's all for today's letter!
- YB
2 comments
Good read🫶🏻
👍🏻👍🏻