Data Availability Sampling is a major breakthrough in blockchain scaling, but many struggle to understand it.

Don't worry, it's actually pretty simple. You can think of it like tossing a coin 🧵👇 

Imagine I have two coins, one that always lands on heads and one that has lands 50/50 on heads or tails.

I give you one of the two coins. How do you figure out which coin I gave you?

Simple. You just flip the coin multiple times.

If the coin is heads-only, it will always land heads.

If the coin is 50/50 then with each toss, it only has a 50% chance of landing heads. 

So each time you toss the coin and it lands heads, you gain 50% more confidence that it's the heads-only coin.

Repeat the process 20 times and you will have 99.9999% confidence. 

This is how Data Availability Sampling works.

There are two types of blocks: data available &amp; data unavailable blocks (heads-only and 50/50 coins).

The block producer gives you a block and - for your security - you need to know which it is!

How do you do it? By sampling that block multiple times (tossing the coin).

With each successful sample (the coin lands heads) you gain 50% more confidence that you have the available block.

You repeat this until you have 99.99% confidence the block is available. 

Why is this such a big breakthrough in blockchain scalability?

Well bigger blocks means more throughput, however it also means more data for nodes to download. 

Typically to make sure a block is available, you have to download the entire thing.

Data Availability Sampling allows you to make sure that a block is available by only downloading a few tiny samples of it 

If you have a 4MB block and you need 20 samples of 1kB each, then you only need to download ~.5% of the total block to be 99.9999% sure it's available.

That's a 200x improvement.

And it only becomes more efficient as blocks get bigger.

This is one of the core technologies we're building @CelestiaOrg and a fundamental piece of the modular blockchain stack.

To learn more I recommend checking out our Learn Modular page https://celestia.org/learn

&amp; stay tuned for more explainer threads from me 🤝

How Data Availability Sampling Works

The shared sequencers' paradox:

💰 Extract MEV but be less attractive for rollups to use
🙅‍♂️ *OR* don't extract MEV but lose value capture

Wat do? 🧵

A shared sequencer that wants to remain attractive to build on and capture value, should extract *some* MEV, but not all.

❌ Intra-rollup MEV
✅ Inter-rollup MEV

Why is this a better approach?

Rollups will want to capture their chain's intra-rollup MEV, a concept called "Sovereign MEV" pioneered by @SkipProtocol

This feature is so valuable that if a shared sequencer takes it away, rollups may opt to do their own sequencing

Inter-rollup MEV is a different story.

Rollups on their own don't have much means to capture cross-domain MEV

So they wouldn't be giving up much if the shared sequencer captures their inter-rollup MEV 🤷‍♂️

Plus rollups benefit from lower overhead & better interoperability provided by a shared sequencer

So they should be willing to pay something for those benefits, e.g. give up some MEV

The shared sequencers' paradox isn't so bad after all.

A shared sequencer that strictly extracts inter-rollup MEV can be attractive to use AND capture some juicy value 😋

But they better leave that Sovereign MEV alone...

Just some shower thoughts.

Wondering how the teams building these systems like @AstriaOrg and @EspressoSys are thinking about this

Shared Sequencing MEV

Is a rollup on @ethereum more secure than a rollup on @CelestiaOrg?

Yes and no.

Here are the tradeoffs and why security isn't what you think 🧵

To understand why rollups on Ethereum aren't necessarily more secure than on Celestia, we need to define "security".

There are two aspects of security:
⛓️ Immutability
✔️ Verifiability

Let's unpack them

⛓️ Immutability is how hard it is to re-org a chain.

The more expensive it is to re-org the chain, the more secure the chain is against a double spend or similar attack.

@nic__carter wrote a great post on this topic: https://medium.com/@nic__carter/its-the-settlement-assurances-stupid-5dcd1c3f4e41


✔️ Verifiability is how easy it is for users to verify that the blocks in the chain are valid.

The cheaper & simpler the chain is to verify, the more secure it is against block producers breaking the rules, e.g. printing money.

@hasufl explains this here 👇

https://twitter.com/hasufl/status/1395742194766602241?s=20

In my opinion, verifiability is the more important of the two.

If the rules of the chain can be broken, the downside of an attack is virtually unlimited.

A double spend is still bad, but the downside is tiny in comparison.

Now, back to Ethereum and Celestia security.

Let's look at immutability first.

In PoS protocols, immutability increases with the amount of value at stake.

If a chain has more $$ staked, then more stake gets burned if the block producers re-org.

With $38B of ETH staked, Ethereum will be significantly more costly to re-org than Celestia when it launches.

So on the dimension of immutability, Ethereum will be more secure than Celestia, hands down.

What about verifiability?

If you need $$$ for a powerful computer and fast internet connection to verify the chain, verifiability is low.

If you can verify the chain on a smart phone like @CelestiaOrg, verifiability is extremely high.

https://twitter.com/musalbas/status/1480901457633239048?s=20

Celestia requires significantly fewer resources to verify than Ethereum.

Ethereum requirements:
• 4 cores
• 16GB of RAM
• 1 TB SSD
• 25 Mbps bandwidth

Celestia requirements:
• 1 core (1/4th)
• 2 GB of RAM (1/8th)
• 5 GB SSD (1/200th)
• 56 Kbps bandwidth (1/500th)

So on the dimension of verifiability, Celestia is more secure than Ethereum.

With Danksharding Ethereum may become cheaper to verify, but users may still have to run a normal execution full node in order to verify the DA chain.

Anyone know the answer to this?

To be clear, I'm not saying one is better than the other. I'm pointing out the tradeoffs.

Modularism not maximalism means that developers have the option to choose which tradeoff is best for their use case.

Devs should build whatever they want.

So is a rollup on Ethereum more secure than on Celestia?

It depends.

If you want immutability, Ethereum is better.

If you want verifiability, Celestia is better.

Immutability and Verifiability

There are two *very* different definitions of the term “modular blockchain” being conflated on CT right now

1. Modular blockchain software 
2. Modular blockchain protocols
 
Both are important for the future of the space, but people should not mix them up 🧵👇

First, where did the term “modular blockchain” come from??

The term was first coined by @musalbas in this blogpost announcing the @CelestiaOrg project in 2019

Referring to separating data availability and execution to build modular blockchain protocols
https://twitter.com/epolynya/status/1449954691622727683?s=20

So what are modular blockchain protocols?

They are a new blockchain architecture where the key functions of consensus, data availability and execution are split into separate layers within the stack

This is in contrast to monolithic chains which do everything on their own

Since 2019, the idea for modular blockchain protocols has grown in popularity due to better scalability, security &amp; decentralization over monolithic chains (and thanks to @epolynya's writing)

Projects like Ethereum are now pursuing a modular roadmap
https://ethereum-magicians.org/t/a-rollup-centric-ethereum-roadmap/4698

As the term grew in popularity, it started to be used interchangeably to describe a different kind of modularity - modular blockchain software

Modular blockchain software breaks apart the blockchain software stack into different components, namely consensus and execution

Examples of this software design pattern first emerged in the @cosmos ecosystem with Tendermint (consensus software) and the Cosmos SDK (execution software)

But it’s starting to appear everywhere - such as the Ethereum &amp; Avalanche ecosystems

https://blog.ethereum.org/2022/01/24/the-great-eth2-renaming/ 

A modular blockchain software stack makes it much easier to design and build a new chain

In fact, @celestiaorg is built using one - Tendermint and the Cosmos SDK (can highly recommend!)

However, a blockchain built using modular software is *not* necessarily a modular blockchain protocol.

How is this possible?

Simple, a blockchain built using modular software might still do consensus, data availability &amp; execution all on its own, making it a *monolithic protocol*

Examples of this would be all current cosmos zones like the Hub or Osmosis, as well as the Avalanche C-chain

By comparison, Celestia is a modular protocol bc it only does consensus &amp; DA and is designed for execution layers to plug into it

Rollups like Starkware or Arbitrum are also modular protocols bc they only do execution and rely on other chains for consensus, DA and settlement

So, when someone uses the term “modular blockchain” be aware that this could mean two very different things.

One decomposes the blockchain *software* stack into modular pieces

The other decomposes the blockchain *protocol* stack into separate specialized modular layers

One lets you build software for a new chain easily.

The other enables you to deploy a new blockchain without bootstrapping a new consensus network (among other benefits)

BOTH are important for the future development of the space. /fin

Inspiration for this thread came from this discussion w/ @el33th4xor 
https://twitter.com/nickwh8te/status/1485456468530933762?s=20&t=Y_rUH4KJSbPzwaVWuhqpeA

Modular Blockchain vs Modular Software

Let’s clear this up once and for all

Alchemy, the Graph and all other blockchain indexers are NOT data availability services

Neither are file storage protocols like Filecoin and Arweave 👇🧵 https://twitter.com/cdixon/status/1485303906154467330 

Data availability is best thought of as protection against *data withholding*, a nasty attack by a block producer

To prevent this attack you need to make sure that no data in a block has been hidden/withheld

Specifically you want to know that no data was withheld *without* having to download all the data yourself

The only solution to this problem is Data Availability Sampling (DAS) and only @CelestiaOrg, @0xPolygon Avail and Ethereum are building this atm

So now back to indexers and storage protocols

Does an indexer help me know that all the data in that block was available without downloading it all?

No it just organizes and serves that data. I’d still have to download it all to verify the availability

Does a storage network help me know the block data has not been withheld?

No, a storage network just ensures that some data will not disappear, not that the data was there in the first place.

Again if I wanted to make sure the data is there I’d have to download it all myself

Data availability is a very nuanced concept and often misunderstood.

That’s likely due to the fact that the name is confusing and vague

I hope this helps clear up the misconceptions