Celestia is a project that revises the architecture of classical blockchains and proposes a more flexible approach where blockchain is returned to its sole purpose: order data and make it available.
Its modular architecture, where the execution is dissociated from Layer 1, allows it to evolve in an ecosystem of scalable and interoperable blockchains where developers are no longer limited by their execution environment.
Key concepts
Full nodes: they download and verify the validity of the transactions entered. In other words, they verify that a miner hasn’t entered an invalid transaction or circumvented the protocol rules. These nodes are the most secure despite the number of resources required.
Light clients: require fewer resources because they only download the block headers assuming that the block body contains only valid transactions.
Only the block headers are downloaded because they contain the Merkel root, which is a kind of concise summary of all transactions.
Their low resource demand makes them less secure.
The current blockchains have a so-called monolithic architecture**,** the three main functions of a blockchain being grouped on the same layer:
Consensus: how the nodes in the network agree on which transactions take place and in what order.
Execution: the processing of these transactions (determining which ones are valid and their impact on the state of the network).
Data availability: how transaction data are made available to nodes for download.
Having execution and consensus on the same layer means that the full nodes must, in addition to checking whether there has been a consensus for the validity of a certain block, ensure that the transactions are valid by executing them.
However, this architecture is a real bottleneck because as the number of transactions increases the full nodes will have large blocks to download and verify.
Few people will have the resources to set up such nodes and this will naturally lead to centralization.
Mustafa offers his vision of things in 2019: to build a lazy Layer 1 (the project was formerly called "Lazy Ledger") that focuses only on consensus and data availability by decoupling execution.
This change in architecture will modify the role of nodes within Layer 1: since transaction execution is performed on Layer 2, full nodes no longer focus on the content of transactions (the content of a transaction being its associated data).
Their role is to simply order this data and make it available to Layer 2.
Light clients as we know them today suffer from a lack of security: they assume that the chain privileged by the consensus algorithm contains only valid blocks and that most block producers are honest.
Celestia's premise is to increase the security of light clients while keeping their resource demands low.
And the idea of increasing the security of light clients goes back to Satoshi's paper. He mentioned that light clients could be more secure if full nodes sent them "alerts" when an invalid block was published.
If you have read my article on rollups, you know that fraud proofs are used in Optimistic rollups and can be submitted by any user on Layer 2.
But this operation is specific to Optimistic rollups in the Ethereum ecosystem.
With Celestia, Optimistic rollups can be implemented as applications with client-side execution, where fraud proofs are distributed via a peer-to-peer network instead of being published in a smart contract.
The full nodes will then notify the light clients if there is an invalid transaction in the chain by sending a fraud proof. The light client will verify this proof and thus be able to reject the transaction.
As Mustafa shows in this tweet light clients of such a scheme require so few resources that they can be run with a smartphone.
https://twitter.com/musalbas/status/1480901457633239048?s=20&t=8uhLnMQx7FVcsQ3igjA2ZA))
The problem with the above scheme is that the block producer can make a data withholding attack: they will "deliver" the block, but it will only contain its header.
The transaction data will voluntarily not be communicated (fraud proof cannot be constructed without this data).
One can legitimately wonder how to detect the presence of an invalid transaction if the block producer omits to share certain data?
The most obvious solution would be to force all nodes (including light clients) to download the entire blockchain, but this would prevent scalability in the long run.
The other solution is to allow light clients to check that the block is fully "available", but without needing to download the entire block.
Once the block data is considered available, full nodes will be able to verify it and build their fraud proofs.
Mustafa and Vitalik assemble their knowledge and present their solutions in a paper in 2019, which we will popularize below.
Erasure Coding is a technology initially used with CD-ROMs or satellites.
It allows reading a CD despite it being partly scratched or recovering the elements of communication if this one stopped suddenly between the satellite and the earth.
Applied to blockchain, we will use erasure coding to create new types of blocks allowing the reconstitution of data even if a certain part is missing (in the case where a fraudulent miner hides some data).
The principle is as follows: let's take a 1 Mo block. We will add an extra 1 Mo of erasure codes. The block is now 2 Mo in size, where 1 Mo is the "real data" and extra Mo is the erasure codes.
Erasure coding can recover up to 50% of a missing block.
This means that only 50% of a block must be published by the block producer for it to be considered available.
A fraudulent miner wanting to retain 1% of the block will therefore be obliged to retain at least 50% of the block since this 1% can be reconstituted from the 50% of erasure codes.
Based on this principle, light clients will sample the block: they will randomly download small pieces of the block repeatedly to ensure that no part of the block has been retained.
On the first try, they have a 50% chance of picking from the unavailable part of the block. We will know that this part is unavailable.
But if this action is repeated 16 times, light clients have a 99.99% chance of knowing that the block is unavailable and will be able to reject it.
Data sampling combined with erasure coding greatly enhances the role of light clients within Layer 1.
The first advantage of architecture like Celestia's is practicality. A plug-and-play ecosystem could then be put in place, where blockchains will no longer have to build a consensus layer and deal with data availability (which is a long and expensive process).
They will be able to delegate this part by using Celestia as Layer 1 and focus on their execution environment.
The second advantage is the gain in sovereignty. Blockchains with a modular architecture give more freedom to developers because they can choose their preferred execution environment, and thus no longer must deploy smart contracts limited by the restrictions of the EVM.
We also find this problem in DAOs: a blockchain that depends on the execution of Ethereum also depends on the choices of its community regarding decision making.
First of all, the narrative on monolithic blockchains is becoming less and less current.
Many projects moving Ethereum execution to Layer 2 are in development and the ecosystems are growing fast.
What is the future of Celestia in a world where transaction execution on Ethereum is fully operational on Layer 2? This thought also applies to blockchains whose value proposition relies solely on Ethereum congestion.
Despite exponential fees, Ethereum continues to dominate most of the market and give birth to the biggest blockchain use cases (DeFi, NFT…) overshadowing other Layers 1 trying to reappropriate these "hypes".
What will happen when Layer 2s have decongested Ethereum?
I think that wanting to compete with Ethereum can be a good value proposition for a while (as Solana and Avalanche have demonstrated with significant upside) but ends up falling in the long run.
There are many examples: the rate of transactions on the Binance Smart Chain, the exponential increase in gas costs of Avalanche or Fantom when blockchains are used intensively, the repeated technical problems at Solana, etc.
Second, Vitalik had already talked about data availability issues at a conference in 2017, which he posted on his GitHub, two years before the paper with Mustapha.
He is therefore aware of this problem and knows the solutions.
I think that Celestia has more to gain by offering itself as a data availability solution for Ethereum but using the Ethereum consensus layer.
But does Ethereum need it?
And if so, would Ethereum prefer to outsource this task to Celestia rather than directly implement data availability solutions at its native Layer 1?
Third, the data sampling performed randomly by the light clients is dependent on the number of participants in the network. If Celestia's value proposition does not convince the community, there will not be enough light clients to perform the block sampling.
However, since light clients require few resources, the network can be easily expanded.
Mustafa Al-Bassam (CEO): passionate about programming since childhood, he will have fun hacking the servers of his school.
He will make different attacks within the Anonymous collective and his own collective LulZec by hacking the Tunisian prime minister, the American government, or the FBI (he is only 15-16 years old).
Mustafa began his interest in blockchain in 2015-2016 developing on Ethereum and began a Ph.D. on Layer 1 scalability in 2016.
He wrote one of the first articles on sharding.
Then he will co-found Chainspace, a company focused on sharding bought by Facebook in 2019 to be integrated into the Diem project (ex-Libra).
However, he will not join the Facebook team and will start Celestia under the name “Lazy Ledger”.
John Adler (CRO): creator of Optimistic rollups, former scalability researcher at Consensys.
Ismail Khoffi (CTO): previously a Tendermint core engineer.
Nick White (COO): co-founder of Harmony.
Despite no official information, Nick White had given some details on @bigdsenpai's Twitch channel.
The token will a priori be used to secure the network via the Proof of Stake (staking & delegation) and to pay transaction fees.
There will likely be an inflationary reward system.
Like the gwei used to pay gas fees on Ethereum, the Celestia token can be used for displaying data on the blockchain, as well as making it available.
This mechanism is interesting because the more a large number of blockchains will be built on Celestia, the more their blockchain will be requested to provide consensus and prove the availability of data, which will naturally request the token.
A token burn mechanism similar to EIP-1559 could also be implemented.
As a modular blockchain, Celestia paves the way for a future ecosystem of interoperable blockchains where tasks are skillfully shared.
The data availability solutions are innovative, and the team is recognized in the blockchain community.
Like any project, Celestia's success will depend on the enthusiasm of its community, its marketing strength, and the attraction of developers to its technology.
Celestia is a project that revises the architecture of classical blockchains and proposes a more flexible approach where blockchain is returned to its sole purpose: order data and make it available.
Its modular architecture, where the execution is dissociated from Layer 1, allows it to evolve in an ecosystem of scalable and interoperable blockchains where developers are no longer limited by their execution environment.
Key concepts
Full nodes: they download and verify the validity of the transactions entered. In other words, they verify that a miner hasn’t entered an invalid transaction or circumvented the protocol rules. These nodes are the most secure despite the number of resources required.
Light clients: require fewer resources because they only download the block headers assuming that the block body contains only valid transactions.
Only the block headers are downloaded because they contain the Merkel root, which is a kind of concise summary of all transactions.
Their low resource demand makes them less secure.
The current blockchains have a so-called monolithic architecture**,** the three main functions of a blockchain being grouped on the same layer:
Consensus: how the nodes in the network agree on which transactions take place and in what order.
Execution: the processing of these transactions (determining which ones are valid and their impact on the state of the network).
Data availability: how transaction data are made available to nodes for download.
Having execution and consensus on the same layer means that the full nodes must, in addition to checking whether there has been a consensus for the validity of a certain block, ensure that the transactions are valid by executing them.
However, this architecture is a real bottleneck because as the number of transactions increases the full nodes will have large blocks to download and verify.
Few people will have the resources to set up such nodes and this will naturally lead to centralization.
Mustafa offers his vision of things in 2019: to build a lazy Layer 1 (the project was formerly called "Lazy Ledger") that focuses only on consensus and data availability by decoupling execution.
This change in architecture will modify the role of nodes within Layer 1: since transaction execution is performed on Layer 2, full nodes no longer focus on the content of transactions (the content of a transaction being its associated data).
Their role is to simply order this data and make it available to Layer 2.
Light clients as we know them today suffer from a lack of security: they assume that the chain privileged by the consensus algorithm contains only valid blocks and that most block producers are honest.
Celestia's premise is to increase the security of light clients while keeping their resource demands low.
And the idea of increasing the security of light clients goes back to Satoshi's paper. He mentioned that light clients could be more secure if full nodes sent them "alerts" when an invalid block was published.
If you have read my article on rollups, you know that fraud proofs are used in Optimistic rollups and can be submitted by any user on Layer 2.
But this operation is specific to Optimistic rollups in the Ethereum ecosystem.
With Celestia, Optimistic rollups can be implemented as applications with client-side execution, where fraud proofs are distributed via a peer-to-peer network instead of being published in a smart contract.
The full nodes will then notify the light clients if there is an invalid transaction in the chain by sending a fraud proof. The light client will verify this proof and thus be able to reject the transaction.
As Mustafa shows in this tweet light clients of such a scheme require so few resources that they can be run with a smartphone.
https://twitter.com/musalbas/status/1480901457633239048?s=20&t=8uhLnMQx7FVcsQ3igjA2ZA))
The problem with the above scheme is that the block producer can make a data withholding attack: they will "deliver" the block, but it will only contain its header.
The transaction data will voluntarily not be communicated (fraud proof cannot be constructed without this data).
One can legitimately wonder how to detect the presence of an invalid transaction if the block producer omits to share certain data?
The most obvious solution would be to force all nodes (including light clients) to download the entire blockchain, but this would prevent scalability in the long run.
The other solution is to allow light clients to check that the block is fully "available", but without needing to download the entire block.
Once the block data is considered available, full nodes will be able to verify it and build their fraud proofs.
Mustafa and Vitalik assemble their knowledge and present their solutions in a paper in 2019, which we will popularize below.
Erasure Coding is a technology initially used with CD-ROMs or satellites.
It allows reading a CD despite it being partly scratched or recovering the elements of communication if this one stopped suddenly between the satellite and the earth.
Applied to blockchain, we will use erasure coding to create new types of blocks allowing the reconstitution of data even if a certain part is missing (in the case where a fraudulent miner hides some data).
The principle is as follows: let's take a 1 Mo block. We will add an extra 1 Mo of erasure codes. The block is now 2 Mo in size, where 1 Mo is the "real data" and extra Mo is the erasure codes.
Erasure coding can recover up to 50% of a missing block.
This means that only 50% of a block must be published by the block producer for it to be considered available.
A fraudulent miner wanting to retain 1% of the block will therefore be obliged to retain at least 50% of the block since this 1% can be reconstituted from the 50% of erasure codes.
Based on this principle, light clients will sample the block: they will randomly download small pieces of the block repeatedly to ensure that no part of the block has been retained.
On the first try, they have a 50% chance of picking from the unavailable part of the block. We will know that this part is unavailable.
But if this action is repeated 16 times, light clients have a 99.99% chance of knowing that the block is unavailable and will be able to reject it.
Data sampling combined with erasure coding greatly enhances the role of light clients within Layer 1.
The first advantage of architecture like Celestia's is practicality. A plug-and-play ecosystem could then be put in place, where blockchains will no longer have to build a consensus layer and deal with data availability (which is a long and expensive process).
They will be able to delegate this part by using Celestia as Layer 1 and focus on their execution environment.
The second advantage is the gain in sovereignty. Blockchains with a modular architecture give more freedom to developers because they can choose their preferred execution environment, and thus no longer must deploy smart contracts limited by the restrictions of the EVM.
We also find this problem in DAOs: a blockchain that depends on the execution of Ethereum also depends on the choices of its community regarding decision making.
First of all, the narrative on monolithic blockchains is becoming less and less current.
Many projects moving Ethereum execution to Layer 2 are in development and the ecosystems are growing fast.
What is the future of Celestia in a world where transaction execution on Ethereum is fully operational on Layer 2? This thought also applies to blockchains whose value proposition relies solely on Ethereum congestion.
Despite exponential fees, Ethereum continues to dominate most of the market and give birth to the biggest blockchain use cases (DeFi, NFT…) overshadowing other Layers 1 trying to reappropriate these "hypes".
What will happen when Layer 2s have decongested Ethereum?
I think that wanting to compete with Ethereum can be a good value proposition for a while (as Solana and Avalanche have demonstrated with significant upside) but ends up falling in the long run.
There are many examples: the rate of transactions on the Binance Smart Chain, the exponential increase in gas costs of Avalanche or Fantom when blockchains are used intensively, the repeated technical problems at Solana, etc.
Second, Vitalik had already talked about data availability issues at a conference in 2017, which he posted on his GitHub, two years before the paper with Mustapha.
He is therefore aware of this problem and knows the solutions.
I think that Celestia has more to gain by offering itself as a data availability solution for Ethereum but using the Ethereum consensus layer.
But does Ethereum need it?
And if so, would Ethereum prefer to outsource this task to Celestia rather than directly implement data availability solutions at its native Layer 1?
Third, the data sampling performed randomly by the light clients is dependent on the number of participants in the network. If Celestia's value proposition does not convince the community, there will not be enough light clients to perform the block sampling.
However, since light clients require few resources, the network can be easily expanded.
Mustafa Al-Bassam (CEO): passionate about programming since childhood, he will have fun hacking the servers of his school.
He will make different attacks within the Anonymous collective and his own collective LulZec by hacking the Tunisian prime minister, the American government, or the FBI (he is only 15-16 years old).
Mustafa began his interest in blockchain in 2015-2016 developing on Ethereum and began a Ph.D. on Layer 1 scalability in 2016.
He wrote one of the first articles on sharding.
Then he will co-found Chainspace, a company focused on sharding bought by Facebook in 2019 to be integrated into the Diem project (ex-Libra).
However, he will not join the Facebook team and will start Celestia under the name “Lazy Ledger”.
John Adler (CRO): creator of Optimistic rollups, former scalability researcher at Consensys.
Ismail Khoffi (CTO): previously a Tendermint core engineer.
Nick White (COO): co-founder of Harmony.
Despite no official information, Nick White had given some details on @bigdsenpai's Twitch channel.
The token will a priori be used to secure the network via the Proof of Stake (staking & delegation) and to pay transaction fees.
There will likely be an inflationary reward system.
Like the gwei used to pay gas fees on Ethereum, the Celestia token can be used for displaying data on the blockchain, as well as making it available.
This mechanism is interesting because the more a large number of blockchains will be built on Celestia, the more their blockchain will be requested to provide consensus and prove the availability of data, which will naturally request the token.
A token burn mechanism similar to EIP-1559 could also be implemented.
As a modular blockchain, Celestia paves the way for a future ecosystem of interoperable blockchains where tasks are skillfully shared.
The data availability solutions are innovative, and the team is recognized in the blockchain community.
Like any project, Celestia's success will depend on the enthusiasm of its community, its marketing strength, and the attraction of developers to its technology.
My articles will focus on projects analyzes and technological advances in the crypto ecosystem. Your feedbacks are welcomed !
My articles will focus on projects analyzes and technological advances in the crypto ecosystem. Your feedbacks are welcomed !
Share Dialog
Share Dialog
Subscribe to Lucas
Subscribe to Lucas
<100 subscribers
<100 subscribers
No activity yet