Exploring the Modularisation of Blockchains: An Introduction to Rollups

Ben Harvey
23 min readDec 3, 2021

On the 26th October I listened to a live Bankless podcast, hosted by David Hoffman and Ryan Sean Adams, on monolithic vs. modular blockchains. If neither of those terms mean anything to you, that’s absolutely fine, because the concept of modular blockchains struck such a nerve with me that I’ve decided to break it down over a few articles. I believe that modular blockchains represent the future of blockchain technology, and they’re crucial to understand should you wish to grasp the blockchain ecosystem in the future. The purpose of this blog is to allow you to learn with me, so I’ll be boiling all associated concepts down to basics, in order to paint a clear picture of what modular blockchains are, why they’re important, and how to invest in them.

There are two key concepts that require understanding before modular blockchains can be fully grasped: (1) rollups, and (2) shards. This article will focus on rollups; what they are, how they work, and where they’re headed in the future. An article in the not too distant future will focus on shards. I’ll then be completing this three part article series with an article tying everything together to explain modular blockchains and the future of the blockchain ecosystem.

The Blockchain Trilemma and the Birth of Layer 2

Inherently, blockchains face a trilemma, dubbed the blockchain trilemma. This refers to a trade-off faced by all blockchains, between decentralisation, scalability and security, in which a blockchain can only optimise for two, at the expense of the third. Clearly security is vital to a blockchain, so this generally leads to a compromise on either decentralisation or scalability.

The Blockchain Trilemma (Source)

Why is Decentralisation Important?

You’re probably wondering, why does decentralisation even matter? This is often a misunderstood concept, with the suggestion frequently arising that web3 advocates decentralisation to resist government censorship. This is in fact not the main reason decentralisation is important. The answer lies primarily in the incentivisation of innovation.

The issues with centralised platforms lie predominantly around their predictable life cycles. In the beginning, platforms are incredibly inviting, as they have to be to attract the users, developers, businesses and media organisations that will inevitably make their platform valuable. They lower barriers to entry and extract little value to create a positive-sum environment to encourage participants. As Chris Dixon states, centralised platforms do this to, “make their services more valuable, as platforms (by definition) are systems with multi-sided network effects.” The issues arise when platforms move up their S-curve, and gain more power over their users and developers.

Centralised Platform Life Cycle (Source)

Once centralised platforms reach the top of their S-curve life cycle, the nature of its relationship with participants of the platform diminishes from positive- to zero-sum. Generally speaking this means that the centralised platform will switch to extracting and monetising data from its users, and competing for attention with complementary businesses developing on its platform (for example, Twitter killing support for third-party-clients). This is bad news for innovation and the continued development of these platforms.

We’ve seen this life cycle play out over the past decade with multiple platforms; Facebook, Twitter, Microsoft, YouTube and Google to name a few. These are the very platforms that essentially own the internet today. Ultimately, this drives entrepreneurs, developers, and investors to be wary of these types of platforms, and many refuse to build in relation to centralised platforms altogether. Why would they, when the all powerful centralised platform has so much control over their creation? Why would they build with the risk of their creation being taken away? Think of content creators developing brands on YouTube, only to have their viewer figures halved by a recommendation algorithm out of their control, or developers building a game with a global community as a Facebook add-on, only for Facebook to remove their third-party-client API. On top of this, you have issues with centralisation from a user perspective, where users are vulnerable to security breaches, and give up both control of their data and privacy.

Decentralised platforms on the other hand, pose no censorship risk. Entrepreneurs and developers are free to build with a clear conscience. This has close parallels with the first era of the internet (1980s to 2000s) when internet services were built primarily on open-source protocols, owned by the internet community itself. By definition, decentralised platforms are unable to extract value from users and developers, rather they allow users and developers to generate value. Just to nail this point home, Google has associated itself with the motto “Don’t be evil,’ while in a decentralised platform, this shifts to quite literally, “Can’t be evil.”

This all sounds wonderful, and while I think we have a compelling case for decentralised platforms, do they really win, or even occur in reality? The first era of the internet was won by decentralised, open-sourced protocols. The second era was dominated by centralised applications, such as Google, Youtube, and Twitter. I believe the third era of the internet is likely to be decentralised. The simple reason for this: it’s where the builders are. Having been repelled from building on centralised platforms for reasons mentioned, entrepreneurs and developers are flooding to build on decentralised platforms such as Ethereum. What would the third era of the internet be without entrepreneurs and developers? Sure, you could argue that centralised platforms may be able to simply pay developers, but in my opinion they won’t be able to surpass those building on decentralised platforms, and they certainly won’t be able to match the rate of innovation of completely free entrepreneurs. This is particularly true when considering the tokenised nature of a decentralised third era of the internet, where entrepreneurs are able to raise capital at alarming rates, reducing the impact of centralised platforms simply paying their way to a developer community.

Blockchain technology is a breeding ground for the web3 ultimate platform. One in which there is no need for third party involvement, and inefficiencies are largely removed. If we’re to encourage entrepreneurs, developers and investors to build on these blockchain platforms, we must make them decentralised. For more detail on this loaded topic, I’d highly recommend reading this article.

So great, from an innovation, user privacy and data ownership perspective, it makes sense for blockchains to be decentralised. Unfortunately, this means foregoing scalability within the blockchain trilemma.

Components of a Blockchain

The way these blockchains have worked in the past has been that when someone posts a transaction, a miner verifies it and subsequently posts it on a blockchain, but then all of the other miners in the network have to verify it as well. Given in a decentralised network there can be large volumes of miners, this clearly represents a lot of inefficient and wasted work.

Decentralisation, security and scalability are however merely properties of a blockchain, dictated by the components which make up said blockchain. As a fundamental concept, it’s important to outline these components: (1) consensus, (2) data availability, and (3) execution. The consensus component dictates the collective verification of transactions by nodes in the network, providing security and truth about data stored on the blockchain. Data availability refers to the data that the settlement layer guarantees is available to be used for transacting on the blockchain (be that transfers, holding states etc.). Execution refers to the computation required to update the blockchain with the new information.

Simply put, the consensus component is where the record of transactions on the blockchain is agreed upon and stored. The data availability component is how much room the blockchain has at any one point in time to add additional data. The execution component is where new transactions are registered on the blockchain, prior to being verified and written into the permanent record.

As a solution to the blockchain trilemma, Layer 2 scaling solutions were developed. Layer 2 protocols can be thought of as scaling solutions for the Ethereum settlement layer, which allow the settlement layer to optimise for decentralisation and security. I’ll be expanding on the different type of scaling solutions and how they work later in this article. Layer 2 scaling solutions focus purely on the execution and (occasionally) the data availability components of the blockchain. The consensus component inherently must live with the settlement layer to leverage its decentralisation and security features.

To date, ~$6.62 billion worth of value has been locked in Layer 2 solutions, with leading solutions such as Arbitrum and dYdX (powered by StarkEx) attracting the lions share.

Total Value Locked Layer 2 Protocols (Source)

There are numerous types of scaling solutions, ranging from state channels, plasma, sidechains and rollups. For the purpose of this article, I will focus on rollups, due to the inherent limitations of state channels, plasma, sidechains and validium scaling solutions. In the context of the three components of the blockchain, rollups focus primarily on the execution layer.

Types of Rollups

Rollups are scaling solutions that bundle, compress, and send transactions to a consensus layer to be verified. By verifying multiple transactions at once, they dramatically increase efficiency, while allowing Ethereum to go from 15 to 3000+ transactions per second (TPS) without sacrificing security.

Essentially, instead of sending transactions to Layer 1 miners, users send their transactions to a rollup server, which verifies that these transactions are valid. So they’re performing transaction execution outside of the main Ethereum chain, but these rollups then post the transaction data to the main Ethereum chain. When posting the transaction to the main chain, they inherit the security properties of the layer 1.

There are two types of rollups, with different security models: (1) optimistic rollups and (2) zero-knowledge rollups.

Optimistic Rollups

Optimistic rollups bundle hundreds of transfers and publish only the bare minimum information needed on-chain with no proofs. This essentially assumes that no fraud or malicious behaviour has been committed, hence the ‘optimistic’ name. Optimistic rollups will only provide proof if a transfer or state is challenged.

The benefits of optimistic rollups, beside the obvious scaling enhancements, include EVM and solidity compatibility, meaning anything you can do on Ethereum layer 1, you can do on optimistic rollups.

However, there can be long wait times for on-chain transactions due to potential fraud challenges. You have to wait ~1-week to withdraw assets from optimistic rollups, since it gives time for fraud to be detected and resolved. This is an inherent issue with the assumption that no fraud or malicious behaviour has been committed.

Another downside of optimistic rollups is that there is a need to post all witness data, such as signatures, oracle feeds etc., on-chain. This limits scalability, especially when compared to zero-knowledge rollups, which I’ll discuss shortly.

Large optimistic rollup protocols include Arbitrum and Optimism. I’ll be exploring these protocols later in this article.

Zero-Knowledge Rollups

Zero-knowledge (ZK) rollups differ from optimistic rollups due to their integration of privacy by utilising zero-knowledge technology. Zero-knowledge refers to the ability to prove something (a transaction or state) to another party without disclosing the information necessary to prove it.

Instead of sending transactions to Layer 1 miners, users send transactions to a rollup server, where the transactions are executed. The two most common verification proofs used in ZK rollups are ZK-SNARKS and ZK-STARKS, I’ll touch on ZK-STARKS in a later section. ZK-SNARKS (Zero-Knowledge Succinct Non-Interactive Argument of Knowledge) are then generated as proofs for the validity of the transactions, and posted to the Layer 1 main chains. If you’re interested in exactly what ZK-SNARKs are, and how they’re formed, I’d highly recommend reading this short article. These proofs are fast to generate, they’re convincing, and they’re incredibly fast to verify. For context, hundreds of transactions can be rolled up into a single proof, and verifying that proof takes ~5 milliseconds.

By utilising the zero-knowledge feature of this technology, miners on the layer 1 will never know what the data in the transactions are. Only the rollup miners will be aware of the transaction data. This is a really exciting piece of technology, because it means data can be private, even on public blockchains. This would be crucial for companies like Apple, who may want to keep private how much they’re paying certain entities etc.

With ZK rollups, there are no delays when withdrawing assets, because a ZK-SNARK proof accepted by the ZK rollup contract has already verified the assets. This is one of the major feature differences between optimistic and ZK rollups.

The pros of ZK rollups include faster time to finality, as described above, as well as less vulnerability to economic attacks than optimistic rollups. However, ZK-SNARK proofs are intense to compute, and could be argued to be not worth it for applications with little on-chain activity.

I’ll be touching on four of the major ZK rollup protocols on mainnet to date later in the article, but for a full directory of ZK rollup protocols, check out this link.

It is worth noting here that currently the majority of ZK rollups don’t actually utilise the privacy aspect of ZK technology yet. Instead, they they focus on the scalability aspect, i.e. a proof can be exponentially smaller and leaner to verify than a native computation.

Technical Comparison Between Optimistic and ZK Rollups

Vitalik Buterin, co-founder of Ethereum has written extensively about rollup technologies, and the comparison of optimistic and ZK rollups. Below is a quote on Vitalik's outlook in terms of the two technologies, based on a technical comparison.

In general, my own view is that in the short term, optimistic rollups are likely to win out for general-purpose EVM computation and ZK rollups are likely to win out for simple payments, exchange and other application-specific use cases, but in the medium to long term ZK rollups will win out in all use cases as ZK-SNARK technology improves. — Vitalik Buterin (Source)

Below I've summarised the technical comparison between optimistic and ZK rollups.

Complex Tradeoffs Between Rollup Flavours (Source)

While optimistic rollups are currently more practical than ZK rollups, simply due to their lesser complexity, ZK rollups clearly hold the possibility to provide greater benefits to users. I will explore the future roadmap of each rollup type in the next section of this article.

Future of Rollups

While both rollup types currently provide benefits above the execution layer of the native Ethereum chain, there are some challenging aspects that require either improvement, or complete development.

Challenges Faced

EVM stands for Ethereum Virtual Machine, and can be thought of as a large decentralised computer that computes every type of task on the blockchain. It’s essentially the bedrock of Ethereum’s entire operating structure, and runs execution and smart contract deployment.

EVM compatibility means that another chain, for example, a rollup, would have the ability to deploy Ethereum smart contracts on said chain. So without EVM compatibility, it would be impossible for another chain to run an Ethereum smart contract. EVM compatibility is obviously hugely beneficial, because it means that projects already deployed on Ethereum, think DeFi giants such as Uniswap etc., can be deployed on the compatible chain. This means it’s relatively easy for the compatible chain to siphon off users at the execution module, should they have a significant competitive advantage over the Ethereum execution module. It’s also a huge advantage for the compatible chain, given the magnitude of developers within the Ethereum community. By tapping into that community through becoming EVM compatible, the compatible chains are able to benefit from the power of adoption and network effects of said developers, and the innovations that will be produced going forward.

The drawbacks of EVM compatibility stem from complexity, and the associated risk of security loopholes. Hence why EVM compatibility development has been slow for rollups in general, however some breakthroughs have been made, particularly in the context of optimistic rollups, which I’ll refer to in a bit.

A further challenge for rollups, and actually the one that I am most excited about seeing developments in going forward, is direct fiat on-ramps. Fiat refers to traditional currencies, such as the US Dollar, Great British Pound, and the Japanese Yen. On-ramp refers to a service in which you have the ability to directly convert fiat money in exchange for cryptocurrency. Fiat on-ramp services are currently typically found on centralised exchanges (CEXs), such as Coinbase and Binance.

To date, there have been several bottlenecks in the process of purchasing cryptocurrency, purely due to the difficulty of getting your fiat money from your traditional bank account, and into the world of crypto. Should you wish to interact with a decentralised exchange (DEX), it is common-practise to deposit fiat money into a CEX, and then have to exchange this fiat into crypto, only to transfer it across to your wallet of choice, for example MetaMask, and from there interact with your DEX of choice. This is clearly a bit of a hassle, and diminishes the user experience significantly. It’s worth noting that there is a project called Dharma, which is the only Ethereum wallet that can seamlessly move fiat money between your bank account and decentralized exchanges like Uniswap, on both Ethereum and Polygon, a sidechain with lower fees. However, this solution is only available with US banks, and is relatively new.

The challenge is heightened for rollups. In order to interact with crypto on rollups, you face the bottlenecks of getting crypto into your wallet of choice, and from there you must bridge your crypto assets across to the rollup itself. This is a seriously bad user experience, and requires a lot of patience. It is worth noting that Ramp, a fiat on-ramp solution, recently announced a partnership with zkSync, a Zk rollup. I’ll be touching on this later in the article.

I truly believe the mass adoption of direct fiat on-ramps, i.e. a path from your traditional bank account, straight onto a rollup, will dramatically increase usage. This is why I’m so excited for their development; the enhanced user experience will no doubt give users a reason to never directly interact with layer 1’s again, and will mark the mass migration from layer 1 execution, to layer 2 execution, leaving layer 1’s to focus on consensus and security. This will be the true beginning of modular blockchain adoption.

It’s also worth noting that for a rollup to reach its lowest fee rates, it needs to maximise transactions being settled on the Layer 1 chain. This is to evenly distribute the fees among the many transactions rolled together. So rollup fees can be thought of as a product of the Layer 1’s fees, but also a product of the demand on the rollup. Therefore, for rollups to truly reach their potential in terms of fee reduction versus Ethereum, they require higher levels of demand. This is commonly referred to as economies of scale, meaning the marginal transaction cost decreases for the next marginal transaction.

This almost creates a chicken and egg scenario, where users are required to lower fees dramatically, but may only switch across if fees are dramatically low. However, I don’t see this being a particular issue. In fact, it is a breakthrough in the sense that it inverts the traditional cost structure of a blockchain, which typically gets more expensive with each marginal transaction. This is also an instance in which ZK rollups outperform optimistic rollups, given optimistic rollups will always need to include signatures with each transaction, adding another cost factor. ZK rollups on the other hand don’t require signatures, and have better data compression, resulting in superior scalability potential.

Existing Optimistic Projects



The first optimistic rollup I’ll touch on is Arbitrum, an optimistic rollup protocol build by Offchain Labs. Arbitrum fully launched to Ethereum mainnet in August 2021, dubbing the mainnet network Arbitrum One. The full launch coincided with a funding raise of $120m for Offchain Labs, led by Lightspeed Venture Partners.

Arbitrum supports all EVM languages, and natively supports all Ethereum tooling without the need for any special adapters, thus making it fully EVM compatible. This includes the deployment of smart contracts and dApps. Under the hood however, this is all possible due to the Arbitrum Virtual Machine, although this is never exposed to developers or users.

To date, this EVM compatibility has enabled some of the largest dApps in the Ethereum ecosystem to deploy to Arbitrum, including DeFi giants Uniswap, SushiSwap, Aave, Curve and 1Inch. A full list of the ecosystem of wallets, dApps and more can be found at this link. The result of these deployments is over $2.65bn of total value locked on the network.

From a throughput perspective, Arbitrum One has a theoretical maximum TPS of 40,000, although in reality the maximum recorded TPS currently sits at around 49. The cost benefits to using Arbitrum One compared with Ethereum mainnet depend greatly on the transfer type, for example, for an ordinary transfer the fees on Arbitrum One are ~36% that of Ethereum Mainnet, whereas for a Uniswap v3 transaction they’re ~6.9% in comparison.



Optimism is the second optimistic rollup I’ll touch on today, which launched to mainnet in July 2021. Since launching, Optimism has accumulated $466m in total value locked in the network, processing over 2.2m transactions.

According to Optimism, their optimistic rollup solution has helped save over $100m in gas fees when transacting versus the Ethereum mainnet, stating you can transact in milliseconds, and save 10–100x on fees. The network itself is home to over 100,000 unique wallet addresses (users).

In November 2021, Optimism went live with EVM equivalence on Optimistic Ethereum, replacing the previously used Optimistic Virtual Machine solution. EVM equivalence is complete compliance with the Ethereum stack; every debugger, toolchain, node implementation, and of course, allows for the deployment of smart contracts and dApps.

This EVM equivalence has enabled the deployment of many key players in the Ethereum ecosystem to the Optimism network. These key players include, but are not limited to, Uniswap, Synthetix and 1Inch.

It is worth noting that currently Optimism is running with no fraud proofs. In other words, this means that the optimistic rollup technology aspect is not yet turned on.

Existing ZK Projects



The first ZK rollup project I’ll touch on is StarkNet, an Ethereum mainnet ZK rollup by StarkWare. StarkNet Alpha (an unaudited version of StarkNet) launched on mainnet on the 29th November 2021, having been deployed to testnet, a self-explanatory instance of a blockchain powered by a newer version of the underlying software, for the purpose of testing and debugging, in June 2020.

StarkNets predecessor StarkEx launched on mainnet in June 2020, and offered support for general smart contracts. Applications deployed on StarkEx are dYdX, ImmutableX, DiversiFi and Sorare. To date, they have processed 65m transactions, with a cumulative trading volume of $215bn across the four protocols. The StarkEx network also demonstrated a rate of 9,000 TPS for trades, and 18,000 TPS for transfers, which equates to approximately a 600x increase in scalability for trades from Ethereum mainnet execution layer.

StarkNet followed the path of its impressive predecessor, by featuring general smart contracts on a fully composable network. Composability refers to the ability of applications to coordinate, build on top of one another, and interconnect. This composability is supported both with other StarkNet contracts and with Layer 1 contracts via L1<>L2 messaging. StarkNet will also be permissionless, meaning it enables dApp to leverage StarkNets unlimited scale for its computation, without the need to request permission to deploy on the network. However, for the time being StarkNet Alpha is available only to whitelisted dApps.

One of the greatest benefits that StarkEx and StarkNet have over competing ZK rollups is their ZK-STARK (Zero-Knowledge Scalable Transparent Argument of Knowledge) technology. Both ZK-SNARKs and ZK-STARKS are privacy enhancing technologies, since they reduce the amount of information required to be communicated between users. They’re also both scaling solutions, since they allow proofs to be verified at a faster rate than the native Ethereum execution module because they don’t contain the full amount of information for non-private systems. However, while ZK-SNARKS require a trusted setup phase, ZK-STARKs use publicly verifiable randomness to create trustlessly verifiable computation systems. The result is that ZK-STARKs are more scalable in terms of computational speed and size when compared to ZK-SNARKs.

On top of this, ZK-STARKS have soundness against a computationally unbounded provers, due to strong cryptographic primitives like has functions. This means there is zero chance a verifier can be convinced of a false ZK-STARK. The result: ZK-STARKs are post quantum because their security isn’t reliant on prover computation power. For more detail, check out the tweet thread below, explaining the comparison between ZK-SNARKS and ZK-STARKS in this context.

StarkNet however, is not directly EVM compatible. Instead, they have developed a STARK optimised programming language called Cairo. Cairo is used for writing provable programs on blockchain, enabling developers to use proof technology. A software solution team called Nethermind have subsequently developed a solution to transpile Solidity (the EVM language) into Cairo, and so there is indirect EVM compatibility. Both STARK and Cairo were developed in-house by StarkWare and have powered all their production-grade applications.

As you can see from the StarkNet roadmap below, the final step is to take StarkNet to a decentralised governance model. StarkNet currently isn’t represented by a token, meaning there is no liquid way for everyday investors to invest in the project and therefore technology. However, it is likely that a decentralised governance model will introduce an associated token, which may be used to speculate on the technology. Having said this, StarkNet core devs repeatedly refuse to answer the question ‘wen token’, so this is pure speculation.

StarkNet Roadmap (Source)



The second ZK rollup project I’ll take a look at it zkSync, a mainnet ZK rollup by Matter Labs. Matter Labs received a $50m investment from a16z a few months ago to help them bring ZK rollup scalability to Ethereum. zkSync was launched to mainnet in June 2020, and was built primarily for scalable payments. This means users can deposit onto the network and transfer between other zkSync accounts at a fraction of the cost of on Ethereum.

To date, zkSync has processed 4m transactions, with a total value locked in the network of $24m. In July 2020, the network added NFT functionality with zkNFT.

Matter Lab’s iteration of zkSync, zkSync 2.0 is scheduled to launch on mainnet, and aims to feature fully composable smart contracts on a ZK rollup. zkSync 2.0 will introduce zkEVM, a virtual machine that simulates an environment like Ethereum, thus making it EVM compatible and enabling Ethereum smart contract deployment. zkEVM testnet has been successfully introduced with UniSync, a fork of Uniswap V2, to demonstrate its functionality. Since its launch in October, UniSync has handled over 3,000,000 transactions on its testnet.

One huge development with zkSync has been the announcement of a partnership with Ramp, to provide a fiat on-ramp directly onto the zkSync protocol. This is a solution to one of the major drawbacks of rollups in general, and no doubt will drive a significant volume of adoption to the zkSync network.

The zkSync roadmap below is similar to that of StarkNet. We’ve already seen the first phase with the introduction of zkSync to mainnet. zkSync 2.0 will complete the second phase when deployed to mainnet. The final phase, censorship resistance, will ultimately require the platform to decentralise and become permissionless, which likely means the introduction of a governance token. zkPorter, a proof-of-stake chain which will provide off-chain data availability for zkSync 2.0 is expected to have its own token.




Loopring is a ZK rollup protocol that allows for building highly scalable exchanges and payment solutions, which allows high through-put, low cost trading and payment on Ethereum. Loopring 3.0, their first ZK rollup iteration was launched to mainnet in December 2019, and was the first zero-knowledge rollup DEX protocol to deploy to the Ethereum mainnet. The latest version of Loopring, v3.6 was deployed to mainnet in December 2020.

To date, Loopring has over $650m in total locked value in the protocol. According to Loopring themselves, their ZK Rollup throughput reaches approximately 1000x of Ethereum for trades, or as high as 2,025 trades per second, which is a significant enhancement on the user experience of the native Ethereum execution module. Loopring utilises ZK-SNARKS to achieve this, much like zkSync.

Loopring is the only ZK rollup protocol with their own smart wallet that’s available to download as an app on both iPhone and Android. It allows you to store your crypto assets on their rollup, and also interact with the functionality of their ZK rollup, with trading and payments. However, there is currently a fee of ~$300 to set up a Loopring native wallet. This is expected to be replaced by the release of a completely free Loopring wallet in the not too distant future. In August 2021, Loopring announced that it now supports NFT minting, trading, and transfers, directly on its ZK rollup protocol.

Loopring has a native token, $LRC, which is available on many exchanges, including Coinbase and Binance. The token was released during a three phased airdrop throughout 2018, to users early adopters of Loopring v1. $LRC acts as a governance token to the Loopring protocol, making it completely decentralised. The performance of $LRC YTD can be seen below.

$LRC Performance in USD YTD (Source)

In terms of roadmap for Looping, their team have outlined a few things to look out for in Q4 2021. Given we have merely less than a month left of this timeframe, we should expect to see some of these features being added soon. The first of which will be the development of an NFT marketplace on Loopring. Distributed Automatic Market Makers have also been teased, in collaboration with StarkWare.

Polygon Hermez


In August 2021 Ethereum scaling protocol Polygon announced the acquisition of and merger with Hermez. Polygon is a protocol and framework for building and connecting Ethereum-compatible blockchain networks, with the goal of aggregating scalable solutions on Ethereum, thus supporting a multi-chain Ethereum ecosystem. They’re essentially supporting the modularisation of the Ethereum ecosystem, by developing capabilities for the execution and data availability modules.

Hermez itself is part of the Polygon ecosystem, and is an open-source ZK rollup optimised for secure, low-cost token transfers, secured by the Ethereum network. By utilising zero-knowledge technology, Hermez state that they’re able to increase throughput by 133x relative to Ethereum mainnet, while gaining over 90% a reduction in token transfer costs.

Below is an intuitive diagram produced by Hermez, which outlines the general rollup concept we’ve discussed in this article, while explaining how 2,000 transactions are processed in a single Hermez network batch.


Hermez is a completely decentralised network, much like StarkNet and zkSync plan to be in their roadmaps. This means it has a native token, $HEZ, which is used for governance, amongst other things within the network. The $HEZ token is available on DEXs including Uniswap v2. The performance of $HEZ YTD can be seen below. However, given Hermez has been acquired by Polygon, which has an existing native token, $MATIC, it is unclear what will happen to $HEZ going forward.

$HEZ Performance in USD YTD (Source)

Upcoming Rollup Catalysts


An important mention in this article (and if you’ve made it this far I’m assuming you’re incredibly interested in understanding ZK rollups!) is EIP-4488 . For those of you unfamiliar with EIP’s, they stand for Ethereum Improvement Proposal, and are standards which specify potential new features or processes for Ethereum. Anyone within the Ethereum community has the ability to create an EIP, though they must be sufficiently agreed upon to be implemented.

To summarise, EIP-4488 reduces the calldata cost from 16 to 3 gas per byte, with a cap on calldata per block to mitigate security risks. Essentially, this means that rollup overhead costs decrease, which for the end user of a rollup means fees are even lower. For a more detailed dive into EIP-4488, check out the below thread on Twitter by proto.eth.


Another important catalyst for rollups will be sharding. Sharing is a key concept to understand in the context of modular blockchains, and is essentially a solution to the data availability constraints of Ethereum. I’ll be releasing an article that focuses purely on sharding as the second article in this three part series on the modularisation of blockchains. However, if you can’t wait until then to learn about sharing, I’d recommend checking out this article.

Closing Remarks

Rollups are no doubt crucial to the future of blockchains, and will arguably be the networks in which everyday users of dApps will interact with blockchain technology. It’s possible these users won’t even know which rollup they’re using, much like the internet where we often don’t know which cloud-hosting service a website utilises.

It’s this reason that I’m extremely excited by rollups. We’re still very early in their lifecycle, hence the complexity of the technology and the missing features in some cases. This means we’ll be able to watch as they evolve, and may even be able to speculate on various rollup technologies over the coming months and years.

If you’d like to learn more, or keep up to date with the technical progression of rollups, and their adoption, I’d advise following these people on Twitter:

And of course, I’ll be delivering content regularly here on Medium, and on my Twitter account @0xbenharvey. Give me a follow or subscribe to this newsletter to keep up to date. Thanks, and see you in two weeks. WAGMI.



Ben Harvey

I’m a student of the blockchain and broader Web 3.0 ecosystem, documenting my learning on this blog in the hope that you find some value in it too.