A Deep Dive into Rollups, Their Current Centralized Nature, and the Ongoing Efforts Towards Decentralization.

High Rollers:

Welcome to a deep dive into the world of rollups – a technology that’s reshaping Ethereum as we know it. Today, rollups are more centralized than you might think, a trade-off made for immediate feedback loops, accessibility, and user acquisition in the race to scale Ethereum.

Whether you’re familiar with Arbitrum, OptimismFND, or new to rollups, this post is for you.

Discover the tech behind rollups, the quest for L2 decentralization, and the teams leading the charge. Plus, explore the rapid expansion of L2s in the onchain dark forest.

Let’s unravel the mysteries of rollups together????

Cheers,

The Rollup

---

???????????????????????????? ???????? ???????? ???????????????? ???????????????? ???????????????????? ???????????? ???????????????? ???????????????????????????????????????????? ???????????????? ???????????????????????? ???????????? ????????????????????????????…

In the quest to scale Ethereum with rollups, decentralization has been put to the backburner in exchange for immediate feedback loops, accessibility and user acquisition.

For better or for worse, the rollup stack today looks far different than the ideal version in the future.

If you’re reading this, you may have heard of rollups before, used @arbitrum or @optimismFND, or perhaps others…but, if you haven’t?

Don’t worry.

In today’s post, you’ll learn about the underlying technology behind rollups, the search for decentralization in the L2 sphere, and several key teams pioneering these initiatives

As we all know, the onchain dark forest is witnessing a rapid expansion of L2s to address the scalability issues of Ethereum.

In the diagram below from @l2beat, you can see there are already 15+ highly active rollup ecosystems with tons of users, transaction activity, and TVL.

I think by EOY, we’ll be at 100+ rollups with thousands of daily active users.

By the end of 2024?

???????????????? ???????????? ????????????????????????????????. Some highly knowledgable people in the space, like @neelsalami, have put out pieces arguing against this.

In fact, tomorrow I’ll have a piece summarizing Neel’s recent post & also post some counterpoints to his theses.

Gib follow @ayyyeandy to not miss that one tomorrow, then keep reading…

Amidst this Cambrian Explosion of rollups, as quoted by @0xkatz of @Calderaxyz in the most recent DeFi By Design Podcast on @therollupco set to go live this week, a key component of rollups, known as sequencers, has sparked a debate about centralization and decentralization within the ecosystem.

???? ???????????????????????????? ????????????????????????, ???????? ????????????????. First, what are sequencers and why are they important?

Sequencers play a crucial role in the functioning of rollups. They are responsible for ordering transactions within rollups, making the network cheaper, faster, and more user-friendly.

As you can see below, sequencers work on the ordering of TXs and batching into one TX to be posted to Ethereum L1.

???????????????????????????? ????????????????????????????????????????, ???????????????????????????? ???????????????????? ???????????????? ???????? ???????????????? ???????? ???????????????????????????????? ???????????? ???????????????????????????????????????? ????????????????????????????????????????????????, ???????????????????? ???????????????????? ???????? ???????????????? ???????????????????????????????????? ???????????? ???????????????? ????????????????????????????????????.

However, the use of centralized sequencers is not without its downsides:

‣ ????????????????????????????????????????:
Theoretically, sequencer operators could choose which TXs to include in blocks, and which to exclude. This is not ideal for immutability or trustlessness, two core ethos’ of crypto.

‣ ???????????????????????? ???????????????????????? ???????? ????????????????????????????:
With just one operator in a hardware database somewhere around the world, there is little to no backup. If unforeseen events happen, for example Arbitrum / Offchain labs forgot to add gas money to their sequencer and it stopped processing TXs for sometime (???????? ???????? ????????????????????, ???????????????? ???????????? ???????????? ????????????????????), then the chain can stop operating effectively.

‣ ???????????????????????????? ???????????????????????????????????????????? ???????????????????? (????????????) ????????????????????????????????????????:
Due to the fact that sequencers are responsible for TX ordering, preferential treatment is not out of the picture. Sandwich attacks, or frontrunning become real possibilities due to the centralized control of TX ordering.

‣ ???????????????? ????????????????????????????????????????????????????????????????:
With just one sequencer, its tough to interoperate with other rollups in a cost-effective fashion. You have to eat all the costs for sequencing and likely data availability as well. Less interop = less liquidity = less users.

These are four major topics to address.

???????????????????????? we enter the solutions segment, I want to highlight that I am still learning about sequencers myself and I am very open to feedback.

????????????????, according to other people much smarter than me, there are much more important challenges facing rollups than sequencers at the moment.

????????????????????????????, ???????????????????? ???????? ???????????????????? ???? ???????????????????????????? ???????????????? ???????????? ???????????????? ???????????????????????????????????????????????????? ????????????????????????????????????…

???????????? ???????????????????????????? ???????????????????????????????????????? ???????????????? ????????????????????????????: ???????????????????????? ???????????????????????????????????????? ???????????? ???????????????????????? ???????????????????????????????????????????????????? ????????????????????????????????????????.

Shared sequencers refer to a network of sequencers that service multiple rollups. They sort out transactions and create a mega block from which individual rollups derive their blocks.

This approach promotes interoperability and composability between rollups, as they can all utilize the shared sequencing layer.

(???????????????????? ???????????????? @CelestiaOrg, reminder: make sure to claim your $TIA before Oct. 17th!)

One of the prominent shared sequencer networks going live very soon is@EspressoSys. Espresso is aiming to achieve lower costs and greater decentralizations for teams which choose to utilize their network of shared sequencers.

On the other hand, direct decentralized sequencing involves each rollup having its own set of sequencers. This approach allows rollups to design mechanisms that cater specifically to their needs. It also makes it easier for network participants to work together, as they operate within their own ecosystems.

Both shared and direct decentralized sequencing have their merits and drawbacks.

Shared sequencing can save rollups the hassle of setting up their own sequencing node army and promote better liquidity across rollups.

However, it is still limited by the L1’s data and transaction ordering throughput. Direct decentralized sequencing allows for more customization and control, but it can be harder to bootstrap and may isolate rollups from each other.

As you can see the diagram below, the tradeoffs between trustlessness and functionality is highlighted.

Shared sequencers appear to be the outlier…

???????????? ???????????????????????? ???????????????????????????? ???????????????????????? ???????????? ???????????????????????? ???????????????????????????????????????????????????? ???????????????????????????????????????? ???????????????? ???????????????????????? ???????? ???????????? ???????????????????????????????? ???????????????????? ???????? ???????????????? ????????????????????????.

Rollups with complex features may prefer direct decentralized sequencers, while lightweight rollups may opt for shared sequencing. As the rollup ecosystem continues to evolve, it is clear that the future will see a mix of both approaches.

???????????? ???????????????????????????????????? ???????????????? ???????? ???????? ???????????????????????? ???? ???????????????????????????? ???????????????????????????? ????????????????????????????????????????, ????????????????-????????????????????????????????????????????????, ???????????? ????????????????????????????????????????????????????????????????.

The good news?

Solutions exist already: