The message arrived at 3:47 AM Cape Town time—a developer I mentor had spotted something troubling in the transaction ordering logs of a prominent zkEVM rollup. Blocks were being finalized not by the distributed validator set the project had marketed to investors and users, but by a single infrastructure endpoint operated by the founding team. When he flagged this to community moderators, the response was swift: silence. The documentation that once described "decentralized sequencing" had quietly been revised overnight. The code still worked. The users were still transacting. But the promise had quietly been broken.
This is not an isolated incident. Over the past eighteen months, as Layer2 ecosystems have matured from whiteboard concepts to multi-billion-dollar mainnet deployments, a pattern has emerged that should concern anyone who entered this space seeking genuine decentralization. The sequencing layer—the component responsible for ordering transactions, a function that carries profound implications for MEV extraction, censorship resistance, and economic fairness—remains concentrated in single points of control across nearly every production rollup currently operating.
The Gap Between Marketing and Reality
The sequencing problem is not a secret among protocol engineers. If you attend the right working groups, the conversations are remarkably candid. Projects that raised hundreds of millions on the thesis of "Ethereum scaling without sacrificing decentralization" are now running centralized sequencers because the engineering reality of distributed transaction ordering remains extraordinarily difficult. The technical challenges are real—achieving consensus on transaction order across geographically distributed nodes while maintaining the sub-second finality users expect from modern applications is genuinely hard. But the difficulty of a problem does not excuse misrepresenting whether you have solved it.
Consider the data. A comprehensive audit I commissioned for a mid-tier rollup last quarter revealed that 94% of transaction ordering authority rested with three co-located server instances operated by the protocol's core team. When I raised this with the project leads, their response was instructive: "We are working toward decentralization, and the current architecture is a necessary stepping stone." This framing has become the industry's standard deflection. But let me be direct about what this actually means in practice. A "stepping stone" that concentrates ordering power for years while the team collects sequencer fees is not a technical necessity—it is a business choice dressed in engineering language.
The implications extend far beyond abstract architectural concerns. When a single entity controls transaction ordering, they control the sandwich attack surface. They determine which MEV opportunities get captured and who captures them. They can, under regulatory pressure, selectively reorder or censor transactions in ways that users cannot detect or resist. The promise of Ethereum's base layer—immutable, uncensorable, neutral infrastructure—dissolves when the rollups built on top of it delegate these properties to trusted intermediaries.
The Economic Logic Driving Centralization
To understand how we arrived here, you need to follow the money. Decentralized sequencing requires coordinating a validator network, distributing sequencing rights, and building infrastructure for permissionless participation. Each of these introduces latency, cost, and complexity. A centralized sequencer operated by the founding team is fast, cheap, and controllable. The economics are straightforward: build centralized first, generate revenue, then theoretically decentralize when "the protocol matures." The problem is that protocols rarely voluntarily surrender control once they have tasted the economic benefits of centralized sequencing. The fees collected from transaction ordering are substantial—on a busy rollup, sequencer revenue can exceed several million dollars monthly. Why would any team voluntarily dilute that revenue stream by distributing sequencing rights to a permissionless validator set?
The honest answer is that most will not, unless forced. This is where I believe the Ethereum community's governance mechanisms have failed to evolve with the economic reality. The original vision for optimistic rollups, articulated in publications I read during MakerDAO's early days, included specific timelines for decentralized sequencing as a governance milestone. Those timelines have come and gone. The language has shifted from "decentralized by launch" to "decentralized over time" to, in the most egregious cases, removing the commitment entirely from public roadmaps while maintaining the marketing impression.
I want to be careful here not to paint all projects with the same brush. There are genuine technical efforts underway. Projects like Espresso Systems and Astria are building shared sequencing layers that could provide true decentralization across multiple rollups. The Ethereum Foundation's roadmap includes sequencing decentralization as a priority. But these solutions remain in testnet or early mainnet phases while production systems continue operating on centralized infrastructure.
The Regulatory Overlay Changes the Calculus
Here is what keeps me awake at night. We are not merely discussing a technical shortcoming—we are discussing it against a backdrop of accelerating regulatory pressure on blockchain infrastructure. The same centralized sequencers that represent a decentralization failure are also single points of regulatory vulnerability. When the SEC, CFTC, or international regulatory bodies demand compliance from rollup operators, they will not negotiate with a distributed validator set. They will locate the entity controlling the sequencer and apply pressure there.
I have seen this pattern before. During the ICO era, projects marketed "decentralized governance" while running token distributions from single AWS regions. When regulatory attention intensified, those centralized points became the exact vectors through which enforcement actions were delivered. The promise of decentralization provided legal cover during the permissive period, while the technical reality remained centralized. I fear we are watching this movie repeat at a larger scale with rollups.
The implications for users in emerging markets are particularly stark. When I ran SoulBound's educational cooperatives, the participants who most needed censorship-resistant financial infrastructure were often those living under governments with the most aggressive capital controls. A centralized sequencer can be compelled to censor their transactions. The promise of blockchain becomes hollow when the ordering layer remains controllable.
What Genuine Progress Would Look Like
I do not write this to condemn the Layer2 ecosystem. The scaling achievements of the past three years are real and valuable. Ethereum's L2 landscape has delivered meaningful throughput improvements that have benefited millions of users. The engineering teams building these systems are, by and large, working with integrity toward important goals. But integrity requires acknowledging when the gap between promise and delivery has grown too wide.
So what would actual progress on sequencing decentralization require? First, binding commitments with enforcement mechanisms—not roadmap language, but on-chain commitments that trigger protocol-level consequences if decentralization milestones are missed. Second, genuine permissionless participation in sequencing, meaning anyone can operate a sequencer node without requiring team approval. Third, transparent real-time dashboards showing sequencing authority distribution, so the community can hold projects accountable to their commitments.
Some will argue that demanding perfect decentralization now would sacrifice the user experience necessary for mainstream adoption. I understand this argument, but I reject its premise. Users in emerging markets, activists under authoritarian governments, and financial participants who need genuine censorship resistance are not fringe use cases—they are the populations who benefit most from what blockchain technology promises. Trading their protection for a smoother UX for degens in San Francisco is not a acceptable trade-off.
The Question We Must Answer Together
The sequencing problem is ultimately a governance problem—a collective choice about what properties we believe decentralized infrastructure should guarantee and how we will enforce those guarantees. Code is law, but ethics is conscience. The code exists to serve human flourishing, not the other way around.
I am not naive about the difficulty of building distributed systems at scale. The teams working on this problem are dealing with genuinely hard computer science. But the difficulty of achieving decentralization does not excuse its abandonment. Solidarity over speculation means building systems that serve everyone, not just those who happen to live in regulatory jurisdictions that allow them to trust centralized intermediaries.
The next eighteen months will determine whether Layer2 sequencing decentralization becomes a genuine roadmap priority or remains a marketing talking point. The projects that will earn lasting trust are those willing to make binding commitments, accept short-term friction for long-term integrity, and remember that the users who most need this technology are often those with the least recourse when its promises are broken.
The sequencer runs. The blocks finalize. The fees flow. But the question remains: who controls the order of things, and who decides? That question deserves an honest answer, not another revision to the documentation.