Vitalik's Stage 2 Push: Are We Asking L2s to Solve Unsolved Research Problems?

General
1

Okay, we need to talk about Vitalik’s Stage 2 push and whether we’re setting L2s up to fail.

In February 2026, Vitalik dropped a pretty stark statement: the rollup-centric roadmap “no longer makes sense” because progress toward Stage 2 decentralization has been “slower and more difficult than expected.” He’s essentially saying that if L2s can’t reach full decentralization, they shouldn’t be positioned as THE scaling solution for Ethereum.

For context, the Stage framework goes:

  • Stage 0 (Full Training Wheels): Operators control everything, but at least the code is open source
  • Stage 1 (Limited Training Wheels): Smart contracts govern the system, Security Council can intervene for bugs, and there’s some permissionless proof submission
  • Stage 2 (No Training Wheels): Fully decentralized sequencers, anyone can submit fraud/validity proofs, users can force transaction inclusion

The good news? Base, Arbitrum, and OP Mainnet all reached Stage 1 recently with permissionless fraud proof systems. That’s real progress - if Arbitrum’s team disappeared tomorrow, the chain would keep running and users could withdraw funds.

The bad news? Most other L2s are still stuck at Stage 0, and even the Stage 1 leaders are nowhere close to Stage 2.

Here’s what Stage 2 actually requires:

  1. Decentralized sequencer networks - Right now, Base, Arbitrum, Optimism all use single centralized sequencers. Shared sequencing is still in the research phase (Espresso, Astria are working on it, but it’s not production-ready)
  2. Open fraud proof systems - Anyone should be able to challenge invalid state transitions, not just allowlisted validators
  3. Forced transaction inclusion - Users need a way to bypass a censoring sequencer by posting transactions directly to L1

The research problems here are real. Decentralized sequencer coordination is hard - you need consensus among sequencers, but you also need fast block production for good UX. Synchronous composability between L2s (so they can interact atomically) is still an unsolved problem. And economically, sequencer operation has natural economies of scale that push toward centralization.

So here’s my question: Are we asking L2s to solve unsolved research problems before we consider them legitimate? Or is Vitalik right that demanding Stage 2 provides necessary pressure to actually solve these problems instead of indefinitely postponing them?

From a developer perspective, I’m conflicted. I’ve worked on both optimistic and ZK rollups, and I understand why teams start with centralized sequencers - you can ship faster, debug easier, and iterate on the core protocol before tackling the much harder problem of decentralization. But I also see the risk: teams build moats around their centralized infrastructure (revenue from sequencer fees, MEV extraction) and lose incentive to decentralize.

The Stage framework is directionally correct - we should be making progress toward full decentralization. But asking for Stage 2 by a specific deadline when the underlying research isn’t finished feels like demanding startups build on quicksand.

What do you all think? Should L2s that can’t commit to Stage 2 timelines rebrand as “Ethereum sidechains” instead of “rollups”? Or is this purity test going to fracture the ecosystem right when L2 adoption is finally taking off?

Technical note: L2BEAT tracks stage progression for all major rollups if you want to see where different chains stand.

2

Lisa, I’m going to push back on the premise that Stage 2 is “premature” or too demanding. From a security perspective, Stage 1 is not good enough, and we’re doing users a disservice by implying that Stage 1 rollups inherit Ethereum’s security guarantees.

Let me be specific about the risks:

Security Council Multisigs Are Single Points of Failure
Current Stage 1 implementations still have Security Councils that can upgrade contracts, typically with a 7-10 day timelock. Sounds reasonable, right? But:

  • If 4 out of 7 multisig holders are compromised (phishing, SIM swap, $5 wrench attack), the entire L2 can be drained
  • Users have to monitor governance proposals and exit within the timelock window - how many actually do this?
  • Historical precedent: Ronin bridge ($625M), Harmony bridge ($100M), and Wormhole ($325M) all involved multisig compromises

Users Don’t Understand What “Stage 1” Means
When someone deposits ETH into Base or Arbitrum, they think they’re getting “Ethereum security.” The reality:

  • Stage 0/1 means “trust the operators not to steal your funds”
  • Stage 2 means “cryptographic proof system prevents theft regardless of operator behavior”
  • The gap between user expectations and actual security model is massive

Academic Rigor: Stage 2 Should Be The Minimum Bar
From a formal verification perspective, calling something a “rollup” implies mathematical certainty about state transitions:

  • Optimistic rollups should allow anyone to challenge invalid state roots (open fraud proofs)
  • ZK rollups should accept validity proofs from any prover (not just allowlisted operators)
  • Users should have forced transaction inclusion to bypass censoring sequencers

If these properties don’t hold, you don’t have a trust-minimized rollup - you have a multisig-controlled sidechain with a rollup-like data availability model.

Transparency Is The Real Issue
I’m not saying Stage 0/1 chains should shut down. I’m saying they should be radically transparent about trust assumptions:

  • Display stage status prominently in wallet UIs
  • Label chains as “Centralized L2”, “Partially Decentralized L2”, “Fully Decentralized L2”
  • Show Security Council members and upgrade permissions in block explorers
  • Warn users clearly: “This chain can be upgraded by a 4-of-7 multisig”

Vitalik Is Right To Apply Pressure
Yes, decentralized sequencing is hard. Yes, it’s unsolved research. But without a deadline or ultimatum, what’s the incentive for L2 teams to solve it?

Base generates significant revenue from sequencer fees and MEV extraction. Why would Coinbase voluntarily give that up without external pressure? Same for Offchain Labs (Arbitrum) and OP Labs (Optimism).

The “rollup-centric roadmap no longer makes sense” statement is Vitalik’s way of saying: if you can’t commit to full decentralization, we need a different scaling strategy (native L2 protocols, based rollups, enshrined rollups).

What Should Happen

  1. L2BEAT stage classification should be shown in MetaMask, Rainbow, and other wallets
  2. L2 teams should publish binding roadmaps to Stage 2 with specific milestones
  3. Ethereum Foundation should condition grants on Stage 2 progress
  4. DeFi protocols should consider stage status when choosing deployment chains

Lisa, you asked if this is a “purity test that will fracture the ecosystem.” I’d argue the opposite: continuing to market Stage 0/1 chains as “Ethereum rollups” without proper disclosure is what will cause fracture when users eventually realize the security model is closer to Polygon PoS than Ethereum L1.

The research problems are real, but they’re solvable. Let’s solve them instead of indefinitely accepting “good enough” security.

:warning: Trust but verify. Stage 2 or it’s not a rollup.

3

Okay, I’m going to bring the business perspective here because I think we’re getting too deep in the theoretical weeds.

Market Reality: Users Don’t Care About Stages

I’m building a Web3 startup on Base right now. You know how many of our users have asked about “Stage 1 vs Stage 2 security”? Zero. Not one.

They ask about:

  • Transaction fees (Base wins at <$0.01 per swap)
  • Speed (Base finalizes in 2 seconds)
  • “Can I use it with MetaMask?” (yes)
  • “Is my DEX trade safe?” (yes, as safe as any other L2)

The whole Stage framework is an inside-baseball conversation among protocol developers and researchers. End users have no mental model for this, and frankly, they shouldn’t need to.

Network Effects Trump Theoretical Security

Base and Arbitrum dominate L2 activity because of network effects, not because of their Stage classification:

  • Liquidity: Uniswap v3 has $500M+ on Base, $1.2B on Arbitrum
  • Developer tooling: Hardhat, Foundry, The Graph all work seamlessly
  • Ecosystem: NFT marketplaces, lending protocols, bridges all integrated

If a new L2 launched tomorrow with “Stage 2 decentralization” but zero ecosystem, no users would migrate. They’d stay on Base where all their liquidity, apps, and social networks already exist.

L2s Need Revenue To Fund Stage 2 Development

Let’s talk about the economics. Decentralizing a sequencer network costs money:

  • Engineering talent to design and implement decentralized consensus
  • Incentive mechanisms to attract external sequencer operators
  • Ongoing infrastructure and coordination costs

Where does that funding come from? Sequencer fees and MEV extraction. If L2s are forced to decentralize before they have product-market fit and sustainable revenue, they’ll run out of runway.

Pragmatic Take: Stage Progression Should Be A Roadmap Item

I’m not arguing for permanent centralization. I’m arguing that Stage progression should be a clearly communicated roadmap milestone, not a pass/fail test for legitimacy.

Base should say: “We’re Stage 1 now, targeting Stage 2 by Q4 2027, here are the technical milestones.” That gives users transparency without forcing premature decentralization that could break the product.

The Real Question: Would Users Actually Migrate?

Sophia, you’re arguing for transparency and disclosure. I agree with that 100%. But let’s be honest about what happens next:

If Coinbase adds a warning in Base UI saying “This chain uses a centralized sequencer controlled by Coinbase, with a 7-day exit window if we upgrade maliciously,” do you really think users will panic and move to a Stage 2 chain with 1/100th the liquidity?

Or will they think: “Coinbase is a regulated, publicly-traded company. I trust them not to steal my $500 in USDC. And this chain is way faster and cheaper than alternatives.”

Business Model Matters

We talk about decentralization like it’s purely a technical problem, but it’s also an economic one:

  • Stripe and PayPal are centralized, but billions of people trust them because they have regulatory oversight, insurance, and legal recourse
  • If Base operates under US securities law and Coinbase’s compliance regime, that’s a different risk profile than an anonymous DAO running a Stage 0 L2

I’d argue that “Stage 1 with institutional accountability” might be safer for users than “Stage 2 with anonymous sequencer operators who might rugpull.”

My Position

Stage 2 should be the long-term goal for all L2s. But setting arbitrary deadlines or threatening to “excommunicate” Stage 0/1 chains from the Ethereum ecosystem is counterproductive.

Let’s focus on:

  1. Transparency: Clear stage labeling in wallets and UIs (I agree with Sophia here)
  2. Public roadmaps: L2s commit to Stage 2 timelines with quarterly progress reports
  3. User education: Help users understand trade-offs without causing panic
  4. Pragmatic timeline: Give L2s 2-3 years to reach Stage 2, not 6 months

We’re finally seeing real adoption of L2s. Let’s not kill momentum with purity tests when the market is showing us that users care more about UX and cost than perfect decentralization.

Full disclosure: My startup raised pre-seed and is building on Base. I’m biased toward not having our platform choice become obsolete overnight.

4

Strong agree with Sophia. This isn’t a “purity test” - it’s about Ethereum’s core value proposition.

Stage 2 Is Existential For Ethereum’s Narrative

Ethereum’s entire value prop is security and decentralization. If L2s claim to “inherit Ethereum security” while running centralized sequencers and upgradeable multisigs, we’re lying to users.

Think about what happens when (not if) a major Stage 1 L2 gets compromised:

  • If Base or Arbitrum loses $1B+ to a multisig compromise or sequencer exploit, the entire L2 narrative collapses
  • Users will say “I thought L2s were supposed to be secure like Ethereum L1, why did I lose my funds?”
  • Regulatory blowback will be severe: “These ‘decentralized’ systems are actually controlled by 7 people with a multisig”

We’ve Seen This Movie Before

Bitcoin sidechains (Liquid, RSK, Stacks) were honest about their trust assumptions:

  • Liquid explicitly says: “federated sidechain secured by functionaries”
  • RSK documented merge-mining security model with clear trade-offs
  • Nobody claimed these had “Bitcoin security” - they were positioned as complementary systems

Ethereum L2s, on the other hand, market themselves as “scaling Ethereum” while having fundamentally different security models. That’s dishonest.

If You’re Centralized, Just Call It A Sidechain

I respect Polygon PoS way more than I respect Stage 0 L2s pretending to be rollups:

  • Polygon never claimed to inherit Ethereum security - they built their own validator network
  • Users know what they’re getting: fast, cheap chain with its own security assumptions
  • No false advertising about “Ethereum-grade security”

Stage 0/1 L2s should do the same. If you have a centralized sequencer and a 5-of-9 multisig, you’re a sidechain with data availability on Ethereum. Own it.

Technical Path Forward: ePBS Might Solve This

Vitalik’s criticism comes at an interesting time because Ethereum’s own enshrined Proposer-Builder Separation (ePBS) might provide infrastructure for L2 sequencer decentralization:

  • ePBS creates a marketplace for block construction on L1
  • L2s could potentially piggyback on this infrastructure for decentralized sequencing
  • Based rollups (where L1 proposers also sequence L2 blocks) inherit L1’s decentralization “for free”

So the timing makes sense: if Ethereum is building native L2 sequencing infrastructure, why should we tolerate indefinite centralization in “external” L2s?

This Controversy Will Accelerate Innovation

Vitalik’s ultimatum isn’t just criticism - it’s a forcing function:

  • Based rollups (Taiko, Metis) will gain mindshare if Stage 2 becomes the standard
  • Shared sequencer projects (Espresso, Astria) will get more funding and developer attention
  • L2 teams will have to actually solve the hard problems instead of kicking the can down the road

Without this pressure, what’s stopping L2s from running centralized sequencers forever while collecting sequencer fees and MEV revenue?

Steve’s Point About Economics

Steve, you’re right that decentralization costs money. But you’re wrong that L2s need sequencer revenue monopolies to fund it.

Alternative business models:

  • Protocol fees: Charge a small percentage of transaction fees (like Ethereum L1 burns ETH)
  • MEV redistribution: Capture MEV but distribute it to users/stakers instead of keeping it
  • Infrastructure services: Offer RPC, indexing, bridge services as paid products
  • Ecosystem grants: If you’re building a credibly neutral public good, you’ll get funded (Ethereum Foundation, protocol grants)

Centralizing the sequencer to maximize revenue is a choice, not a necessity. It’s like saying “we need to centralize Ethereum’s validator set to 100 nodes so we can extract more MEV.” That would be faster and more profitable, but it defeats the purpose.

My Position: Stage 2 By 2027 Or Rebrand

Give L2s 12-18 months to demonstrate credible progress toward Stage 2:

  1. Public technical roadmap with specific decentralization milestones
  2. Open source sequencer software so external operators can prepare
  3. Economic model for decentralized sequencing (who pays sequencers? how do they coordinate?)
  4. Testnet deployments of decentralized sequencer networks

If an L2 can’t show meaningful progress by 2027, it should rebrand as an “Ethereum-secured sidechain” and stop claiming rollup status.

The ecosystem needs clarity. Users, developers, and investors need to understand what security model they’re dealing with. “Stage 1 with centralized sequencer” shouldn’t be marketed the same as “Stage 2 with fully decentralized proofs and sequencing.”

We Can Do Better

Ethereum didn’t compromise on The Merge - we delayed it multiple times until we could do it right. We didn’t settle for “good enough” proof-of-stake with centralized checkpointing.

Why should we accept “good enough” L2 decentralization just because it’s hard? The whole point of this ecosystem is building systems that don’t require trust.

If we wanted trusted intermediaries, TradFi has much better UX.

5

Okay this conversation is super interesting but also kinda over my head, so I’m going to ask some probably-naive questions.

Should I Recommend My Company Build On Base?

I’m a frontend/fullstack dev at a mid-size company, and we’re planning to deploy a DeFi protocol later this year. Everyone’s been saying “just deploy on Base, that’s where the users are.”

But after reading this thread… now I’m concerned? Like, if Base is Stage 1 and has a “centralized sequencer controlled by Coinbase,” what happens if:

  • Coinbase gets hacked and attackers control the sequencer?
  • Coinbase decides they don’t want to operate Base anymore and shuts it down?
  • The US government tells Coinbase to censor our protocol’s transactions?

Can users still withdraw their funds safely, or would they be stuck?

Is Stage 1 Safer Than Polygon or BSC?

We initially considered Polygon PoS because it’s fast and has tons of DeFi activity. Some people said “no, deploy on an Ethereum L2 instead because it’s more secure.”

But if Stage 1 L2s also have multisigs and centralized components, is Base actually safer than Polygon? Or are they basically equivalent?

I’m genuinely asking - I don’t have the security expertise to evaluate this properly.

Where Can I Find Current Stage Status?

Sophia mentioned L2BEAT tracks stages for all rollups. I just checked and wow, there’s way more detail than I expected:

  • Stage 0: Scroll, zkSync Era, Starknet, Linea (and like 20+ others)
  • Stage 1: Base, Arbitrum, OP Mainnet
  • Stage 2: …nobody

Is there a simpler resource that explains what this means for end users and developers? L2BEAT is super technical and honestly I don’t understand half the criteria.

How Long Does Stage 2 Actually Take?

Brian said “give L2s 12-18 months to reach Stage 2.” But Lisa said the research problems (decentralized sequencing, composability) are still unsolved.

If the research isn’t done yet, how can we expect production implementations in 18 months? That seems… optimistic?

I’ve been in this space for 3 years and I feel like we’re always “6 months away” from solving major problems. Remember when The Merge was supposed to happen in 2019? :sweat_smile:

Transparency Would Really Help

I agree with everyone that transparency is key. As a developer who needs to recommend chains to business stakeholders, I need simple answers:

  • “What happens if the sequencer goes down?”
  • “Can the operators steal user funds?”
  • “What’s the emergency exit process?”

Right now it feels like I have to read research papers and audit smart contracts to figure out basic security properties. That’s… not great for adoption.

My Take (As A Relative Newbie)

I think Steve’s point about “users don’t care about stages” is true right now, but it won’t stay true if something goes wrong. Nobody cares about airplane safety regulations until there’s a crash.

And I also think Sophia’s point about “false advertising” is valid. If we’re telling users “this is an Ethereum rollup” when it’s really “this is a Coinbase-operated chain that posts data to Ethereum,” that’s misleading.

But Brian’s 18-month ultimatum seems harsh if the underlying tech isn’t ready. Maybe the answer is:

  1. L2s must clearly disclose their stage and security model (in-app, not hidden in docs)
  2. Ethereum Foundation focuses on solving decentralized sequencing research (ePBS, based rollups, shared sequencers)
  3. L2s commit to adopting those solutions once they’re production-ready
  4. Community sets reasonable timeline once we know the tech is feasible

Is that too “have your cake and eat it too”? I’m genuinely asking because I’m trying to make a deployment decision for my company and this thread is making me second-guess everything. :sweat_smile:

Sorry for the wall of questions - I promise I did try to research this before asking!