Client implementation dominance is the primary lever. The protocol is defined by its execution and consensus clients, like Geth and Nethermind. A supermajority of validators running a single client creates systemic risk, as seen in past Geth bugs, making client teams de facto gatekeepers.
the-ethereum-roadmap-merge-surge-verge
Blog
Who Decides Ethereum Protocol Changes
Ethereum's governance is a messy, multi-stakeholder bazaar, not a boardroom. This analysis breaks down the real power dynamics between core devs, client teams, stakers, and the community in shaping The Merge, The Surge, and beyond.
introduction
THE POWER MAP
The Illusion of Decentralized Control
Ethereum's governance is a multi-layered power structure where client diversity and capital concentration dictate protocol evolution.
Capital controls the fork choice. While EIPs are debated openly, the final adoption signal comes from validator staking weight. Entities like Lido, Coinbase, and Figment, which control massive ETH stakes, hold decisive influence over which upgrades activate on the mainnet.
The Core Devs are coordinators, not rulers. Their role is to build consensus among client teams and large stakers. A contentious hard fork without their buy-in fails, but so does one without backing from the staking pools that run the nodes.
Evidence: The 2022 Merge required near-unanimous client team alignment and was ultimately enacted by validators, 86% of whom were running the dominant Geth client at the time, highlighting the centralized points of failure.
key-trends
WHO DECIDES PROTOCOL CHANGES
The Four Pillars of Ethereum's Power Structure
Ethereum's governance is a multi-stakeholder game where code, capital, and consensus collide.
01
The Core Devs: The BDFL and the Client Teams
The Ethereum Foundation and client teams (Geth, Nethermind, Besu) hold soft power through technical implementation. They propose, debate, and ship code via Ethereum Improvement Proposals (EIPs).
- Key Power: Control the canonical client software used by ~90% of nodes.
- Key Constraint: Cannot force a hard fork; require broad network adoption.
~90%
Client Market Share
EIP-1559, 4844
Key Proposals
02
The Capital: Stakers and Exchanges
With Proof-of-Stake, validators (32 ETH minimum) are the ultimate veto. Lido, Coinbase, Binance and solo stakers collectively control the chain's finality.
- Key Power: Can refuse to run new client software, stalling upgrades.
- Key Dynamic: Centralization risk with Lido's ~30% of staked ETH.
32 ETH
Validator Minimum
~30%
Lido's Stake Share
03
The Users: Applications and Token Holders
Uniswap, Aave, MakerDAO and major token holders exert pressure via social consensus and economic weight. They signal support on forums and can threaten chain splits.
- Key Power: Drive demand for specific EIPs (e.g., ERC-4337 for account abstraction).
- Key Constraint: Diffuse and uncoordinated; capital is the final voice.
$10B+
Top App TVL
ERC-4337
User-Driven EIP
04
The Process: Rough Consensus and Running Code
No on-chain votes. Governance happens through All Core Devs calls, Ethereum Magicians, and research forums. The rule: "Those who do the work decide."
- Key Mechanism: Social consensus precedes client code.
- Key Result: Slow, messy, but resilient to capture compared to DAO voting.
0
On-Chain Votes
ACD Calls
Primary Forum
deep-dive
THE POLITICS
The Bazaar of Influence: How an EIP Actually Gets Shipped
Ethereum protocol changes are a multi-stakeholder negotiation where code is the final argument.
EIPs are social contracts first. An Ethereum Improvement Proposal (EIP) starts as a GitHub issue, but its success depends on social consensus before technical consensus. Developers from core teams like Consensys (Besu) or the Ethereum Foundation must champion it through All Core Devs calls.
Client diversity dictates feasibility. A change requires implementation in all major execution clients (Geth, Nethermind, Erigon, Besu). If one client team objects on technical grounds, the proposal stalls. This creates a veto-by-implementation dynamic that prioritizes stability over speed.
The roadmap is a forcing function. Proposals aligning with the official rollup-centric roadmap (e.g., Verkle trees for statelessness) gain momentum. Proposals outside it, like certain EIP-1559 extensions, face an uphill battle for mindshare and developer resources.
Evidence: The Proto-Danksharding (EIP-4844) rollout demonstrated this process. It required coordinated upgrades across all execution and consensus clients (Lighthouse, Prysm), layer-2s (Arbitrum, Optimism), and tooling providers, a multi-year effort brokered by core developers.
DECISION-MAKING AUTHORITY
Power Matrix: Stakeholder Influence on Roadmap Pillars
Mapping the formal and informal power structures that determine the trajectory of Ethereum's core protocol. This matrix reveals who has the final say, veto power, or merely advisory influence over key technical pillars.
| Roadmap Pillar / Decision Type | Core Developers (Ethereum Magicians) | Client Teams (EF, ConsenSys, etc.) | Ethereum Foundation (EF) | Stakers & Token Holders (via Social Consensus) |
|---|---|---|---|---|
Formal Governance Mechanism | Ethereum Improvement Proposal (EIP) Process | Client Implementation & Adoption | Funding & Coordination | Social Consensus / Fork Threat |
Veto Power on Protocol Changes | ||||
Direct Code Commit Access | ||||
Primary Influence on The Merge (PoS) | High (Design & Spec) | Critical (Client Execution) | High (Funding & R&D) | Medium (Approval via Staking) |
Primary Influence on The Surge (Scalability) | High (EIP-4844, DANKS) | Critical (Client Optimization) | High (Grant Funding) | Low (End-User Demand) |
Primary Influence on The Scourge (MEV) | Medium (Research, PBS) | High (Builder/Relay Implementation) | Medium (Research Funding) | Very Low (Opaque to Users) |
Primary Influence on The Verge (ZK-SNARKs) | High (Verkle Trees, EIPs) | Critical (Prover Integration) | Very High (PSE, Grant Funding) | Very Low (Technical Abstraction) |
Time to Enact Non-Consensus Change | 3-12 months (EIP Process) | Immediate (Client Update) | N/A (Not a Dev Org) | N/A (Social Process) |
Time to Enact Consensus Change (Hard Fork) | 12-24 months (Coordinated Upgrade) | 12-24 months (Client Coordination) | N/A (Coordinates, Doesn't Code) | Indefinite (Requires Overwhelming Agreement) |
counter-argument
THE POWER STRUCTURE
The Steelman: Isn't This Just Meritocracy?
Ethereum's governance is a de facto oligarchy where influence is concentrated among a few key entities, not a pure meritocracy of ideas.
Core developers hold veto power. The Ethereum Foundation and client teams like Geth and Nethermind control the implementation of protocol changes. A proposal without their technical buy-in is dead on arrival, regardless of community sentiment.
Capital concentration dictates outcomes. Large staking pools (Lido, Coinbase) and Layer 2 teams (Arbitrum, Optimism) possess overwhelming voting weight in social consensus forums. Their economic interests shape the upgrade roadmap.
The 'rough consensus' process is opaque. Final decisions emerge from closed-door discussions between core devs and major stakeholders, not from transparent, on-chain voting. This creates a governance bottleneck that prioritizes stability over radical innovation.
Evidence: The swift adoption of EIP-1559 and The Merge contrasted with the decade-long stagnation of proposals like account abstraction showcases this dynamic. Execution follows the aligned interests of the technical and capital elite.
risk-analysis
WHO DECIDES ETHEREUM PROTOCOL CHANGES
The Cracks in the Foundation: Governance Risks
Ethereum's governance is a complex, informal system where influence is concentrated, creating systemic risk for a $400B+ ecosystem.
01
The Client Team Oligopoly
Core protocol changes are dictated by a handful of client teams (Geth, Nethermind, Erigon). A single team's veto can stall upgrades, creating a centralized bottleneck. This is a single point of failure for the world's largest smart contract platform.
- Geth dominance: ~85% of execution layer clients.
- Coordination failure risk: Hard forks require near-unanimous client team sign-off.
~85%
Geth Dominance
5 Teams
Key Veto Power
02
The Whale-Driven Signaling Farce
On-chain governance signals (like Coinbase's cbETH or Lido's stETH) are gamed by large holders and protocols, not a true measure of community sentiment. This creates a facade of decentralization while voting power follows capital.
- Lido DAO controls: ~$30B+ in staked ETH voting weight.
- Exchange dominance: Centralized exchanges often vote with user-deposited assets.
$30B+
Protocol-Controlled Vote
Wealth = Voice
Core Flaw
03
The Informal Core Dev Capture
The Ethereum Improvement Proposal (EIP) process and All Core Devs calls are opaque and inaccessible. Influence accrues to those with social capital and full-time dedication, systematically excluding smaller stakeholders and creating potential for regulatory or corporate capture.
- Barrier to entry: Requires deep technical knowledge and endless meeting attendance.
- Opaque prioritization: EIP progression depends on core developer consensus, not transparent metrics.
Opaque
Decision Process
High
Capture Risk
04
The Layer 2 Governance Black Box
Major scaling solutions like Arbitrum, Optimism, and Starknet have their own centralized upgrade keys or multi-sigs, creating a meta-governance risk. Ethereum's security is only as strong as the least decentralized L2's upgrade mechanism.
- Security downgrade: Users often trade Ethereum's security for an L2's weaker governance model.
- Bridge risk: Billions in bridged assets depend on L2 team integrity.
Billions
At Risk
Multi-Sig
Common Control
05
The Staking Centralization Time Bomb
Liquid Staking Derivatives (LSDs) like Lido and centralized exchanges control ~35% of all staked ETH. This gives a few entities disproportionate influence over consensus-layer decisions (e.g., attestation voting), threatening Ethereum's credible neutrality and censorship-resistance.
- Lido's >30% share: Nears the critical threshold for consensus safety.
- Validator centralization: Top 3 entities run a majority of consensus clients.
>30%
Lido's Stake Share
Critical
Threshold Risk
06
The Application-Layer Coup
Mega-protocols like Uniswap, Aave, and MakerDAO now wield more economic power and user influence than many core development teams. Their potential to fork the chain or mandate client behavior (e.g., via MEV relays) presents a new, unaccountable form of governance pressure.
- Protocol-controlled TVL: Top 5 DeFi protocols command ~$50B+.
- Soft power: Threats of migration can force core dev hand.
$50B+
DeFi TVL Leverage
Economic Coup
Threat Vector
future-outlook
THE POWER SHIFT
The Verge of Centralization? Governance's Next Evolution
Ethereum's protocol governance is evolving from rough consensus to a formalized, multi-stakeholder model that risks institutional capture.
Formalized governance replaces rough consensus. The EIP process is now a structured battlefield where Layer 2 teams like Arbitrum and Optimism, wallet providers like MetaMask, and large staking pools like Lido and Coinbase hold disproportionate influence over client development and proposal prioritization.
Core developers retain a technical veto. While the community signals direction, the final implementation gate is held by client teams like Geth (Go-Ethereum) and Nethermind. A coordinated client refusal to implement a popular EIP creates a de facto governance deadlock.
The risk is stakeholder ossification. The current model incentivizes proposals that benefit the largest incumbents—scaling solutions and staking services. This creates a path dependency where protocol changes that threaten these entities' business models face organized resistance.
Evidence: The rapid adoption of ERC-4337 (Account Abstraction) was driven by a coalition of wallet and bundler services (e.g., Safe, Biconomy, Stackup), demonstrating how aligned commercial interests can fast-track standards that benefit their infrastructure.
takeaways
GOVERNANCE MECHANICS
TL;DR for Protocol Architects
Ethereum protocol changes are not dictated by a CEO but emerge from a complex, multi-layered governance system where influence is earned, not granted.
01
Core Devs Hold the Merge Key
Client teams like Geth (Go-Ethereum), Nethermind, and Lodestar implement the spec. A single dominant client (e.g., Geth's ~85%+ historical share) creates centralization risk. Their consensus to deploy is the final gate.
- Key Benefit: Technical rigor and security-first execution.
- Key Risk: Implicit veto power and potential for client-level centralization.
~85%
Client Share
Unanimous
Deploy Threshold
02
The EIP Process: Where Ideas Are Forged
All changes start as Ethereum Improvement Proposals (EIPs). Authors must navigate EIP-1 bureaucracy, peer review on GitHub, and debate in All Core Devs (ACD) calls. High-stakes EIPs (e.g., EIP-1559, EIP-4844) face intense public scrutiny.
- Key Benefit: Transparent, meritocratic ideation with wide input.
- Key Risk: Process can be slow, political, and bottlenecked by core dev bandwidth.
1000+
Total EIPs
<1%
Live on Mainnet
03
Stakers Are the Ultimate Upgrade Enforcers
Since The Merge, Proof-of-Stake validators (~1M+ entities) must run new client software to activate a fork. Their collective action—or inaction—is the final vote. This creates a credible neutrality check against developer overreach.
- Key Benefit: Decentralized, economic finality for upgrades.
- Key Risk: Upgrade coordination failures can cause chain splits.
>66%
Supermajority Needed
32 ETH
Stake Minimum
04
Layer 2s & Apps: The Shadow Government
Major ecosystems (Arbitrum, Optimism, Base) and apps (Uniswap, Aave, Lido) wield immense soft power. Their adoption requirements and public support can make or break an EIP (see: ERC-4337 for account abstraction). They represent the user base's economic interests.
- Key Benefit: Real-world utility drives prioritization.
- Key Risk: Corporate interests can skew development towards rent-seeking features.
$20B+
Collective TVL
Top 10
DApp Influence
05
The Ethereum Foundation: Catalyst, Not Commander
The EF funds research (e.g., Vitalik Buterin, Dankrad Feist), coordinates ACD calls, and stewards the protocol's vision. It has moral authority but no on-chain power. Its greatest leverage is allocating its ~$1B+ treasury to critical development.
- Key Benefit: Provides stability, funding, and long-term vision.
- Key Risk: Perceived centralization point for regulators and critics.
$1B+
Ecosystem War Chest
0 Votes
On-Chain Power
06
The Social Layer: The Final Arbiter
When consensus breaks, community sentiment on forums (EthResearch, Twitter) and node operator signals determine the canonical chain. This was tested in the DAO Fork and Shanghai Upgrade coordination. Code is law, until it isn't.
- Key Benefit: Resilient to capture by any single group.
- Key Risk: Subjective, messy, and vulnerable to narrative warfare.
1
Historic Chain Split
Always On
Background Pressure
ENQUIRY
Get In Touch
today.
Our experts will offer a free quote and a 30min call to discuss your project.
NDA Protected
Confidentiality24h Response
Time to ValueDirectly to Engineering Team
No Sales Fluff10+
Protocols Shipped
$20M+
TVL Overall