Ethereum has just released its most detailed upgrade plan in history. Seven upgrades. Five goals. One massive rebuild. Sketch: https://strawmap.org/ This analogy is worth exploring. The Ship of Theseus is an ancient Greek thought experiment: if you replace every single plank of a ship one by one until every single plank is replaced, is it still the same ship? This is exactly the plan Strawmap has proposed for Ethereum. By 2029, every major part of the system will be replaced. But there will be absolutely no "downtime rewrite" planned. The goal is to achieve backward-compatible upgrades, keeping the blockchain running in real time while replacing the "planks," although each upgrade will still require node operators to update their software, and certain edge cases may change. This is actually a complete rebuild disguised as an incremental upgrade. Strictly speaking, while consensus and execution logic are being rebuilt, the state (user balances, contract storage, and history) will be preserved at all forks. The "ship" is being rebuilt while still carrying cargo. Board up! “Why not start from scratch?” Because you can’t restart Ethereum without losing its core value: the applications already running on it, the funds already flowing, and the trust already established. You have to replace the planks while the ship is sailing. The name “Strawmap” is a combination of “strawman” and “roadmap.” “Strrawman” refers to an initial proposal, known from the outset to be imperfect, intended for criticism. So, it’s not a promise, but a starting point for discussion. But this is the first time Ethereum’s builders have detailed a structured, time-bound upgrade path with clear performance goals. The work involves the world’s top cryptographers and computer scientists. And it’s all open source. No licensing fees, no vendor contracts, no enterprise sales teams. Any company, any developer, any country can build upon it. The upgrades that JPMorgan Chase can enjoy are exactly the same as those that a three-person startup in São Paulo can get. Imagine if a global consortium of world-class engineers were rebuilding the internet’s financial pipeline from scratch, and you only needed… direct access. How Ethereum Actually Works (60-Second Version) Before we discuss its future direction, let's look at what it's like today. Ethereum is essentially a shared global computer.Instead of a single company running servers, Ethereum is comprised of thousands of independent operators around the world, each running a copy of the same software. These operators independently validate transactions. Some of these are called validators, and they stake their own funds (ETH) as collateral. If a validator attempts to cheat, they lose this collateral. Every 12 seconds, validators reach a consensus on which transactions occurred and their order. This 12-second window is called a "slot." Every 32 slots (approximately 6.4 minutes) constitute an "epoch." True finality, the moment a transaction becomes irreversible, takes about 13 to 15 minutes, depending on where your transaction lands in the validation cycle. Ethereum processes approximately 15 to 30 transactions per second, depending on the complexity of each transaction. In comparison, the Visa network can process over 65,000 transactions per second. It is because of this gap that most Ethereum applications today run on a Layer 2 network. Layer 2 networks are independent systems that process large volumes of transactions in batches and then send the aggregated information back to the underlying Ethereum network for security. The system that gets all operators to agree is called a "consensus mechanism." Ethereum's current consensus mechanism works well and is proven, but it was designed for an early era and limits the network's capabilities. Strawmap aims to solve all these problems. One upgrade at a time. Strawmap's Five Core Goals This roadmap revolves around five goals. Ethereum is already operational, with billions of dollars flowing through it daily. But it has real limitations on what can be built on it. These five goals are to eliminate these limitations. 1. Fast L1: Second-Level Finality Today, after sending a transaction on Ethereum, it takes approximately 13 to 15 minutes to achieve true finality, meaning the transaction is irreversible, completed, and cannot be withdrawn. Solution: Replace the engine that gets all operators to agree. The goal is to achieve finality through a single round of voting within each time slot. Minimmit is one of the leading candidates currently under research; it's a protocol designed for ultra-fast consensus, but the specific design is still being refined. The key objective is to achieve finality within a single time slot. Next, the time slot itself will be compressed: the proposed path is 12 seconds → 8 seconds → 6 seconds → 4 seconds → 3 seconds → 2 seconds. Finality is not just about speed, but more importantly, about certainty.Consider wire transfers; the time between "sending" and "settling" is the potential window of opportunity. If you're transferring millions of dollars, settling bond transactions, or completing a real estate transaction on the blockchain, that 13-minute uncertainty is a major problem. Reducing that to a few seconds would fundamentally change the network's capabilities. This applies not only to crypto-native applications but to anything involving value transfer. 2. Gigagas: 300x scaling. The Ethereum mainnet processes approximately 15–30 transactions per second. This is a bottleneck. Solution: Strawmap aims to achieve an execution capacity of 1 gigagas (billions of gas) per second, roughly equivalent to 10,000 TPS for a typical transaction (the exact number depends on the complexity of each transaction, as different operations consume different amounts of gas). The core idea is a technology called "zero-knowledge proofs." The simplest way to understand this is: currently, every operator on the network must re-execute every calculation to check its correctness. This is like every employee in a company independently recalculating the math problems of all their colleagues. Is it secure? Yes. Is it extremely inefficient? Yes. ZK proofs allow you to verify a compact mathematical "receipt" that proves the calculation process is correct. The same level of trust, but with minimal effort. The software that generates these proofs is currently too slow. Existing versions take anywhere from minutes to hours to handle complex tasks. Reducing that time to seconds (an improvement of approximately 1000x) is an active research problem, not just an engineering challenge. Teams like RISC Zero and Succinct are making rapid progress, but it's still cutting-edge. A mainnet with fast finality and up to 10,000 TPS means a simpler system with fewer moving parts. The chances of things going wrong are also lower. 3. Teragas L2: Tens of Millions of TPS Across the "Fast Lane" For truly massive transactions (and customization), you still need a Layer 2 network. Currently, L2 is limited by the amount of data the Ethereum mainnet can handle. Solution: A technique called "Data Availability Sampling" (DAS). Instead of requiring each operator to download all data to verify its existence, they are allowed to check random samples and use mathematical methods to verify that the complete dataset is intact. Imagine this as checking if a 500-page book is actually on the shelf by randomly flipping through 20 pages; if those 20 pages are present, statistically you can be certain that the rest are also present.PeerDAS has been delivered in the Fusaka upgrade, laying the foundation for the infrastructure that Strawmap relies on. Expanding from here to the ultimate goal means iterative scaling: increasing data capacity with each fork and stress-testing network stability at every step. A processing capacity of 10 million TPS across the L2 ecosystem will open doors that no current blockchain can achieve. Imagine a global supply chain where every product and item has a digital token; millions of connected devices generating verifiable data; or a micropayment system handling fractions of a cent. These workloads are overwhelming for any existing network. But with 10 million TPS, they are not only easily accommodated, but handled with ease. 4. Post-Quantum L1: Preparing for Quantum Computers Ethereum's security relies on mathematical problems that are extremely difficult for today's computers to solve. This applies to the entire system, including the signatures users use when sending transactions and the signatures used by validators to reach consensus. Once quantum computers become powerful enough, they could break both of these signatures, potentially allowing attackers to forge transactions or steal funds. Solution: Migrate to a new cryptographic approach (hash-based scheme) that is believed to be resistant to quantum attacks. This is a late-stage upgrade because it touches almost every part of the system, and the new method uses a much larger amount of data (kilobytes instead of bytes), which will change the economics of the entire network's block size, bandwidth, and storage. Quantum attacks against today's cryptography may still be years or even decades away. But if you are building a long-lasting infrastructure—one that could be worth trillions of dollars—"later" is not a real answer. 5. Privacy L1: Achieving Transaction Confidentiality All information on Ethereum is public by default. Unless you use a privacy application like Railgun, or a privacy-focused L2 like ZKsync or Aztec, every transaction, every amount, and every counterparty is visible to anyone. Solution: Built confidential transfer functionality directly into the Ethereum core. The technical goal is to allow the network to verify the validity of a transaction, whether the sender has sufficient funds, and whether the mathematical calculations are correct without revealing the actual details. You can prove "this is a legitimate $50,000 payment" without revealing who the payer is, who the recipient is, or what the purpose of the payment is. There are currently some workarounds available.In February 2026, EY and StarkWare announced Nightfall on Starknet, bringing privacy-preserving transactions to a Layer 2 environment. However, this stopgap measure increases complexity and cost. Building privacy into the infrastructure can completely eliminate the need for middleware. This is also where post-quantum work intersects: whatever privacy scheme is built, it must be resistant to quantum attacks. These are two problems that must be solved simultaneously. Once solved, a major obstacle to the widespread adoption of the technology will disappear. Seven Forks (Upgrades) Strawmap proposed a seven-upgrade scheme, carried out at a pace of about six months, starting with Glamsterdam. Each upgrade was deliberately kept to a limited scope, changing only one or two major things at a time, because if problems arise, you need to know exactly what caused them. The first upgrade after Fusaka (already released and laid the foundation through PeerDAS and data tuning) was Glamsterdam, which restructured how transaction blocks are assembled. This was followed by Hegotá, which brought further structural improvements. The remaining forks (from I to M) will continue until 2029, progressively introducing faster consensus mechanisms, zero-knowledge proofs, expanded data availability, quantum-resistant cryptography, and privacy features. Why wait until 2029? Because some of these issues are indeed unresolved. Replacing the consensus mechanism is the most difficult. Imagine changing an engine mid-flight on an airplane, with thousands of co-pilots needing to agree on every single change. Each change requires months of testing and formal verification. The effort to reduce cycle time to under 4 seconds ultimately encounters a physical challenge: a signal takes approximately 200 milliseconds to travel across the Earth and back. In a sense, you're battling the speed of light. Making ZK provers fast enough is another cutting-edge challenge. The current speed (in minutes) is about 1000 times faster than the target speed (in seconds). This requires both mathematical breakthroughs and specially designed hardware. Expanding data availability, while difficult, is relatively easier to handle. The mathematical logic is sound. The challenge lies in how to operate cautiously on a real-time network carrying hundreds of billions of dollars in value. Post-quantum migration is an operational nightmare because the new signature features are so large that they alter the economic models of all aspects. Native privacy is not only technically challenging but also politically sensitive. Regulators worry that privacy tools could facilitate money laundering. Engineers must build a system that is private enough to ensure usability, transparent enough to meet compliance requirements, and resistant to quantum attacks. These upgrades cannot be implemented simultaneously.Some upgrades depend on others. Without mature ZK proofs, scaling to 10,000 TPS is impossible. Without addressing data availability, scaling to L2 is impossible. These chains of dependencies dictate the timeline. Three and a half years is actually quite aggressive for all of this. 2029? First, there's a variable. Strawmap explicitly states: "The current draft assumes a human-led development model. AI-driven development and formal verification could significantly compress the timeline." In February 2026, a developer named YQ bet Vitalik that one could write a complete Ethereum system for the 2030+ roadmap using an AI agent. Within weeks, he delivered ETH2030: an experimental Go execution client claiming approximately 713,000 lines of code, implementing all 65 projects on Strawmap, and marked as runnable on the testnet and mainnet. Is it ready for production? No. As Vitalik pointed out, there are almost certainly fatal flaws in the code, and in some cases, these may just be stub implementations; the AI didn't even attempt a full version. But Vitalik's response is worth reading carefully: "Six months ago, this was even pure fantasy, but what matters is the trend… People should be open to this possibility (just a possibility, not a certainty!): the Ethereum roadmap may be completed much faster than expected, and security standards will far exceed expectations." Vitalik's core insight is that the correct way to use AI shouldn't be solely for speed. Instead, half of the benefits of AI should be used to improve speed, and the other half to improve security: more testing, more mathematical verification, and more independent implementations of the same functionality. The Lean Ethereum project is working on formal verification of parts of the cryptography and proof stack through machine checks. "Flawless code," long considered an ideal fantasy, may actually become a fundamental expectation. The Strawmap is a coordination document, not a promise. Its goals are ambitious, its timeline is visionary, and its execution relies on the participation of hundreds of independent contributors. The real issue isn't whether each goal can be achieved on time. It's whether you want to build on a platform with this kind of development trajectory, or compete with it.And all of this—including all the research, breakthroughs, and cryptographic migrations—was done in a public environment, free and open to everyone… This is the part of the story that truly deserves far more attention than it does now. [James | Snapcrackle]