# Being Right Isn't Enough - Lessons for Ethereum from a 90s OS War #Web3 *Last Updated: February, 2026* *(This essay is part of a series of reflections I am writing while pursuing the 2026 Ethereum Protocol Study Group (EPS). It was inspired by [Lecture 1](https://study.epf.wiki/mod/hvp/view.php?id=17), which covers Ethereum's predecessors and the culture it was built on. Explore the [EPF Wiki](https://epf.wiki/#/) to learn more or check out this [thread](https://x.com/joshdavislight/status/2023860142098596320) by [@joshdavislight](https://x.com/joshdavislight))* GNU Hurd was the better answer. Linux won anyway. That's the short version. The longer version is more unsettling, and if you're building anything meant to outlast the next wave of competition, it's worth sitting with. Richard Stallman's [GNU Hurd](https://www.gnu.org/software/hurd/) was architecturally elegant, built on a [microkernel design](https://www.gnu.org/software/hurd/hurd-paper.html#design) that separated concerns beautifully. It was what computer scientists of the era [considered the correct way to build](https://groups.google.com/g/comp.os.minix/c/wlhw16QWltI/m/T1L667lVAAAJ) an operating system. Linus Torvalds' Linux was the opposite: a monolithic kernel, fast and loose, the kind of thing that earned [public mockery](https://en.wikipedia.org/wiki/Tanenbaum%E2%80%93Torvalds_debate) from academics. Tanenbaum famously called it obsolete before it was even finished. Yet Linux is everywhere. Hurd is a [historical footnote](https://www.gnu.org/software/hurd/). > ***History doesn't grade on a curve for architectural virtue.*** I keep coming back to this story when thinking about Ethereum. Not to predict doom. Ethereum is in a far stronger position than Hurd ever was, but because the comparison functions as a useful memento mori. This write-up is about the lessons from the Hurd/Linux saga I can't shake. ## Before the Lessons A confession. The Hurd/Linux analogy, taken solely and seriously, does not predict that Ethereum learns its lesson in time. It predicts that Ethereum loses. The historical pattern is not "architecturally sophisticated platform gets warned, reforms, and wins." The pattern is "simpler, faster, more pragmatic entrant displaces the incumbent regardless of technical merit." If you run the analogy straight, Ethereum is Hurd in terms of a system with a better answer that real developers found too slow and too complicated to wait for. These lessons matter precisely because there is no guarantee Ethereum is the platform that reads them in time. Secondly, each lesson below includes what Ethereum has announced or initiated in response to the risk described. Those announcements are genuinely positive signals. They indicate the right problems are being named. But announcements are where Hurd supporters also lived. The Hurd project had roadmaps, mailing list threads, and architectural documents that were entirely correct about what needed to happen. The question is not whether Ethereum's leadership has diagnosed the problems accurately. The question is whether those diagnoses are translating into shipped, adopted outcomes. Where the evidence below is an EF blog post or a stated intention, read it as a leading indicator, not a lagging one. The real measure will be developer retention numbers, transaction fee trends, and whether the tooling improvements are in users' hands. ## Lesson 1: Running Software Beats Correct Software Hurd spent nearly a decade with almost nothing you could actually use. No real hardware support. No drivers worth mentioning. Theoretically superior but practically absent. Linux shipped. It ran on real machines, supported real peripherals, and let developers build real things. > ***There's a version of this risk lurking inside Ethereum's roadmap.*** The [previous version](https://x.com/VitalikButerin/status/1588669782471368704?s=20) of the Ethereum Roadmap, showcasing The Merge, The Surge, The Scourge, The Verge, and The Purge, was technically justified with a coherent long-arc vision. But when the "real" Ethereum is always the next version, users and developers start hedging elsewhere. Layer 2s have bought time here by shipping usability now, but the fragmentation that followed is its own kind of Hurd trap. Progress deferred is opportunity handed to competitors. That risk hasn't gone away. But the last twelve months are harder to dismiss than they used to be. The most honest answer to the "always the next version" critique is what Ethereum actually shipped. In 2025, [Pectra](https://blog.ethereum.org/2025/04/23/pectra-mainnet) landed in May with 11 EIPs, including [EIP-7702](https://eips.ethereum.org/EIPS/eip-7702) (programmable wallets), which saw [over 11,000 onchain authorizations within a week of launch](https://www.theblock.co/post/354414/smart-wallet-adoption-surges-after-pectra-upgrade) from wallets including MetaMask, OKX, and WhiteBIT. [Fusaka](https://blog.ethereum.org/2025/11/06/fusaka-mainnet-announcement) followed in December with PeerDAS and an 8x increase in theoretical blob capacity. The mainnet gas limit doubled from 30M to 60M, the first significant increase since 2021. This makes more room for transactions and more room for L2s, the two numbers that determine whether Ethereum stays cheap under pressure. The EF's [February 2026 Protocol Priorities update](https://blog.ethereum.org/2026/02/18/protocol-priorities-update-2026) named the next two forks, Glamsterdam (H1 2026) and Hegotá (H2 2026), with a defined scope and a [strawmap](https://strawmap.org) through 2029. More importantly, it institutionalised a twice-yearly upgrade cadence, the first time Ethereum has had a predictable delivery rhythm rather than one defined by R&D readiness. ## Lesson 2: Ecosystem Gravity Is Hard to Build, Easy to Lose Once Linux accumulated device drivers, package ecosystems, and community tooling, Hurd couldn't catch up, even as Hurd improved. The switching cost wasn't technical. It was the accumulated weight of everything built *around* Linux. > ***Ethereum has this moat today. Solidity developers, EVM tooling, DeFi composability, institutional familiarity. That's real. But it's not permanent.*** Solana isn't just competing on throughput; it's building a parallel ecosystem of tooling, developer habits, and user onboarding flows. Every time a developer chooses Rust on Solana because the experience feels better, that's a small Hurd moment. Individually negligible. Collectively, the kind of thing that turns a dominant platform into a cautionary case study. The risk is real. So is the response, though whether it's fast enough is a separate question. One of the leading causes weakening Ethereum's ecosystem gravity is the fragmentation trap across L2s. The [Open Intents Framework](https://docs.openintents.xyz/docs) addresses this directly: it lets users declare desired outcomes (move funds, swap, pay) and lets solvers find optimal cross-chain routes. It's no longer a specification; Eco Routes, Uniswap's cross-chain swaps via Across, and Optimism's Superchain bridge interfaces are all running on it today. The [Ethereum Interoperability Layer (EIL)](https://blog.ethereum.org/2025/11/18/eil) goes further, explicitly targeting cross-chain finality of 15–30 seconds, down from 13–19 minutes, designed to make multi-L2 interactions feel like a single chain. But infrastructure alone doesn't explain why a developer or user stays on Ethereum rather than migrating elsewhere. That's what the recent ["Defipunk" commitment](https://blog.ethereum.org/2026/02/23/commitment-to-defi) is attempting to answer. The EF's position, announced in February 2026, is that Ethereum's DeFi should be permissionless, open-source, privacy-first, and built to pass the [walkaway test](https://x.com/VitalikButerin/status/2026338053162414545?s=20): A protocol should keep running even if its founding team disappears or turns hostile. No admin keys, no emergency backdoors. It is a credible-neutrality play, an attempt to make the ecosystem itself the differentiator, not just the underlying chain. ## Lesson 3: Governance Coherence Beats Governance Purity GNU Hurd suffered from slow, ideologically filtered decision-making. Linux had Torvalds who was opinionated, sometimes abrasive, but decisive. Patches got merged or rejected. Things moved. Ethereum's governance is deliberately decentralized, and that's philosophically coherent. But the EIP process can produce years-long delays on changes that look routine from the outside. The counter-argument I find most honest: decentralized governance isn't a bug in Ethereum's design, it is the product. Trustlessness requires no single point of control. Still, the Hurd lesson stands. Ideological purity in governance can become a bottleneck that competitors exploit. > ***The question isn't whether to decentralize. It's how to coordinate faster without betraying the core commitment.*** The twice-yearly cadence is a real improvement, but it proves a narrower thing than it seems. Knowing *when* decisions will ship is different from knowing *how* contested decisions get made. The cadence solves a coordination problem. It doesn't touch the harder one. [FOCIL (EIP-7805)](https://eips.ethereum.org/EIPS/eip-7805) makes this a little more concrete. A [technically contentious change](https://www.theblock.co/post/390682/vitalik-buterin-is-building-a-cypherpunk-principled-non-ugly-ethereum-as-devs-officially-add-focil-to-upgrade-roadmap) debated hard before landing in Hegotá's confirmed scope isn't evidence that governance got faster. It's evidence that the mechanism works, albeit slowly, through disagreement, by rough consensus. That's not a failure. But it's also not Torvalds. Ethereum cannot have a Torvalds and shouldn't want one. The real risk isn't the absence of a benevolent dictator. It's that coordination costs compound: across hundreds of independent teams, across client diversity requirements, across a community that treats legitimacy as a precondition for any change. Competitors with simpler governance structures will sometimes move faster, and some of those changes will matter. Whether the speed disadvantage is worth the trustlessness guarantee is a question Ethereum keeps deferring. The fork cadence helps at the margins. That question remains open. ## Lesson 4: "Worse Is Better" Is Real, and It's Uncomfortable Richard Gabriel [wrote about this](https://www.dreamsongs.com/RiseOfWorseIsBetter.html) directly. Unix and C were simpler, less correct, more pragmatic who beat Lisp's more elegant systems because good-enough solutions that ship win over perfect solutions that don't. Linux was "worse" than Hurd by the measure of OS theory. It won because of that, not despite it. This may be the sharpest edge for Ethereum. If a simpler blockchain with lower costs, more accessible, somewhat less decentralized delivers enough utility to capture the next wave of developers and users, Ethereum's theoretical superiority in trust model may not save it. This isn't an argument for abandoning principles. > ***It's an argument for being brutally honest about where "good enough" serves users better than "correct."*** L2s are partly an acknowledgement of this: keep the base layer trustworthy, let the execution layers be fast and cheap. The [Trillion Dollar Security initiative's Phase 2](https://blog.ethereum.org/en/2025/08/20/trillion-dollar-sec-2) extends this logic into UX: the EF surveyed the ecosystem, identified wallet security and blind signing as the highest priority friction points, and launched a minimum wallet security standard, complete with a public grant to [Walletbeat](https://walletbeat.xyz) and a push for transaction decoding that shows users human-readable outcomes instead of raw calldata. Another pillar is less visible but equally pragmatic: a shared open-source database of smart contract vulnerabilities, designed to integrate with IDEs so developers get flagged for known vulnerability patterns before deploying. No new chain required. No architectural elegance. Just the EF doing coordination work that no individual team has enough incentive to do alone. That's Ethereum choosing "good enough, right now" over "architecturally correct, eventually." Sometimes winning looks like plumbing. ## Lesson 5: Being Right Doesn't Mean You Win GNU Hurd was right about microkernel architecture in many ways. Modern macOS (XNU) and various server architectures vindicate what Hurd was attempting. But being right didn't matter, because Linux had won. That was that. Ethereum's decentralization and censorship resistance may be genuinely correct for the long-term architecture of a global financial coordination layer. > ***At the risk of sounding fatalistic: correctness doesn't guarantee adoption.*** The protocol that gets embedded into the habits, tools, and institutions of the next billion users will be extraordinarily hard to displace, regardless of whether a better one exists. That makes user and developer experience not a secondary concern downstream of protocol correctness. It makes it existential. The clearest signal that Ethereum is building for permanence isn't a treasury decision, it's the [Trillion Dollar Security initiative](https://blog.ethereum.org/2025/05/14/trillion-dollar-security), an explicit, named, multi-year program to make Ethereum secure enough for sovereign and institutional-scale value storage. The target is a world where billions of individuals are comfortable holding $1,000 onchain, and institutions are comfortable storing $1 trillion in a single contract. There's now a [public dashboard](https://trilliondollarsecurity.org) measuring progress across six security domains (UX, smart contracts, infrastructure, consensus, incident response, and governance) so the community can hold the initiative accountable in public. The post-quantum readiness work (hash-based signatures, STARKs, quantum-resistant precompiles) and the Hegotá-targeted Verkle tree transition (reducing node hardware requirements and enabling stateless clients) are investments in the same direction: they have no short-term marketing payoff, and they look unnecessary until they aren't. (Whether these commitments survive the organizational pressures of a bear market cycle is the open question the dashboard itself cannot answer.) ## The Real Lesson GNU Hurd teaches that the gap between "building the right thing" and "building the thing that wins" is where ecosystems go to die. Ethereum has what Hurd never did: real users, real capital, real network effects. Those are genuine advantages. They are also exactly the kind of advantages that look decisive right up until they don't. Stay obsessed with shipping. Lower friction relentlessly. And never mistake being architecturally correct for being actually used.