Best of vitalik.eth.limo on Kiwi News
Most upvoted by Kiwi News readers
70 stories from vitalik.eth.limo have been shared on Kiwi News since 9 June 2023, with 401 upvotes and 50 comments in total. The latest was shared on 27 September 2026. The 50 most upvoted are listed below.
Text under a title is an AI summary of the linked article.
- Does digital ID have risks even if it's ZK-wrapped?
ZK-wrapped digital ID solves important problems, such as proving age or humanity without revealing legal identity, but it still carries risks. These mainly stem from enforcing one identity per person: applications may use a single persistent ID per user, weakening pseudonymity, and governments or employers can compel users to reveal their secrets. Zero-knowledge proofs do not address non-privacy errors such as statelessness, injury, or spoofed biometrics. Pluralistic identity, where multiple ID types or social-graph links coexist, is presented as the best realistic solution, since proof-of-wealth alone is insufficient for UBI-like uses.
- The end of my childhood
Vitalik Buterin reflects on how his role in the crypto and Ethereum community has changed over the past decade. He describes meeting younger organizers and developers at hackathons, hacker houses and Zuzalu in Montenegro, and notes that he is now one of the older people in the room. He also discusses how his thinking has shifted from mathematical governance design toward less economics-focused arguments, how the 2022 FTX collapse and his 2017 meeting with Vladimir Putin shaped his sense of responsibility, and how finite experiences give life meaning.
- Make Ethereum Cypherpunk Again
Vitalik Buterin argues that Ethereum should be steered back toward the cypherpunk values it was originally built around, namely open participation, decentralization, censorship resistance, auditability, and credible neutrality, rather than becoming a clone of existing web2 services. He attributes the drift away from these ideals mainly to high transaction fees, which he says pushed the chain toward degen gambling use. He points to 2023 progress in rollups, account abstraction, light clients, and zero-knowledge proofs as a basis for an open alternative.
- Degen communism: the only correct political ideology
Vitalik Buterin argues that current political ideologies are failing and proposes "degen communism," an ideology that embraces chaos while tilting its rules and incentives toward the common good. He applies the idea to crypto, where he proposes that project terms of service prioritize making smaller users whole first, up to a threshold such as $50k, when projects become insolvent. For government policy, he proposes land value taxes, looser building rules citing Austin, Texas, and Harberger taxes on copyrights and patents.
- Against choosing your political allegiances based on who is "pro-crypto"
Vitalik Buterin argues that crypto supporters should not choose political allegiances based on whether a politician is "pro-crypto," but on the deeper goals that drew people to crypto, such as freedom, decentralization, privacy and internationalism. He says that current pro-crypto initiatives, including the bills tracked by StandWithCrypto, judge politicians only on cryptocurrency and not on these wider freedoms. He also notes that a politician's crypto-friendly stance today may not predict their views five years from now, and suggests checking their earlier positions on topics like encrypted messaging and copyright.
- What I would love to see in a wallet
Vitalik Buterin argues that Ethereum wallets should be a priority for security and privacy, since users only gain the protections of Ethereum and its applications to the extent their wallet provides them. He proposes built-in cross-L2 sends using chain-specific addresses, and a social recovery design where a primary key handles low-value actions while a majority of N guardians (for example, N = 5) is required for high-value operations or key changes. He suggests ZK-wrapped centralized IDs such as zk-email as an easy-to-use guardian option, but advises using it for only one guardian rather than a majority. He also says private transfers should be built into wallets instead of requiring a separate privacy wallet.
- The promise and challenges of crypto + AI applications
Crypto and AI intersect in several promising ways, and the most viable is AI acting as a player in on-chain mechanisms. Its main example is prediction markets, where AIs that work for under $1 per hour could make small markets function, even with a $50 liquidity subsidy that would not attract humans. Using AI as a user interface, such as explaining transactions before signing, is described as high potential but risky, because an open-source AI assistant gives attackers the same access to tune scams against it.
- "I support it only if it's open source" should be a more common viewpoint
The author argues that the view "I support it only if it's open source" is undervalued as a middle path between rapid, unrestricted deployment of powerful technologies and control by trusted gatekeepers. The argument holds that open source improves equality of access, makes producers out of consumers, improves verifiability, and prevents vendor lock-in. Applied to proprietary life extension, for example, the author expects open source would let anyone use and offer the technology cheaply, whereas proprietary versions might be limited to billionaires and political leaders. The author also says open source carries misuse risks for things like nuclear weapons, and is skeptical of gatekeeper control because militaries have a poor history of constraining themselves, citing the Soviet bioweapons program.
- What else could memecoins be?
Vitalik Buterin argues that memecoins could be redirected toward positive-sum uses such as charity funding and genuinely fun games, rather than being purely speculative tokens. He notes that memecoins have drawn unease in 2024, partly because some Solana memecoins were openly racist and many others rise and fall in price while contributing nothing of value. Buterin cites GiveWell Inu, which donated proceeds to GiveWell, and says that half the supply of Dogelon Mars was gifted to the Methuselah Foundation, turning $ELON into a charity coin. He concludes that more fun projects that support public goods should gain attention, and he praises 0xPARC for making crypto games that players joined for fun.
- Why I support privacy
Vitalik Buterin argues that privacy is essential to decentralization and freedom, and that the Ethereum ecosystem should support it more explicitly, particularly in the digital realm. He cites the rise of AI-driven data collection and analysis, and the rise of corporate personalized pricing, as reasons privacy is becoming more urgent. He rejects government backdoors as a workable alternative, pointing to insider abuse, such as Twitter employees convicted in 2019 for selling dissidents' data to Saudi Arabia, and to hacks of sensitive government and telecom data in 2024 and 2025. He also notes that the average 19th-century conversation was never recorded, making full privacy the long-standing historical norm.
- Possible futures of the Ethereum protocol, part 1: The Merge
Ethereum's proof of stake has run stably for almost two years, but its technical design still has areas to improve, chiefly single-slot finality and a lower staking minimum. Finalizing a block currently takes 2-3 epochs (about 15 minutes), and staking requires 32 ETH, while the stated goals are one-slot finality and a 1 ETH minimum without losing economic finality. The leading options are brute-force signature aggregation handling over 1 million signatures in 5-10 seconds, Orbit committees, and two-tiered staking, with Orbit accepting a cost of attack of about $2.5 billion instead of $25 billion.
- Galaxy brain resistance
Vitalik Buterin argues that galaxy brain resistance, meaning how hard a style of reasoning is to abuse for justifying conclusions already reached for other reasons, is an important property of arguments. He contends that arguments with low resistance, such as inevitabilism, longtermism and aesthetic-based bans, are common and often function as rationalization rather than reasoning. He gives the example of "if everyone is going to automate, we should hasten it," which he says can justify disemployment, and says longtermist stories about distant benefits can justify almost anything because they cannot be checked against near-term results.
- From prediction markets to info finance
The author argues that prediction markets are a useful tool today and are one example of a broader category called "info finance," which uses market design to elicit information for public benefit. As evidence, the post says Polymarket gave Trump a greater than 95% chance of victory in the US election as results came in, while other sources had predicted 50/50 before the vote. It also cites a Venezuela market in July where bettors priced a 23% chance that Maduro would be removed, a signal the author says prompted closer attention to protests that ultimately failed to unseat him. The post further proposes AI as a way to make low-volume "micro" markets viable.
- Why I used to prefer permissive licenses and now favor copyleft
The author, Vitalik Buterin, argues that copyleft licenses now deserve preference over permissive ones, reversing his earlier stance. He previously favored permissive licensing, using the WTFPL for his blog, to maximize adoption and to avoid copyright altogether. His change rests on three points: open source is now mainstream, crypto has become more competitive so openness can no longer be assumed, and economies of scale risk power imbalances that copyleft can counter by encouraging diffusion of progress. He concludes that projects choosing permissive licenses 15 years ago should at least consider copyleft today.
- Epochs and slots all the way down
Vitalik Buterin argues that Ethereum's slot-and-epoch architecture is hard to avoid, because reaching approximate agreement takes less time than reaching economic finality. He describes Ethereum L1 transactions as confirming within 5-20 seconds today, while Gasper finality takes 12.8 minutes after two epochs. He proposes exploring based preconfirmations, promoted by Justin Drake, in which proposers sell immediate guarantees of inclusion, and notes that a 1-second slot time would shrink the case for separate L2 committee preconfirmations.
- My techno-optimism
Vitalik Buterin argues that technology is overwhelmingly beneficial and that its benefits should not be discounted, but that the direction of technological development matters and needs deliberate human choices rather than following profit alone. He cites the rise in life expectancy over the last century, which improved across all regions despite wars and other calamities, and notes that the 1952 London Great Smog was followed by a century of declining air pollution. He attributes these improvements to intentional action such as the Montreal Protocol, which helped the ozone layer recover, rather than to an automatic technological progression.
- Mid-term future of improving Ethereum’s permissionlessness and decentralization
Vitalik Buterin argues that Ethereum's permissionlessness and decentralization can be improved through protocol changes already in progress and realistic roadmap tweaks, responding to concerns raised by Geth core developer Peter Szilagyi. On MEV, he favors minimizing MEV alongside "MEV quarantining" via proposer/builder separation, and he supports inclusion lists that let proposers force transactions into blocks, with the view that the builder's role should shrink further. On staking, he says most non-solo stakers cite the 32 ETH minimum, and that verkle trees with EIP-4444 and a larger validator set could address the barriers to solo staking.
- Making Ethereum alignment legible
Vitalik Buterin argues that "Ethereum alignment" should be made legible by breaking it into specific properties that can be measured, rather than judged by social connections. The proposed properties include open source code, open standards, decentralization and security (tested by the walkaway test and the insider attack test), and benefit to the wider Ethereum community and the world. The author suggests that independent monitors like L2beat should track projects against these criteria, with the Ethereum Foundation funding such efforts but not performing them itself.
- Ethereum has blobs. Where do we go from here?
Vitalik Buterin argues that Ethereum's scaling phase is largely complete now that blobs are live, and that the focus should shift toward rollup security, application design and user experience. The Dencun hard fork activated on March 13 with EIP-4844, which cut rollup transaction fees by over 100 times at first, and blob usage later spiked as the Blobscriptions protocol began using them. The post outlines next steps: PeerDAS, which would let each node store about 1/8 of blob data, improved data compression and execution scaling for rollups, and stricter standards so that only projects reaching at least stage 1 on l2beat count as rollups. It also argues that the lack of large-scale applications came from high fees, and that developers now have the tools to build usable apps.
- My response to AI 2027
Defenses in biosecurity and cybersecurity could make the AI 2027 scenario's path to human extinction far less certain than it portrays. Air filtering, UVC, real-time viral sequencing and immune enhancement could together cut airborne disease transmission by perhaps 10-20x, enough to stop existing airborne diseases, even measles, from spreading. The critique also notes that AI 2027's own infographic expects cancer and aging to be cured by 2029, and that cybersecurity's endgame is likely defense-favoring given the rapid technology development the scenario assumes.
- Simplifying the L1
Vitalik Buterin argues that Ethereum can become much simpler, approaching Bitcoin's protocol simplicity within five years, and that simplicity is as important to resilience and scalability as throughput gains. He states that simplicity lowers the barrier to protocol participation, reduces infrastructure and maintenance costs, and lowers the risk of catastrophic bugs and special-interest capture. For the consensus layer, the 3-slot finality redesign removes separate slots and epochs, committee shuffling and sync committees, and a basic implementation takes about 200 lines of code. For the execution layer, he proposes replacing the EVM with RISC-V, which he says could bring a 100x+ performance gain in many cases based on Succinct data, with a staged migration.
- Layer 2s as cultural extensions of Ethereum
Vitalik Buterin's essay "Layer 2s as cultural extensions of Ethereum" argues that blockchains have distinct cultures that shape their technical direction, participants and legitimacy, and that a layer-2-centric Ethereum allows subcultures to pursue their own approaches while sharing common infrastructure. The article identifies three Ethereum subcultures described by Paul Dylan-Ennis: cypherpunks, who build infrastructure and tools; regens, who run governance and public goods experiments; and degens, who trade memecoins and NFTs. It argues that layer 2s are the strongest playing field for such subcultures because each is both an ecosystem and organized around building something. It cites EOS's delegated proof of stake system as an earlier two-level attempt that produced autonomous sub-communities but proved unstable.
- Should Ethereum be okay with enshrining more things in the protocol?
This Vitalik Buterin post from September 2023 asks whether Ethereum should enshrine more features into its core protocol, and it traces why the original minimalist approach was hard to sustain for account abstraction. The early goal was to make a transaction "just a call," with validation moved into user account code, but EIP-86 pushed complexity onto miners and other layers. Account abstraction then evolved through EIP-2938 into ERC-4337, which the post says has a medium-term roadmap to enshrine parts of it into the protocol. Stated reasons include gas efficiency, since ERC-4337 costs about 42,000 gas versus 21,000 for a basic transaction, and censorship resistance, since wrapped user operations are opaque to protocol-level inclusion lists.
- Low-risk defi can be for Ethereum what search was for Google
Vitalik Buterin argues that low-risk defi, covering payments, savings, synthetic assets, fully collateralized lending and exchange between these assets, can do for Ethereum what search and ads did for Google by sustaining the ecosystem economically while staying aligned with its values. He contrasts this with Google's ad-driven incentives, which encourage collecting user data, and says Ethereum's decentralization makes such alignment more important. The post credits a shift in regulation and falling protocol risk for making defi more viable, and it cites Aave deposit rates for major stablecoins and Ethereum L1 defi losses as supporting data.
- We should talk less about public goods funding & more about open source funding
Vitalik Buterin argues that the term "public goods" has become too easily stretched, with commercial projects describing themselves as "building a public good", and that "open source" is a clearer alternative. He notes that central examples of digital public goods, such as academic blockchain and cryptographic research, documentation and tutorials, and open source software like Ethereum clients, are all open source. He adds that open-sourcing is itself a public good with diffuse benefits, and that open source has a well-agreed definition from the FSF and OSI. He also suggests that for physical goods, such as clean air and defense technology, open source is the most efficient route to global scale.
- On idea-driven ideas
Vitalik Buterin argues that good decision-making in complex settings needs both idea-driven reasoning, meaning principles and ideologies that generate intermediate steps, and pragmatic reasoning, and that ideologies are best understood as a mix of means and ends. He says the world is too complex to reason pragmatically through every decision, and that principles give communities a robust basis for coordination, since they act as a brake against clever arguments for harmful actions. He also says that different people holding narrower goals enables specialization and measurement, and he notes that ideologies can cause motivated reasoning, such as climate hawks resisting solar geoengineering and crypto enthusiasts resisting non-blockchain solutions.
- Possible futures of the Ethereum protocol, part 3: The Scourge
Vitalik Buterin's essay on the Ethereum "Scourge" roadmap argues that proof-of-stake centralization driven by economies of scale is one of the biggest risks to Ethereum's base layer, since it could lead to 51% attacks, transaction censorship and value extraction from users. The article says block construction is largely handled by two actors choosing the contents of roughly 88% of Ethereum blocks through MEVBoost. It describes inclusion lists, multiple concurrent proposers such as BRAID, and encrypted mempools as proposed ways to reduce the authority of these specialized builders.
- Review: museums of the future, Dubai and Tokyo
The author compares two museums of the future: the Museum of the Future in Dubai, which imagines bold technological transformation such as a 2064 space elevator and genetic engineering that improves the biosphere, and the Miraikan in Tokyo, which emphasizes environmental crisis, public participation, and social technology like governance. The author concludes the two visions are complementary, and that the ideal museum would combine Dubai's boldness with the Miraikan's inviting spirit. The Dubai museum is described as consumerist and missing governance, while the Tokyo museum is described as inviting people to learn and take part in building the future.
- Multidimensional gas pricing
Vitalik Buterin's post argues that Ethereum's single "gas" resource treats distinct limits as interchangeable, and that multidimensional gas pricing can fix this. Gas merges raw computation, storage access, data bandwidth and ZK-SNARK proving cost, so the gas limit can exclude safe blocks or accept unsafe ones. The post says EIP-4844 blobs already added a separate data dimension, making rollups about 100x cheaper and raising the theoretical maximum block size only from ~1.9 MB to ~2.6 MB. It also proposes separate storage-access limits for stateless clients, such as 1,000 average and 2,000 per-block accesses, and a stateless-client-proof-gas dimension. The post presents EIP-7623's per-transaction max rule as a simpler but weaker step, and says the ideal approach is multidimensional EIP-1559 pricing.
- Vitalik's take on the Network State
Balaji Srinivasan released the first version of his book on "network states" on July 4, 2022, describing communities that start as online clubs, build a physical presence, and may eventually seek political autonomy or diplomatic recognition. The article's author, Vitalik Buterin, frames network states as an ideological successor to libertarianism that departs from The Sovereign Individual by centering moral and community ties rather than individualism and money. The article explores Balaji's examples, such as a health-focused "Keto Kosher" society and a medical sovereignty zone free of FDA rules, and notes that Balaji's book blends the network state concept with a broader megapolitical theory.
- Vitalik: My 40-liter backpack travel guide
Vitalik Buterin describes the travel setup he has used for nine years of nomadic life, built around a 40-liter Hynes Eagle backpack that holds all his supplies and a laptop bag. He says the guiding principle is minimizing cubic centimeters rather than dollars, and that the approach depends on being a USB-C maximalist, so that one charger and one or two cables cover all devices. He also reports that about 70% of his clothing is from Uniqlo, that layering lets three items cover a wide range of temperatures, and that he changes locations no more than once every seven days to avoid disorientation.
- Plurality philosophy in an incredibly oversized nutshell
Vitalik Buterin argues that crypto must shift from circumventing outside governance to building its own, and that a principled framework valuing both freedom and democracy, called Plurality, can handle modern tensions between coordination and freedom. He cites Ron DeSantis's Florida ban on lab-grown meat, which Libertarian Party New Hampshire criticized, and Google's Android feature that warns users of likely scams during calls, which Meredith Whittaker of the Signal Foundation opposed. The post introduces Plurality as the book by Glen Weyl and Audrey Tang, which treats connections between individuals as a first-class design object and describes three ideological poles: Technocracy, Libertarianism and Plurality.
- Possible futures for the Ethereum protocol, part 2: The Surge
Vitalik Buterin's "Possible futures of the Ethereum protocol, part 2: The Surge" argues that Ethereum's scaling depends on completing the rollup-centric roadmap, in which the L1 serves as a robust, decentralized base layer and L2s handle most scaling. The article sets a medium-term target of 16 MB per slot, which it says would give about 58,000 TPS with rollup data compression. It says the March 13, 2024 Dencun upgrade gave the chain three ~125 kB blobs per 12-second slot, and it describes PeerDAS as the next step toward data availability sampling.
- Plinko PIR tutorial
Plinko is a private information retrieval protocol that lets a client read from a large database without the server learning which entry it wants, and it reduces communication and server compute from O(N) to O(√N) where N is the number of cells. The protocol is built on classic two-server PIR, whose weaknesses are the need to trust one of two servers, high communication overhead, and server work proportional to half the data per query. Plinko removes the trust assumption by using preprocessing: the client downloads the whole dataset once and keeps roughly 128·√N hints. For a workload of 10 billion 32-byte values, the article estimates the client would store about 390 MB of hints for a 100000×100000 grid, before backup hints and Merkle branches are added.
- A GKR Tutorial
This post is a tutorial explaining GKR, the family of interactive protocols it describes as behind the extreme speed of many recent ZK proving systems, such as ZK-EVM provers and zk-ML systems. GKR works by running a computation backwards layer by layer, using sumchecks to turn an obligation about one layer into an obligation about the previous layer, and it commits only to the inputs and outputs rather than to every intermediate layer. The author notes that a Merkle-tree commitment to an erasure coding requires hashing 4 to 16 bytes per byte committed, which GKR avoids except at the start and end. The post focuses on a GKR implementation for proving Poseidon2 hashes, with examples computed modulo 89, and notes that real systems use a 32-bit prime field such as KoalaBear.
- The math of when stage 1 and stage 2 make sense
Under a simplified model, stage 0 rollups are almost never justified, and a rollup should launch at least directly into stage 1. The model assumes each security council member has an independent 10% chance of breaking, which gives a stage 0 rollup a fixed 0.2728% chance of failing with a 4-of-7 council. To address concerns that slow 6-of-8 signing could delay fixing a critical bug, the text suggests letting any single council member delay withdrawals by one or two weeks. It also cautions against rushing to stage 2 at the expense of hardening the proof system.
- Scaling Ethereum L1 and L2s in 2025 and beyond
Vitalik Buterin argues that Ethereum should keep scaling mainly through layer 2 rollups while making them meet the security and interoperability standards they were meant to have. He says L2s now secure billions of dollars and are scaling Ethereum's transaction capacity by 17x, and that blobs are the main bottleneck. After EIP-4844 provided 3 blobs per slot, Pectra was scheduled for March to double that to 6, with PeerDAS then expected to add another 2-4x.
- d/acc: one year later
Vitalik Buterin argues that d/acc, or decentralized and democratic differential defensive acceleration, should guide technology so that it favors defense over offense and distributes power rather than concentrating it. He says the philosophy has matured over the past year, citing progress in open-source vaccines, Community Notes, prediction markets, ZK-SNARKs in government ID and social media, and open-source medical and brain-computer interface imaging tools. The first d/acc event, "d/acc Discovery Day" at Devcon, featured a full day of speakers across bio, physical, cyber, info defense and neurotech. He presents the approach as an alternative to e/acc-style acceleration without defense, centralized control, and deceleration.
- Possible futures of the Ethereum protocol, part 4: The Verge
The Verge is Ethereum's roadmap goal of making full verification of the chain cheap enough for ordinary devices, including mobile wallets and smart watches. The article says the main obstacle is state size, which exceeds hundreds of gigabytes and grows by about 30 GB per year. It describes stateless verification, where blocks carry witnesses with the needed state values and proofs, and compares two leading approaches: Verkle trees, with proofs of roughly 2.6 MB worst case, and STARKed binary hash trees, which depend on prover speed. It notes that Polygon proved 1.7 million Poseidon hashes per second on a laptop using circle STARKs.
- Glue and coprocessor architectures
Vitalik Buterin argues that modern computation is increasingly organized as a "glue and coprocessor architecture," in which a general but inefficient glue component handles business logic while specialized, efficient coprocessors perform the bulk of intensive work. In a recent Ethereum transaction updating an ENS hash, about 73% of EVM execution (about 85% including the computational part of the base cost) went to storage reads and writes, logs, and cryptography. The same split appears in Python with torch, where matrix multiplication dominates, and in STARK proving, where specialized modules handle hashes and signatures alongside a RISC-V prover. The article suggests the EVM should be optimized through better precompiles and storage layout rather than by changing the VM itself.
- Some reflections on the Bitcoin block size war
This long-form reflection, written by an author who says he was a "big blocker" during the Bitcoin block size war, compares Jonathan Bier's "The Blocksize War" with Roger Ver and Steve Patterson's "Hijacking Bitcoin" and then gives his own assessment. Bier's account frames the small-block side as defending user control and decentralization against miner and exchange cabals, such as the 2017 New York Agreement. Ver's account frames the big-block side as defending on-chain affordability against self-appointed developers and Blockstream. The author sympathizes with big blockers on Lightning's untested viability and the governance argument, while faulting them for lacking a realistic limiting principle for block size.
- Ask security questions
Vitalik Buterin argues that audio or video of a person can no longer serve as a secure way to authenticate identity, and that security questions based on shared personal experiences are a practical answer. He cites a company that lost $25 million after a finance worker was convinced by a deepfaked video call to send a bank wire to a scammer posing as the CFO, and notes that deepfakes of him have been used in scams and a dog coin promotion. He argues that cryptographic signatures alone fail in multisig approvals, since verifying a signer's key lets an attacker with that key pass as the owner. The article recommends stacking security questions with code words, delays and multi-channel confirmation.
- Different types of layer 2s
Vitalik Buterin's article "Different types of layer 2s" argues that layer 2 projects are becoming more heterogeneous, and that applications should choose a point on a spectrum between rollups and validiums based on their own security and fee needs. Rollups post computation proofs and data to Ethereum layer 1, so assets can always be withdrawn to L1, while validiums store data off-chain, so data availability failures can cause assets to be lost but not stolen. Ethereum's Dencun upgrade, via EIP-4844, is said to provide about 32 kB/sec of onchain data availability, with full danksharding eventually targeting about 1.3 MB/sec. The article also says a bridge that only checks consensus signatures gives no asset guarantees, while adding a ZK-SNARK check makes the chain's security model match a validium's.
- What do I think about biometric proof of personhood?
Proof of personhood aims to tie each registered account to a unique real human, and biometric versions such as Worldcoin's iris-scanning Orb raise tradeoffs that must be weighed against social-graph approaches or giving up on proof of personhood entirely. Worldcoin, co-founded by Sam Altman, scans each user's iris with the Orb and stores only hashes of the scans to check uniqueness. The text lists four major risks for Worldcoin's design: privacy, accessibility, centralization through possible Orb backdoors, and security issues such as hacked phones or 3D-printed fake people.
- My self-sovereign / local / private / secure LLM setup, April 2026
Vitalik Buterin describes an approach to AI that treats privacy, security and self-sovereignty as non-negotiable, built on local LLM inference, local files and sandboxed agents. He argues that the wider AI ecosystem, including local open-source projects, is cavalier about privacy and security, citing criticism of OpenClaw and the claim that roughly 15% of the skills seen contained malicious instructions. His setup runs Qwen3.5:35B through llama-server on a laptop with an NVIDIA 5090 GPU, reaching about 90 tokens per second, and uses the pi agent framework inside a bubblewrap sandbox. He also notes that he found the DGX Spark slower than a good laptop GPU and that the post is meant as a starting point rather than a finished, secure product.
- The tree ring model of culture and politics
The article proposes a "tree ring model" of culture and politics, arguing that how a society treats new things reflects the attitudes and incentives prevalent when they emerge, while treatment of older things is driven mainly by status quo bias. It cites examples including the 1990s deregulatory trend in the USA, which shaped internet freedom, and social media, which came of age in the 2010s and is treated as distinct from the internet. It also argues that China's support for open-source AI in the 2020s reflects a "commoditize the complement" strategy. The author concludes that inventing new norms and technologies is easier than changing entrenched ones.
- AI as the engine, humans as the steering wheel
AI agents would compete as open-market "solvers" while humans steer through simple verification mechanisms, rather than a single AI governing. The author argues that a single AI fails credible neutrality, since LLMs encode specific preferences, are about as complex as all US law, and would need replacing every three months. Futarchy, from Robin Hanson's "vote values, but bet beliefs," fits this pattern, with traders predicting outcomes and voters verifying a chosen metric. In distilled human judgement, a jury answers about 100 of a million questions and checks submitted answer lists; in a toy example, a blend of 0.199 Claude and 0.801 Deepseek answers fit the jury best.
- Possible futures of the Ethereum protocol, part 5: The Purge
Vitalik Buterin's Ethereum roadmap essay "The Purge" argues that Ethereum must reduce growing bloat and complexity while preserving the permanence that makes blockchains valuable. The article says a full-synced node needs roughly 1.1 terabytes of disk space, with about 800 GB of that being history, which grows by hundreds of gigabytes each year. It proposes distributing history storage across nodes, noting that a network of 100,000 nodes each storing a random 10% of history would replicate each piece of data 10,000 times. It also says state adds around 50 GB per year, and that state expiry must keep balances and contracts accessible to users who stay inactive.
- Exit games for EVM validiums: the return of Plasma
Vitalik Buterin's post argues that Plasma-style designs, which were largely superseded by rollups, can be revived for EVM chains using validity proofs, though the security guarantees would not cover all users. Validity proofs would let the operator's invalid-block attacks be removed, leaving only unavailable blocks as a concern, and they would reduce the state a user must download from one branch per block in the last week to one branch per asset. The post also proposes a parallel UTXO graph for ETH and ERC20 tokens, with validity proofs showing equivalence to the EVM state, so that withdrawals from the most recent state could be instant in the normal case. It notes that objects without a clear owner, such as a CDP with 1 ETH locked against 1000 DAI of debt, remain where exit games break down.
- Vitalik: What do I think about Community Notes
Vitalik Buterin argues that Twitter's Community Notes feature is among the closest mainstream approximations of crypto values, because anyone can write and rate notes and an open source algorithm decides which notes appear. He says the notes are informative and valuable both in his view and in that of many people across the political spectrum he talks to. The algorithm prioritizes notes rated helpful by users who usually disagree, using a "polarity" term that absorbs partisan effects and a "helpfulness" score that must reach at least +0.4 for a note to be shown. Buterin says the polarity axis emerged from the data and maps closely to left versus right, and that the full implementation is 6282 lines across 22 files.