Ethereum co-founder Vitalik Buterin says the current moment calls for a new kind of minimalism: digital minimalism. In a recent essay, he argues that the point is not to reject digital technology, but to use it deliberately, so that time online and time inside apps are chosen by the user rather than shaped by recommendation systems and platform habits.
Buterin writes that he has long practiced minimalism in physical life and even published a blog post four years ago about fitting everything he needed for travel into a 40-liter backpack. But he says the material minimalism that gained traction in the 2000s and 2010s was a response to the problems and opportunities of that era. Today, he argues, the pressure points are different, and they sit inside digital systems.
Stepping back from algorithmic social media
Buterin starts with social media. He says X has long been filled with negative content and that the situation has worsened in recent years. The platform is highly attention-grabbing, he writes, but much of what it surfaces carries little informational value and often leaves users feeling worse rather than better informed.
He points to late 2024 and early 2025 as the period when that feeling peaked for him. During that stretch, he says, the platform saw a wave of pro-Hitler posts. Around the same time, crypto Twitter also fell into what he calls an atmosphere of “financial nihilism,” a view that blockchain’s natural endpoint is gambling and that anyone hoping the industry can move in a better direction is merely acting like a self-righteous elitist.
He says he already disagreed with both currents of thought, but the larger realization was that these discussions were easy to keep consuming without actually making him wiser. They pulled attention and drained emotion. For most of this year, he says, he ran an experiment: he did not delete his X account or leave the platform entirely, but he stopped using X’s official client for browsing and posting. Instead, he moved to Firefly.social.
According to Buterin, Firefly is an aggregation client that can connect to X, Bluesky, and the Ethereum-based decentralized social platforms Farcaster and Lens. It lets users read content from all four in one interface and publish a post across all of them at once.
He says his original reason for using Firefly was to support decentralized social media while still remaining active on X, spreading network effects rather than leaving all traffic locked inside a single platform. After several months, though, he found another benefit: the experience felt much better.
His explanation is straightforward. Firefly lets him read and post, but it does not include what he sees as X’s central mechanism: the recommendation algorithm. He describes the loop this way: a user sees a pro-Hitler post, clicks into it, reads the quote posts and replies out of the same kind of curiosity that makes people stare at a car crash, and the algorithm then interprets that behavior as interest. It labels the user accordingly and serves more of the same. The cycle repeats and intensifies.
Buterin notes that over the following year and a half, the pro-Hitler trend on the platform faded sharply. He attributes that partly to the fact that such extreme currents lose momentum once more people recognize them as both evil and foolish, and partly to improvements in X’s recommendation system itself.
Even so, he says that when he occasionally switches back to X, he usually leaves after a short time and returns to Firefly. His own mindset has changed as well. He says he is no longer interested in the endless daily arguments involving online opinion leaders. On one recent visit, he found that the day’s major dispute centered on whether a historical figure should be considered evil. His reaction, he writes, was to smile, log off, and continue with the rest of his day.
His takeaway is practical: reduce and limit the time and methods used for social media. He suggests trying aggregation clients such as Firefly or setting fixed rules, such as only checking X on Tuesdays. The goal is to keep access to information without being pulled into a vortex of discourse that keeps shaping the mind below the level of conscious awareness.
A 100-day experiment with airplane mode by default
The second experiment lasted 100 days. Buterin says he kept his phone in airplane mode by default the entire time. He still allowed Wi‑Fi connections, but turned off cellular data.
He says the test had two goals. The first was privacy. As long as a phone is connected to a cellular network, he argues, it functions as a continuous location-tracking device. Telecom operators have a poor record of protecting location data, he writes, and there have been cases where user location information was reportedly sold. In his view, that creates a real risk that movement data ends up with intelligence agencies or advertising companies.
Using an anonymous SIM card is not enough, he says, for two reasons:
- The phone’s IMEI can link together the device’s usage history and make it traceable.
- Movement patterns and internet habits leave many identifying signals, and AI makes it easier to infer who a person is from those signals.
For that reason, he says the best location-privacy solution is simply not to use cellular service. He adds that he also uses GrapheneOS and has long limited location permissions to only a very small number of apps.
The second goal was mental state. Spending long stretches outside without being continuously online, he writes, helped him rebuild the habit of living normally without relying on the internet at every moment. That left him calmer and gave him more room to think about things he considered more valuable and more positive.
Overall, he says, the experiment worked. During the 100 days, he made only one exception to briefly enable cellular data, and that was to receive a verification code so he could keep the phone number active. There was one other exception at the level of communications planning rather than phone settings: on a remote and somewhat risky hiking route, friends insisted that he carry an emergency communication device they had prepared and check in every few hours.
The biggest practical problem came in Singapore. Because he could not receive an SMS verification code, he was unable to log in to Grab, the local ride-hailing app. That reduced his dining options, he writes, and often forced him to take the subway even when a car ride would have been more convenient.
After the 100-day period ended, he says he occasionally turned cellular data back on, but in broad terms he kept the habit. His phone still stays in airplane mode by default.
Offline maps, local knowledge apps, and public Wi‑Fi
For navigation, Buterin says he uses Comaps.app, a map application designed to run locally first. After downloading open-source map data in advance, he can search routes and navigate offline.
In his experience, Comaps is weaker than Google Maps for urban information such as restaurants, shops, and opening hours. But it performs well on outdoor routes and includes many hiking paths that do not appear on Google Maps. It also makes mistakes. He recounts one case where it marked a trail that did not exist, and he ended up pushing through bushes for several hundred meters before getting back to the main route.
He says he has also started paying attention to local offline knowledge-base apps. In theory, he argues, a useful global information package does not have to be impossibly large. A compressed English Wikipedia without images is not that big, and while the full OpenStreetMap dataset is 2.2 TB, a filtered version that keeps only what a user needs could shrink to 1% to 10% of that size.
He mentions a related bounty this month launched by Statuette, to which he also donated. The goal is to build local mobile apps that let users query useful information conveniently while offline. He names two entries: FieldAtlas and Commonplace.
His assessment is that this category of software is close to becoming practical. Many participants, he says, are trying to run large models locally in a way that takes too long, with a single answer requiring anywhere from 1 to 22 minutes. The better approach, in his view, is to use semantic search first to pull relevant material from a local database immediately, then have a smaller local model turn that material into a readable answer. That gives a more workable balance between speed and quality.
He also acknowledges the limits. When traveling, especially on the way to meetings, there are still moments when live connectivity matters: confirming a location after arrival or checking the next appointment after one meeting ends. Once cellular data is off, public Wi‑Fi becomes the fallback, and that only works when the network does not require SMS verification.
Based on his travel experience, he offers a rough ranking of public Wi‑Fi friendliness:
- Taiwan, China: Taipei had many public hotspots. He says that while walking around and checking his phone, there was more than a 25% chance of finding a Wi‑Fi network that could be joined directly, often provided by the city or by chain convenience stores.
- Ukraine, the United States, Canada, and Singapore: Ukraine surprised him the most, with many open hotspots in a number of cities, some of which could be joined immediately without any confirmation. The United States, Canada, and Singapore had many nominally public Wi‑Fi options as well, though a large share required SMS verification or account login. Even so, he says, there were still plenty of open networks without those barriers.
- Austria and Hungary: He says the two former Austro-Hungarian areas looked similar in practice. There were many hotspots, but most public Wi‑Fi networks required SMS verification. Finding an open network without that step often meant trying several hotspots and walking farther.
- Switzerland: He places Switzerland last. Most public Wi‑Fi in city centers required SMS verification. On a 1.5-kilometer walk through a city center, he found only two exceptions: an unnamed shoe store and Starbucks. Starbucks offered 10 minutes of free access, after which continued use required an email verification code.
Moving away from web apps
Buterin says he has recently stopped relying on web interfaces for a growing number of tasks. For example, the last few times he updated an IPFS hash in ENS, he used a command-line tool written for him by AI. He also says that many privacy-protocol transfers are now done through Kohaku-CLI.
He argues that websites and web apps, especially crypto web apps, carry high risk for several reasons.
- First, every time a web app is opened, the full program is downloaded again from the server. That means the user has to trust the server on every visit. If the server is compromised on just one of those visits, the user can suffer losses. Installed software is somewhat better in this respect because the trust check is concentrated at the moment of download.
- Second, the web is structurally weak on privacy. A site can collect browser version, operating system, screen size, and many other signals. Mouse movements, click patterns, and similar behavioral traces can also be used to identify a user. On top of that, websites often send data and metadata to CDNs, analytics providers, and other third parties by default.
- Third, as with social media, commercial web apps are not necessarily designed around the user’s interests. He says many interfaces rely on dark patterns meant to capture attention, while what users actually need are designs that protect them and help them obtain information.
His alternative is a return to the Unix philosophy, combined with local AI to make the system easier to use. In practice, he says, that means users can tell a computer in natural language what they want done instead of typing complex shell commands manually.
He describes the Unix philosophy as building small, modular tools, each focused on doing one thing well, with text used as the common format for passing information between them. Users can then combine those tools freely. A web UI, by contrast, tends to bundle many functions into one opaque application built by a single team. Taken to the extreme, he says, that model becomes the super app: power concentrates in the hands of developers, internal logic grows harder to understand, and security risk expands with it.
Large language models, in his view, make the modular path viable again. Users can regain choice without everyone needing to become a programmer.
From material minimalism to digital minimalism
In the closing section, Buterin places digital minimalism in a longer technological arc. The material minimalism popular in the 2010s, he writes, was built on an earlier wave of computing and internet adoption. A single computer could replace many physical objects, making it possible to live simply with a laptop, a few sets of clothes, and basic charging and hygiene items. Before widespread internet access, that lifestyle was harder to maintain. Once the internet became common, the barrier dropped sharply.
He says the past decade of software progress and AI development is now producing a new wave: digital minimalism. In his framing, that means people can rely less on social media because AI can help them get information, while human connection can still happen in the oldest way possible, by meeting and talking in person. They can rely less on constant connectivity because many tasks can now be completed locally and offline. And they can rely less on web apps because command-line tools and custom interfaces can handle many operations, with AI helping users build and use those tools.
For Buterin, the timing matters. He argues that the weaknesses of social media, the weaknesses of the traditional web model, and the problems created by bad uses of AI are all becoming visible at once. That is why digital minimalism is becoming more necessary.
He closes with a personal recommendation rather than a universal formula. Different people will move toward digital minimalism in different ways, he writes, because their needs differ. Still, he suggests that anyone can start by choosing at least one direction in which to reduce digital information load, and then continue if the experience proves worthwhile.
