Vitalik's Latest Thoughts: We Need a Digital Minimalism Movement

Written by: Vitalik Buterin

Compiled by: Luffy, Foresight News

For a long time, I have adhered to minimalism in real life: the fewer possessions I have, the better, using fewer things to meet more needs. Four years ago, I wrote a blog post about everything I packed into a 40-liter backpack when I traveled.

However, the philosophy of material minimalism emerged in the 2000s to 2010s to address the issues of that era and to seize the opportunities of the time. I still find it valuable, and of course, I see some opposing viewpoints: having the right material possessions can help you better master your autonomy in life. But I also believe that in the present, we need a brand new form of minimalism: digital minimalism.

The core of digital minimalism is consciously controlling how we use digital technology. Every time we go online or scroll through apps, it should be a choice we actively make, rather than being unconsciously swept along by algorithms and passively indulging in it. In this article, I will share some of my attempts in this regard.

Social Media Minimalism

Twitter (X) has been filled with a lot of negative content, and the situation has continued to deteriorate in recent years. This platform is indeed very attention-grabbing, but much of the content lacks informational value; the way it attracts you often only worsens your mood. Around late 2024 to early 2025, this feeling peaked for me. A lot happened during that time, and if you were immersed in X for long periods, you would likely remember one thing: a sudden surge of comments praising Hitler on the platform.

Around the same time, the crypto Twitter community also fell into an atmosphere called "financial nihilism." In simple terms, it means: the ultimate destination of blockchain is gambling, and that’s just how it is; if you disagree and hope the industry can move in a better direction, then you are just a self-righteous elitist.

I have always had ideological disagreements with both of these trends. But beyond that, I also began to seriously realize: while discussions on the platform are easy to get drawn into, they do not enhance understanding; they only consume emotions and make people uncomfortable. So for most of this year, I conducted an experiment to maintain distance from this chaos. I did not deactivate my X account or completely leave the platform, but I stopped using the official X client to browse posts. Instead, I used Firefly.social to share content.

Firefly is an aggregation client that can connect to X, Bluesky, and decentralized social platforms based on the Ethereum ecosystem like Farcaster and Lens. You can browse content from these four platforms in one interface, and a post can be automatically synced across all platforms.

My initial goal in doing this was to support decentralized social media. I wanted to continue engaging on X but to disperse the network effects, preventing all traffic from being locked into a single platform. However, after using Firefly for a few months, I found it had another effect: it was much more pleasant to scroll through.

I suspect the key reason is that while Firefly allows you to read and post, it lacks X's core element—the recommendation algorithm.

I recalled that when I used the official X client, it was easy to fall into this trap: scrolling past a tweet supporting Hitler → clicking in, looking at the quotes and comments, just like someone unable to look away from a car crash. The algorithm would then determine that this user is interested in this type of content and tag it. Thus, you would see ten similar tweets and could not help but click in to observe. The algorithm would continue to push more... You understand this logic; in the end, it continuously feeds you similar content.

Objectively speaking, over the next year and a half, the trend of praising Hitler on the platform has significantly faded. Part of the reason is that once people realize such extreme ideologies are both evil and foolish, their popularity naturally wanes; another part is that X's recommendation algorithm itself has indeed been optimized.

Even so, when I occasionally switch back to X, I find that after a short while, I exit and return to Firefly. My mindset has also changed: I am no longer interested in the various debates that "opinion leaders" on the platform engage in day after day. Recently, when I logged back in, I found that the hot topic of the day was whether a certain historical figure was evil. Seeing this topic felt like an adult unexpectedly encountering a childhood snack, cartoon, or game: something that was once captivating now completely fails to pique my interest. I chuckled and immediately exited, continuing with my day.

My conclusion: I recommend trying this method, finding ways to reduce and limit your time and manner of scrolling through social media, using aggregation clients like Firefly, or setting a rule to only browse X on Tuesdays, finding a method that suits you. Keep the channels for obtaining information, but do not be swept away by the whirlpool of public opinion—algorithms will continue to influence you subconsciously, and often you may not even realize it.

Cellular Data Minimalism

I conducted a second experiment for 100 days: keeping my phone in airplane mode by default. I could connect to WiFi but turned off cellular data.

I had two goals in doing this: First, to protect privacy. As long as the phone is connected to the cellular network, it is an uninterrupted tracking device. Carriers have historically struggled to properly safeguard location data and have even been exposed for selling user location information. Your movement trajectory is likely to end up in the hands of intelligence agencies or advertising companies.

Using an anonymous SIM card is not enough for two reasons:

• The phone's IMEI serial number can link all usage records of this device.
• Your mobile trajectory and internet habits leave a lot of characteristics, and with AI, it is easy to reverse-engineer who you are.

So the best solution for location privacy is simply not to use cellular networks. Of course, I am also using the GrapheneOS system, granting location permissions to only a few apps, which I have been adhering to.

Second, to improve mental state. Being offline for extended periods while out helps me re-establish the habit of "living normally without relying on the internet," leading to a calmer mind and creating space to think about more valuable and positive things.

Overall, this experiment was a success. During these 100 days, I only made exceptions to turn on cellular data once: briefly to receive a verification code, just to keep this SIM card active. Another exception was for mental reasons: I went on a remote and somewhat dangerous hiking route, and my friends insisted I take the emergency communication device they prepared, reporting safety every few hours.

The biggest trouble occurred in Singapore: I could not receive SMS verification codes, so I could not log into Grab (the local ride-hailing app). As a result, my dining options decreased, and many times, even though taking a taxi would have been more convenient, I could only take the subway.

After the 100-day experiment ended, I occasionally turned on cellular data, but generally continued this habit: keeping my phone in airplane mode by default.

For navigation, I use Comaps.app, which is a map software that prioritizes local operation: after downloading open-source map data in advance, route checking and navigation can all be done offline. My feeling is that Comaps is not as good as Google Maps for urban life information like finding restaurants, shops, and business hours; however, it is very suitable for outdoor routes, including many hiking paths that Google Maps does not have. Of course, it can also make mistakes; it occasionally marks a path that does not exist at all, and once I followed its guidance, struggling through bushes for several hundred meters before returning to the main road.

I have also started paying attention to a type of local offline knowledge base app. Theoretically, a sufficient global information package does not take up much space. The English Wikipedia without images is not large when compressed; the complete OpenStreetMap map package is 2.2TB, but after filtering and retaining only the necessary content, the size can be compressed to 1% to 10% of the original.

This month, there is a related bounty activity (initiated by Statuette, and I also participated in the donation), aimed at developing local mobile apps that allow people to conveniently query useful information offline. Here are two of the participating projects: FieldAtlas and Commonplace.

My feeling is that we are currently at a critical point where this type of software is becoming practical. Many participants' ideas involve running large models locally, with single queries taking a long time to think through, and answering a question taking anywhere from 1 to 22 minutes, which is too slow. The best solutions are those that first use semantic search to immediately pull relevant information from the local database, and then use a small model to organize it into coherent answers, achieving a suitable speed and effect.

Of course, when going out, especially when meeting someone, there are times when you still need to be online: when you arrive at a location, you need to contact the other party to confirm the position; after a meeting ends, you need to check the arrangements for the next one. After turning off cellular data, I could only rely on public WiFi, and this WiFi cannot require SMS verification to log in.

I paid special attention to the friendliness of public WiFi in various places during my travels, simply categorizing them:

Taiwan, China. Taipei has a lot of public hotspots; walking down the street, there is about a 25% chance of finding WiFi that can be connected directly, mostly provided by the government or deployed by chain convenience stores.

Ukraine, the United States, Canada, Singapore. Ukraine was the most surprising to me; many cities have open hotspots, some of which can connect just by clicking without any confirmation. You might wonder, considering the local security situation, is this appropriate? I guess their control focus is on frontline areas and drone communications flying into cities, and ordinary public WiFi on the street is not a priority. The situations in the United States, Canada, and Singapore are similar: there are many nominal public WiFi networks, but a large portion requires SMS verification or account login. Even so, there are still many open hotspots without barriers.

Austria, Hungary. These two former Austro-Hungarian regions are similar; there are quite a few hotspots, but the vast majority of public WiFi requires SMS verification to log in. Finding an open network without verification often requires switching between several hotspots and walking a bit further.

Switzerland. Although people generally think Switzerland values privacy, it ranks last here. Most public WiFi in the city center requires SMS verification. I walked 1.5 kilometers in the city center and only found two exceptions: an unknown shoe store and Starbucks—Starbucks offers free access for 10 minutes, but to continue using it, you need to provide an email verification code.

Website Minimalism

Recently, I have increasingly stopped relying on web interfaces for various tasks. The last few times I updated the IPFS hash on ENS, I used a command-line tool that AI helped me write. Many privacy protocol transfers I now complete through the Kohaku-CLI command-line tool.

Web pages and web applications, especially those related to crypto, carry high risks for several reasons: Every time you open a web application, you have to download the entire program from the server again. With each visit, you must fully trust the server. If the server is compromised during any of your visits, you could suffer losses. Traditional downloaded software is somewhat better: you only need to confirm that the server was not tampered with and did not contain malicious code at the moment of download; one-time trust is sufficient.

Web pages are inherently unfavorable for privacy protection. When you visit a website, it can obtain your browser version, operating system, screen size; when you operate on the page, mouse movements, click habits, and a host of other characteristics can easily identify you. Websites also default to sending your data and metadata to a host of third parties like CDNs and data analytics service providers.

Just like social media, web applications developed by commercial companies are not necessarily for your benefit. Many interface designs are dark patterns (manipulative designs) aimed at capturing user attention; what we want is positive design that protects users and helps them obtain information.

So, is there an alternative? Essentially, it is a return to the Unix philosophy, made easier with local AI: you directly tell the computer what to do in natural language, without having to type complex shell commands.

The core of the Unix philosophy is simple: create small, modular tools, each tool only does one thing well, and they communicate using text as a universal format. Users can freely combine and configure these tools. In contrast, web UIs bundle a lot of functions into an opaque application developed by a single team. Taken to the extreme, this results in super apps, where power is highly concentrated in the hands of developers, and the internal logic is complex and difficult to understand, amplifying security risks. With the advent of large models, we can revisit this modular path, returning choice to users without requiring everyone to become a programmer.

Conclusion

The material minimalism that was popular in the 2010s relied on a wave of new technology at that time: computers and the internet. A computer can replace a large number of physical items. Thus, you only need a laptop, a few sets of clothes, charging and toiletries to live a simple and free life. Before the internet became widespread, material minimalism was not impossible, but maintaining it was much more troublesome; after the internet emerged, the threshold was significantly lowered.

In the past decade, advancements in software capabilities and AI technology have given rise to a new wave of minimalism: digital minimalism. We no longer rely so heavily on social media because we can obtain information through AI; connections between people can still be made through the most traditional means—meeting and chatting offline. We no longer rely on being online all the time; many tasks can be completed offline. We also no longer depend on web applications; many operations can be done using command-line tools or customized interfaces, and AI can help you build and use these tools.

At this very juncture, the shortcomings of social media, traditional web models, and the problems caused by the misuse of AI have all become glaringly apparent, making digital minimalism increasingly necessary.

Everyone's path to digital minimalism is different, and everyone's needs vary. But I strongly recommend that everyone at least choose one direction to try to reduce their digital information burden; if the experience is good, continue to push forward.