A Quieter Internet

The Dream of Decentralization

In 2017, I was living in California. Just a few years earlier, I had dropped out of graduate school and plunged into the startup world. It was the early gold rush of big data, when people declared "data is the new oil."

The startup I helped build offered tools and services for companies to train and deploy machine learning models, and was later acquired by Oracle. I saw firsthand how businesses—big and small—used data to predict and shape our behavior. The power was staggering. And inside the software empire built on cloud services, patents, and licenses, I began to see how such power naturally concentrates and centralizes.

Before coming to the US, I grew up in China, and still read and write in Chinese. While I hopped from the startup battlefield to the corporate ladder, in a parallel world, I watched the collapse and reform of the Chinese content ecosystem—and the relentless yet often futile attempts to build something different.

Social media exploded, sucking up attention while producing little that deepened understanding or compassion. At the same time, great projects struggled. A digital media outlet(opens in a new tab) publishing outstanding investigative work was shut down due to the lack of a state-issued news license. An experimental social platform(opens in a new tab), designed to foster real discussions, saw incredible growth—until it was crushed by state censorship.

Again and again, I ran into the same problem: centralization. I realized that our information systems—imagined as humanity's neural network—are, in many ways, less resilient than spoken or written words. Too often, they have a single point of failure, and are easier to control than any communication technology before them. We built global information systems that turned out to be easier to censor than books.

The internet I grew up with was different. As a teenager, I explored the world through an endless sea of free—often pirated—content on BitTorrent, eMule, and VeryCD, China's version of Pirate Bay. Each magnet link was a portal—to philosophy lectures, obscure documentaries, and bootleg live albums. Before platforms took over, peer-to-peer protocols made up most of the internet's traffic. I've been fascinated with P2P networks ever since, though I couldn't imagine how it might work with copyright or business models.

When I discovered Ethereum and IPFS, and the wave of distributed networks that followed, it was like recalling an old dream—the same dream I saw in the early internet, even the same dream The Whole Earth Catalog sparked before the internet. For the first time, it felt possible to re-decentralize the web—to have a network formed around individuals, not platforms.

I began building prototypes using Ethereum for payments and indexing, and IPFS for storage. They were clunky, but the potential was real. I gathered writers and translators across continents to collaborate in a decentralized way. Our online magazine found an audience, confirming the demand among Chinese-speaking readers. The time felt right.

Then, in 2018, I met two experienced co-founders who knew media better than I did. Together, we built Matters Town(opens in a new tab)—a decentralized social and publishing network.

Building the Decentralized Alternative

We started with the core issues of centralization: the power to control information, the homogenization that erases niche content. So in the beginning, we focused on censorship resilience and direct financial support from readers to writers. Everything a writer published was stored as webpages on IPFS, maximizing accessibility across clients. Readers donated to writers using stable coins—a stamp of approval that wrote the CID (content identifier) on-chain and broadcast the content to the front page.

And it worked. Public events—the Hong Kong protests, COVID, Taiwan's elections—drove waves of writers and readers to this digital refuge. It became one of the few places where Chinese speakers across geographies, ideologies, and identities could still talk. Usage exceeded expectations. With millions of monthly page views, we had to refactor our servers in a hurry.

Even though we prioritized decentralization, servers remained indispensable. Features like comments, search, and recommendations rely on them, and aren't available on clients using only Ethereum and IPFS.

Many decentralized projects face the same dilemma, and most adopt a path of "progressive decentralization." But to us, it felt more like a balance we need to keep—between decentralized infrastructure and centralized services.

During many of the attacks we went through — DDoS, DNS pollution, malicious DMCA takedowns—our services would go down, but the decentralized clients stayed online. These moments keep proving the utility of decentralized networks, and they give people confidence in building on top of them.

Still, day-to-day, most users rely on the official web app for better usability. Decentralization offers long-term resilience, but in the short term, convenience wins.

While I was frustrated that more people didn't care about decentralization, I was surprised by how many were seeking alternative spaces to write, read, and connect. We focused on Chinese-speaking communities, but users came from all over the world. Soon we had content in over 40 languages. At one point, more visitors used Hindi than Chinese. And when the Russia–Ukraine war began, Ukrainian writers shared their experiences with a global audience on Matters Town.

The user base surprised us—and so did the behavior. Donations and subscriptions didn't fit most of the content, which sparked discussion and connection more than financial support. As we began serving the general population, we encountered the same tension faced by platforms like Facebook and Twitter: the attention economy. But it's inseparable from surveillance and centralization—the very forces we set out to resist.

It felt like we were back where we started. Was there a way to break the cycle?

Out of necessity, we began confronting the outdated and messy ecosystem of attention. And then it clicked. All the time we've been trying to decentralize content alone, but decentralizing incentives might just be more effective. Attention economy doesn't have to be something we fight against, but something we might have a chance to fix.

The Fragmentation of Social Media

After we began building Matters, the problems that drove us began to surface more visibly in the US and the English-speaking world.

For many, 2020 was a turning point. Elon Musk brought a bathroom sink to Twitter HQ, flipped the platform's political leaning, and turned it into his personal megaphone. People fled—to Mastodon, Threads, Bluesky. In 2024, the US moved to ban TikTok. Users ran to RedNote—only for TikTok to return, just with different censorship.

Trust in institutions and institutional media hit historic lows. But the alternatives—algorithms and influencers—weren't much better. Fake news flooded timelines. The powerful fought for control.

Our political debates, worldviews, and collective future—all increasingly shaped by opaque algorithms, from social media recommendations to generative AI. We barely understand them, let alone control them. And they are rapidly centralizing.

But the fragmentation of social media may open a door for decentralization. Decentralized protocols such as ActivityPub (Mastodon, Threads) and AT Protocol (Bluesky) have entered the mainstream. These federated systems offer hope: servers for convenience, decentralization for resilience.

Each successful network eventually faces monetization, and risks falling into the same trap: tracking and manipulating users in the attention economy. So if we're decentralizing storage and distribution, we must also decentralize economics.

So let's reimagine online advertisements. Instead of constantly monitoring and targeting, they can be static and passive, more like billboards in physical cities—part of the landscape. At best, a sign of prosperity; at worst, a benign nuisance.

Front pages on Matters, Discourse servers, personal sites with WordPress or Substack—these are the busy streets of the digital world. Many already rent them out despite the monopoly and opacity, delegating the role of surveillance. In the past few years, people minted countless digital assets as NFTs, but these "billboards" have existed for decades without standards. Perhaps it's time they became NFTs, too.

Tradable, on-chain billboards let us auction attention without measuring or commodifying the user. The winning bidder declares a price and gains temporary ownership; in the next period, they receive the payment. The billboard's market value determines profit or loss, discouraging exploitation from the owner. A self-imposed tax, tied to the price, becomes revenue for the billboard owner and discourages inflated pricing¹.

The price discovery of billboards does not rely on measuring attention, so applications can decide what user data, if any, is exposed. Not all applications are willing to protect privacy at the expense of personalized targeting, but with an alternative to selling attention directly, now they do have a choice, and users will in turn have a choice as well.

With on-chain revenue, we can further push the attention economy toward transparency. On Matters Town, for example, we use this as a matching fund for continuous quadratic funding to support creators. Other community-driven projects may benefit too.

What's interesting is that, radical as it sounds, this model works with the existing ad ecosystem. If allowed, traditional ad networks can bid on these billboards and use them as part of their inventory. But when someone else—an individual, community, or advertiser—values the billboard more, they can simply outbid. That extra income supports the application and its community. On-chain trading lowers the barrier to entry, creating a more competitive and diverse market.

It's a promising idea—but like the ad industry itself, it's complex. During my Next Billion Fellowship, with the help of designers and industry veterans, we began to navigate this terrain. We refined our product positioning(opens in a new tab) to appeal to both Web3-native advertisers and community spaces, and we're now integrating Google AdSense, the largest traditional ad player.

We're only beginning to bridge this entrenched industry into a decentralized future. We don't know how big that future will be—but it's an exciting start.

Finding a Quiet Space for Thought

In November 2024, I was in New York. The presidential election had just ended. People walked the streets with hesitation, greetings laced with caution, conversations punctuated by frustration. Trump had won in a landslide. Everyone was still processing what that meant—for themselves, and for the country.

To me, it marked a global shift—nationalism rising, globalization retreating. Local production and consumption may work for physical goods. But in the digital world I've spent years building, I still believe in global connection. That belief felt more fragile than ever—like the bleak wind of the New York winter.

A week later, I was in Bangkok for Devcon. People from across the world—young, brilliant, curious—gathered in the warm tropical air, exploring how technology might drive better cooperation, institutions, incentives.

At a workshop I hosted, a dozen participants and I explored how pattern language—an idea originating in architecture to enable people to build their own environment—could guide the design of digital spaces. We discussed patterns that foster connection and community. I saw sparks in people's eyes as they shared stories. Those are precious moments for me, in a time of aggression and mistrust.

At another workshop, I joined entrepreneurs, researchers, civil servants, and organizers to discuss public goods. During the icebreaker, we formed a circle spanning six continents—from Northern Europe to South America. We talked about problems I once thought unsolvable—like using decentralized identifiers (DIDs) in government ID systems to allow interoperability without surveillance. Suddenly, with people from different countries and walks of life, these ideas felt within reach.

It reminded me of the poster by John Lennon and Yoko Ono: "War is over! If you want it." And now, it seems many people do want a decentralized future. Jet-lagged in the tropical sun, it felt surreal.

In 2014, philosopher Byung-Chul Han wrote:

In a way, I've been looking for a quieter medium for thought. Thinking has always been collective—now it's a collaboration between people and machines. Over the years, I've learned a bit about what such ecosystems should look like. But I've also learned it's beyond the understanding of any one person.

So the only way forward is to find the right people—those who care, and who might figure out a better way. The solution to digital failures won't be found in better algorithms, but in real-world communities. That old dream of a web built by people not platforms, might still be alive after all.

Footnotes

1. This design, referred to as "Harberger Tax" or "Partial Common Ownership", was recently popularized in Radical Markets by Posner & Weyl. Read more in Chapter 1 of Radical Markets. ↩