I was introduced to Ethereum some time in 2017. I remember the excitement and awe I felt for this technology—a network of untrusted machines capable of creating the world’s most reliable computer. But what fascinated me even more was the promise of a new web: Web3.
This new web would be owned by its users rather than controlled by large corporations. It was envisioned as perfectly democratic. Web3 represented the real web—the way the Internet was originally imagined.
Ethereum’s model at the time suggested that this new web would run on a massive virtual computer (the World Computer), with Swarm acting as its hard drive. At the time, Swarm was more of a mystical legend that developers whispered about, but no one knew anything about. The only certainty was that if it ever became a reality, it would bring significant change.
However, there was a small problem with this model—something that only became clear to me later: It was completely wrong.
Ethereum is not a World Computer, and Swarm is much more than just a hard drive.
Ethereum and Swarm
Ethereum’s virtual machine (EVM) is a Turing-complete execution environment, meaning that, in theory, it can run any program. However, in practice, its capabilities are quite limited.
One major limitation is that this machine can only interact with the blockchain—it reads data from the blockchain and can write only to the blockchain. This severely restricts its potential use cases. Another issue is that every operation must be executed and verified by all validators, making the system extremely computation-intensive. While the blockchain’s redundant storage and redundant computation provide high security, they also come at an enormous cost. Because of these constraints, Ethereum cannot function as a general-purpose World Computer.
Ethereum is much more like a database, where smart contracts act as stored procedures.
Stored procedures are highly useful for financial transactions (which is what blockchains were originally designed for), but they are not suitable as a general-purpose backend.
For a long time, Swarm seemed like the neglected child—overshadowed despite being a crucial component of the Web3 vision. Consider this: the web is primarily composed of content. If we want to build a new web, a fundamental question arises—where will this content be stored?
There have been other solutions, such as IPFS, which remains the most popular decentralized storage system today. However, IPFS operates on a fundamentally different principle—it focuses more on content discovery than actual storage.
From a user perspective, Swarm functions very similarly to Ethereum. Just as Ethereum validators stake tokens and receive rewards for their computational contributions, Swarm node operators stake tokens and are rewarded for providing storage capacity and bandwidth. Users pay gas fees for storage and bandwidth on Swarm, just as they pay gas fees for executing smart contracts and storing data on Ethereum. The two systems share a similar logic.
Additionally, Swarm nodes are identified using Ethereum addresses, and content stored on Swarm can be validated through smart contracts, enabling seamless integration between the two networks.
Beyond storage, Swarm also serves as a crypto-incentivized content delivery network (CDN). Thanks to single-owner chunks, it supports addressable mutable content storage. It even features a built-in messaging system, replacing Whisper—Ethereum’s long-promised but ultimately unrealized messaging protocol.
For those interested in a deeper dive into how Swarm works, you can read my articles on the topic:
Understanding the Ethereum Swarm Storage Scaling Mechanism
What’s the Difference Between IPFS and Ethereum Swarm?
But Where Is the World Computer?
We now have a storage solution that is far more than just a hard drive, yet we still lack a general-purpose World Computer. So how does this lead to Web3?
To answer this, let’s look at a simple use case: a decentralized Twitter.
The most popular decentralized Twitter alternative is Mastodon, which is based on ActivityPub. The Mastodon network consists of servers where users can register and read each other’s posts, regardless of which server they signed up on. A Mastodon identifier looks like an email address: user@server.
Users are free to choose their server, but since their identity is tied to a specific domain, switching servers later is problematic. A new server means a new domain name, which changes the user’s identifier, requiring them to redistribute it among their followers. To avoid this, the only surefire solution is to run a personal server—something unrealistic for the average user.
A slightly better approach is BlueSky’s AT Protocol. In this model, users are identified by domain names instead of email-like addresses, and their data can be freely moved between PDSs (Personal Data Servers).
However, Swarm takes a radically different approach—one that is completely silo-free. Here, data always remains with the user, eliminating the need for migration.
According to the Fair Data Society model built on Swarm, every user has their own FairDrive, which acts as their private partition within the global storage network. This is where they store their public feed, which they can share with anyone.
On a Swarm-based decentralized Twitter, following someone means incorporating their public feed into your own feed.
Since users want to access the system from mobile devices and discover interesting feeds beyond those they directly follow, feed aggregator servers are a useful addition. These aggregators provide a service: they aggregate personalized feeds (potentially using sophisticated AI algorithms) while abstracting away Swarm’s underlying mechanics, such as crypto payments.
This approach is very similar to BlueSky’s PDS model, with one crucial difference: Aggregator servers in Swarm are always stateless, as storage is handled by Swarm itself.
This makes it incredibly easy to add new aggregators to the system or switch between them—users don’t need to migrate any data. A user could dynamically select a different feed aggregator each time they refresh the feed or even request feeds from multiple aggregators and merge them locally.
With this setup, censorship and manipulation by aggregators become impossible. Any aggregator attempting to control or manipulate users will simply be ignored.
The World Computer is not Ethereum, but rather a decentralized service network, a collection of stateless servers performing various tasks on top of Swarm’s storage layer.
These servers can act as feed aggregators, run distributed AI models, or power decentralized sharing economy platforms like Uber or Airbnb.
Conclusion
While Ethereum plays a crucial role in building the new web—powering incentive mechanisms, DAOs, and more—it is an overstatement to call it a true World Computer.
The real World Computer consists of stateless servers performing diverse tasks, acting as a general-purpose backend. For the storage layer, Swarm is the ideal choice, providing a decentralized and censorship-resistant way to store and serve data while ensuring privacy protection for users.
Because Swarm can serve as the backbone of the new web, I believe it is perhaps an even more essential component of the Web3 ecosystem than Ethereum itself.