This chapter confronts a quiet embarrassment at the heart of Web3. The preceding chapters described a machinery designed to remove trusted middlemen: a public ledger no single party owns (Chapter 11, The Blockchain), programs that run by fixed rules (Chapter 13, Ethereum and Smart Contracts), tokens that record value (Chapter 16, Tokens), and even artworks claimed to be "owned" on-chain (Chapter 17, Non-Fungible Tokens). Yet most of what users actually see and touch in a Web3 application — the web page, the images, the videos, the bulky data files — still sits on ordinary corporate servers rented from a handful of cloud companies. A movement that promises an internet no one can switch off has, in practice, been running much of itself on infrastructure that any one of a few firms could switch off tomorrow. This chapter examines the technologies that aim to close that gap: decentralized storage and decentralized computation. It explains a deceptively simple idea — addressing data by what it is rather than where it is — and then asks honestly whether these systems can deliver the "unstoppable application" their advocates promise, or whether they remain, for now, the least finished room in the Web3 house.
In plain terms
a blockchain is like a courthouse registry of property deeds. It is the right place to record that you own a house, but you would never try to cram the house itself into the filing cabinet. The deed points to the house; it is not the house.
location addressing is asking for a book by which shelf it sits on, so the request fails the moment the book is moved. Content addressing is asking for a book by an unforgeable description of its exact contents, so anyone holding that book can hand it to you — and you can be certain it is the right book before you read a word.
Filecoin is a global Airbnb for hard-drive space. People with spare storage rent it out; people with data to keep rent it; and a public market sets the price. The blockchain is the booking platform and the escrow agent that makes strangers trustworthy to one another.
Arweave works like a university endowment or a "perpetual care" fund at a cemetery. You make one large gift; it is invested; and the network spends only the proceeds, a little each year, to maintain the data indefinitely — in principle without ever exhausting the principal.
the goal is an app that behaves less like a shop with an owner who can lock the doors and more like a song that has been sung by enough people that no authority can ever fully un-sing it. Once released, it would keep running as long as anyone, anywhere, was willing to host its pieces — because no part of it depends on a particular company's server.
decentralized storage and computation are real, working, and clever — and also slower, fiddlier, frequently pricier, and far less used than the cloud they hope to replace. The vision of the unstoppable application is coherent and partly demonstrated; it is not yet the way most things are actually built.
What the chapter covers
- Why storage is the part everyone forgets
- Content addressing: asking for data by what it is, not where it is
- Persistence and the problem of incentives
- From storage to computation: the rest of the stack
- The vision of the unstoppable application
- The honest limits
- Synthesis
Terms explained in this chapter
The full chapter is in the book.
Buy the full book