Here's a weird fact: if you want to read the very first website ever created—Tim Berners-Lee's 1991 page explaining the World Wide Web—you can't access it at its original URL. It's gone. The server shut down decades ago. CERN recreated it at a new location years later. Now imagine if that URL pointed to the content itself instead of a server location that could disappear. That's IPFS.

The InterPlanetary File System (IPFS) is a peer-to-peer protocol for storing and sharing files across a distributed network. Instead of asking "where is this file?" like HTTP does with URLs, IPFS asks "what is this file?" using cryptographic hashes. When you request a file on IPFS, the network finds copies from any peer that has it, not from a single centralized server. It's BitTorrent meets the web, designed to make the internet more resilient, censorship-resistant, and permanent.

The name isn't just sci-fi branding—IPFS was designed with interplanetary communication in mind. HTTP breaks down completely with the multi-minute latency of space communication. The same properties that make IPFS work for Mars colonies also solve problems we're facing on Earth: link rot, server failures, censorship, and the inefficiency of downloading identical files from distant servers when your neighbor already has them.

The problem IPFS solves

HTTP has served us since 1991, but its limitations are showing. Link rot is killing digital history—50% of Supreme Court opinion links no longer work. When servers shut down, every link breaks. Centralization creates single points of failure. When AWS has an outage, huge swaths of the internet go dark. Bandwidth is inefficient—when videos go viral, millions download identical data from the same distant servers instead of sharing from nearby peers. And you simply trust servers haven't been compromised, with no way to verify content matches what you expected.

IPFS addresses all of these by changing from "where" to "what."

How IPFS works

The core innovation is content addressing. Instead of URLs pointing to servers, IPFS uses Content Identifiers (CIDs)—cryptographic hashes of file contents. When you add a file, you get a CID like QmYwAPJzv5CZsnA625s3Xf2nemtYgPpHdWEz79ojWnPbdG. Change one byte, you get a different CID.

CIDs are self-verifying—you can cryptographically verify what you received matches what you asked for. CIDs are permanent because they're mathematically derived from content. Unlike HTTP URLs that can point to different content over time, an IPFS CID always refers to the exact same data.

Large files break into chunks, each with its own hash. This enables retrieving chunks from different peers simultaneously, speeding downloads. When you request a file, the network finds peers that have it. Your node retrieves chunks from multiple sources. Closer peers respond faster, making IPFS inherently CDN-like. After retrieving content, your node becomes a temporary host, creating organic content distribution without central coordination.

The persistence problem

Adding a file to IPFS doesn't guarantee it stays forever. If no one pins your file, it gets garbage collected. Think of IPFS like the web—you need a server to host webpages. With IPFS, you need someone to pin files permanently.

Popular content stays available because many people pin it. Valuable NFTs get pinned by owners and marketplaces. Random photos might disappear if you stop pinning them. This led to pinning services like Pinata, Infura, and Web3.Storage offering permanent IPFS storage for a fee, becoming the "web hosts" of IPFS.

The irony: decentralization technology relying on centralized services. But the crucial difference is interoperability—if Pinata disappears, your CID still works elsewhere. Switching IPFS hosts doesn't break references the way changing HTTP servers does.

Real-world use cases

NFT metadata storage is the killer app. Most NFTs store references, not artwork on-chain. HTTP URLs become dead links when companies shut down. IPFS solves this—NFTs point to CIDs that can be hosted by anyone. The difference between ipfs://QmXyz... and https://startup-api.com/token/42 is decades of persistence versus dependency on one company.

Decentralized websites use IPFS through services like Fleek. ENS domains point to IPFS content, creating censorship-resistant websites. dApps host frontends on IPFS. Lens Protocol uses IPFS for posts and profiles. Scientific datasets, open-source projects, and academic papers distributed via IPFS become harder to censor and easier to preserve.

The problems IPFS faces

Speed is inconsistent—retrieval times vary wildly. Popular content loads quickly; unpopular content might take forever. The persistence model confuses users. People assume "adding to IPFS" means permanent storage, then are shocked when content disappears.

Privacy is unsolved. IPFS content is public by default. Content addressing makes files hard to remove, creating legal risks for node operators who might cache illegal content. Most users access IPFS through HTTP gateways, reintroducing centralized infrastructure that can fail or be censored.

Changing content requires workarounds. Immutable CIDs are great for integrity, terrible when content needs updating. IPNS provides mutable pointers but is slow. Economic sustainability is unclear. Who pays for storage and bandwidth? Pinning services reintroduce hosting costs. Altruism won't support mainstream adoption.

The honest take

IPFS is genuinely innovative technology solving real problems. Content addressing is elegant. The peer-to-peer distribution model has clear benefits. Resilience to censorship and server failures matters for specific applications.

But IPFS isn't magic. It trades centralized simplicity for distributed complexity. Performance is inconsistent. Persistence requires ongoing effort or paid services. Privacy and moderation remain unsolved.

The real value proposition isn't "better HTTP"—it's "different HTTP for cases where different matters." If you need censorship resistance, content permanence, decentralized hosting, or verifiable integrity, IPFS provides capabilities centralized systems fundamentally can't match. If you just need to load websites quickly, HTTP is simpler and faster.

IPFS is infrastructure for a more decentralized internet. Whether we build that internet remains to be seen. But the technology works, real applications use it, and it's solving problems that matter. That's more than most blockchain-adjacent technologies can claim. And if we ever establish Mars colonies, they'll probably thank us for building internet infrastructure that doesn't assume millisecond latency and permanent server availability.

Further Reading:

• IPFS Documentation - Official technical documentation and implementation guides
• IPFS Whitepaper - Original protocol specification by Juan Benet
• Protocol Labs Research - Academic papers on IPFS, libp2p, and Filecoin
• The IPFS Gateway Problem - Addressing centralization in IPFS infrastructure
• NFT.Storage - Free IPFS pinning for NFT data
• Filecoin and IPFS - Economic layer for IPFS persistence