IPFS vs Arweave: Logic of Permanent Data and the Digital Sovereignty Unhack

You saved the link. You were sure of it — the investigation, the receipt, the post that proved you right. Six months later you click it and a blank page stares back. 404. The thing you needed to exist no longer does, and there is no one to ask, because the server that held it answered to someone else the whole time.

The short version: Your files live on machines you don’t control, so their survival is a decision made by other people. Two networks change that. IPFS addresses a file by its content — a cryptographic fingerprint, not a street address — so it can’t be quietly moved or renamed out from under you, though it stays available only while at least one node keeps seeding it. Arweave takes a one-time payment and commits to storing the file for 200+ years on a blockweave funded by a compounding endowment. Pin frequently-accessed data on IPFS; commit anything that must outlive you to Arweave; encrypt before either, because both are public by default.

Why your data disappears: the server-fragility problem

Here is the thing nobody tells you when you “save” something online: you didn’t save it. You borrowed it.

When you open `google.com/my-file`, you are not reaching a file — you are reaching a machine at an IP address that Google owns. The file exists at that location, on that server, under that account. Take down the server, restructure the URL, close the account, and the file is gone. This is the location-based internet, and it makes you a tenant in digital real estate you will never own a brick of.

The decay is measurable, not paranoid. Roughly 30% of referenced links on the public web eventually break. Academic papers cite sources that have evaporated. A journalist’s documentation vanishes the day a host honours a takedown notice. A deplatformed blog drops out of DNS and stops resolving anywhere. Your photos disappear when a company you’d half-forgotten about quietly shuts a service down.

You don’t lose your data because something went wrong. You lose it because the system worked exactly as designed — to keep the file’s existence in someone else’s hands.

What is content addressing, and how does IPFS fix the broken-URL problem?

Most coverage frames IPFS as “a decentralised file system” and leaves you no wiser. Here’s the thing no one tells you: the real reason a link dies isn’t that the file vanished — it’s that you were addressing the file by its location the whole time, and locations are owned. Move the question from where to what and the ownership problem dissolves.

IPFS (the InterPlanetary File System) stops asking where a file is and starts asking what it is. Upload a file and the network derives a CID — a Content Identifier — straight from the bytes of the file itself. Change one byte and the CID changes. So the address is no longer a pointer to a place; it is a proof of the content. That’s the reversal: the name of the file becomes the evidence that it hasn’t been tampered with.

Because the CID belongs to the content and not to any server, no single machine controls access. The file is replicated across nodes worldwide; if one drops offline, another still answers. There’s no domain to seize, no company to sue, no host to pressure. As long as one node carries the data, the original CID retrieves it — unchanged, or not at all.

The honest trade-off is availability. IPFS rewards files that get accessed; every request gets cached by the nodes serving it. Leave a file un-requested for months and it can be garbage-collected off nodes, going temporarily dark — unless you pin it to a service that keeps a node seeding it on purpose.

What is Arweave, and how does paying once buy permanent storage?

Arweave incidents the same problem from the opposite end. Instead of hoping volunteers keep seeding your file, it builds an economic reason for the network to hold it forever.

You upload a file and pay a single storage fee — typically $0.01–0.10 per GB, depending on network conditions. That fee drops into an endowment that uses compounding interest to fund storage costs in perpetuity. The data lives in a structure called a blockweave, spread across nodes that earn small rewards under a proof-of-access mechanism for demonstrating they’re genuinely storing the data and can retrieve it on demand. The protocol’s commitment is explicit: 200+ years of persistence.

For anything meant to outlast you — family records, legal archives, primary-source documentation — that’s not a feature, it’s the entire reason to use it.

The constraint is speed. Arweave retrieves more slowly than IPFS because files are scattered across a broad geographic network; access can run from milliseconds to a few seconds. For an archive you open occasionally, fine. For a production API serving thousands of requests a second, wrong tool.

IPFS vs Arweave: the head-to-head comparison

The two aren’t rivals — they’re layers. But the differences decide which file goes where.

| Feature | IPFS | Arweave | |—|—|—| | Addressing model | Content-based (CID) | Permanent transaction ID | | Storage cost | Free (relies on pinning services for reliability) | One-time payment (perpetual) | | Persistence guarantee | As long as ≥1 node seeds it | 200+ years (protocol commitment) | | Retrieval speed | Fast (cached by active nodes) | Slower (geographic distribution) | | Best for | Frequently-accessed files, websites, dynamic content | Historical records, legal docs, legacy archives | | Privacy risk | High (unencrypted files visible to all nodes) | High (permanent and public by default) |

The rule that falls out of this table: speed and churn go on IPFS, permanence and weight go on Arweave, and nothing personal goes on either without encryption first.

How to build a permanent data stack: a three-phase protocol

You don’t have to choose. Stack them, and each covers the other’s weakness.

Phase 1 — the pinning layer (IPFS reliability). Upload your files to IPFS, then hand the important ones to a pinning service that seeds them around the clock. Services like Pinata run dedicated nodes for roughly $5–20/month, so your IPFS files don’t go dark when casual nodes churn. This is your active archive.

Phase 2 — the endowment layer (Arweave permanence). Take your irreplaceable files — legal records, family archives, published research — and upload them to Arweave, paying the one-time fee. Tools like ArDrive make it a drag-and-drop: you drop files into a folder, pay once, and they’re preserved. This is your legacy layer.

Phase 3 — multiple gateways (redundancy). Reach your data through more than one door. For IPFS, use Cloudflare’s gateway, Protocol Labs’ public gateway, or your own local node. For Arweave, use the official gateway, Arweave.dev, or an alternative. One gateway falling over should never mean your data is unreachable.

Encrypting before you upload: the non-negotiable step

Here’s the mistake that turns a sovereignty upgrade into a leak: both IPFS and Arweave are public networks. Unencrypted files are visible to anyone, and on Arweave that exposure is permanent.

So encrypt locally first. Use Cryptomator — or any tool that builds an encrypted vault — to wrap sensitive files into a container, then upload the container. You open it with a password on your own machine; the network sees only encrypted noise. Even if someone grabs the ciphertext, the data stays yours.

For genuinely public material — published articles, documentation, open-source code — encryption isn’t needed. For anything personal — medical records, financial documents, private correspondence — encrypt first, every time.

A practical checklist for information sovereignty

Treat this as the operating discipline, not a one-off setup:

• Verify the hash first. Trust the CID or transaction ID, never the file name. If the hash matches your record, the file is authentic — that’s the whole guarantee.
• Distribute across networks. Don’t lean on one protocol. Keep critical files pinned on IPFS, committed to Arweave, and ideally on a local or self-hosted backup too.
• Encrypt anything personal. Before uploading private data to a public network, encrypt it with AES-256 or equivalent and keep the password in a password manager.
• Test retrieval on a schedule. Don’t assume access — prove it. Check Arweave links quarterly, pinned IPFS hashes monthly.
• Document your ownership. Store the CIDs, transaction IDs, and access instructions in a personal knowledge base you also back up across these networks.

The single move that does the most: pin one important file to IPFS this week and actually retrieve it from a second gateway. The habit of verifying access is the sovereignty — the tools are just plumbing.

Why permanent data is a form of resistance

Centralised platforms run on decay. Old posts age out, context evaporates, and institutions quietly benefit from collective amnesia. Permanent storage is where Digital Sovereignty stops being a slogan and becomes a property of your files. Storing data permanently and immutably is a refusal to feed that machine.

A journalist who archives her investigations on Arweave can’t be silenced by a hosting ban. A historian who pins primary sources to IPFS keeps them past any corporate purge. A family that stores its photos permanently outlasts the lifespan of every company that ever offered to “keep them safe.”

That isn’t doomsday thinking. It’s epistemic self-defence — you becoming the primary keeper of your own record.

Real example: Archive.org vs the permanent web

The Internet Archive (archive.org) has preserved roughly 735 billion web pages since 1996, and it is invaluable. But it is also a single organisation — exposed to legal pressure, funding cuts, and data loss. In 2024, the Archive was knocked offline by incidents and a data incident affecting tens of millions of user records; a determined court order or sustained outage can risk signalen access to the whole thing.

A document on Arweave behaves differently. It exists across thousands of nodes, with no central organisation to target — no entity to sue, no server to data incident, no company to pressure. The data is mathematically distributed and cryptographically verified. Which is exactly why some researchers and journalists now archive their work on both Archive.org and Arweave: one for discoverability, one for permanence.

Frequently asked questions

Is my data private on IPFS or Arweave?
Only if you encrypt it first. Both networks are public by default — unencrypted files are visible to anyone with network access, and on Arweave that visibility is permanent. Encryption before upload is non-negotiable for sensitive data.

How much does it actually cost to store data permanently on Arweave?
Roughly $0.01–0.10 per GB, paid once. A 1GB family photo archive might run $0.10–1.00; a 100MB document archive, around $0.001–0.01. These are estimates that move with network conditions. Compare it to $10/month of cloud storage indefinitely — about $1,200 over a decade, $12,000 over a century.

What happens if Arweave the company shuts down?
The protocol keeps running. Arweave is decentralised — thousands of independent nodes operate it, not a single company. If everyone working on it vanished tomorrow, the network would keep verifying storage proofs and serving data. No company can “switch off” the network any more than you can switch off the internet.

Can I modify or delete files after uploading to Arweave?
No — and that’s the point. Files are immutable. A new version gets a new transaction ID while the old one stays accessible forever. It prevents tampering, but it demands you be certain before you upload.

Do I need my own node to use these networks?
No. Public gateways and pinning services are enough to start. Running your own IPFS node is optional but good for privacy and resilience. Running an Arweave node needs serious storage (200+ GB today) and is mainly for advanced users and organisations.

You started reading because a link you trusted turned to a blank page, and something in you refused to accept that as normal. It isn’t normal — it’s a choice, and until now it was someone else’s. Pin one file this week. Commit one archive to Arweave. Retrieve them from a second gateway and watch them answer. You’ll feel the shift the moment they do: you stopped renting space in someone else’s memory and started keeping your own. You’re the architect of your archive now, not a tenant in it.