← All archives

How this archive works

This site keeps exact copies of web pages, frozen at the moment they were captured — and attaches mathematical proof that each copy existed at that moment and has not been altered since. Here's what that means and why you can trust it, in plain language.

The short version: every snapshot is fingerprinted, and that fingerprint is permanently written into public records nobody controls — including the Bitcoin blockchain. If even one letter of a snapshot were changed later, the fingerprint would no longer match. So when the proof checks out, the snapshot is exactly what existed on the day it was stamped.

1. What gets saved

When a page is archived, we don't just save a picture of it. A real browser loads the page, and we keep everything: the text, every image, every video, the fonts and styling — plus the behind-the-scenes evidence of where it all came from:

The raw network traffic (a WARC file — the same format the Library of Congress uses): every request our archiver sent and every reply the website's servers gave, including technical headers that are very hard to fake convincingly.
The site's security certificate — the credential the real website presented, issued to its real owner (e.g. "Red Hat, Inc." or a news network's parent company).
Address-book records (DNS) showing which servers the site pointed to that day — typically big content networks like Akamai or Cloudflare.
Video technical metadata — broadcast videos carry traces of the editing and encoding tools that made them; we preserve those untouched.

2. The fingerprint (what a "hash" is)

Every file in a snapshot is run through a mathematical function called SHA-256, which produces a 64-character fingerprint. Two properties make it special:

The same file always produces the same fingerprint.
Change anything — one letter, one pixel — and the fingerprint comes out completely different. Nobody has ever found two different files with the same SHA-256 fingerprint, despite decades of trying and enormous prize money for doing so.

Think of it like a wax seal on an envelope — except this seal is computed from the letter itself. You can't open the envelope and change the letter without the seal visibly breaking, and you can't forge a matching seal for a different letter.

All the individual fingerprints are collected into one list (the manifest), and that list gets its own fingerprint: the root hash. That single line of text represents the entire snapshot — hundreds of megabytes reduced to one unforgeable stamp.

3. Proving when — the Bitcoin anchor

"OK, the fingerprint proves nothing changed. But how do I know when the snapshot was really made? Couldn't the archive owner fake the date?"

No — because the root hash is written into the Bitcoin blockchain, a public ledger maintained by thousands of independent computers around the world. Bitcoin's entire multi-hundred-billion-dollar security model depends on nobody being able to rewrite its history. Every entry is bound to a block with a globally agreed date.

We use a free, open system called OpenTimestamps to do this. Once the root hash is anchored, claiming the snapshot was made later than stamped is equivalent to claiming Bitcoin's history was rewritten — something no government, company, or individual has ever managed.

It's like publishing a tiny classified ad containing your document's seal in every newspaper on Earth on the same day. To fake the date afterwards, you'd have to swap out every archived copy of every newspaper.

4. A second, independent clock — the timestamp authority

Belt and suspenders: the same root hash is also sent to a timestamp authority — an independent service whose sole job is to digitally sign "I saw this fingerprint at this exact time" (an international standard called RFC 3161, used in legal and financial systems). Their signature can be checked offline, forever, with free standard tools. For a forgery, you'd need to steal that authority's private signing keys and rewrite Bitcoin.

5. Independent witnesses

When possible, each capture also asks unrelated public archives — the Internet Archive's Wayback Machine and archive.today — to take their own copy of the same page at the same time. Those organizations have no connection to this server. A skeptic would have to believe this archive and the Internet Archive were compromised simultaneously, in the same way, on the same day.

6. Why "this is AI-generated" doesn't hold up

Suppose that years from now, someone claims an article or video here was fabricated with AI. The proofs answer directly:

The timeline defeats it. The proof pins the content to its stamp date. Content stamped in 2026 cannot have been made with tools that only exist in 2030. And a 2019 news broadcast, corroborated by third-party archives, predates convincing generative video entirely.
The forensics defeat it. A faked page would need a fabricated-but- coherent WARC file (hundreds of server responses with realistic CDN headers), a matching corporate TLS certificate, plausible DNS records, and broadcast-toolchain metadata inside the videos — all consistent with each other and with how that company's infrastructure actually behaved on that date.
The witnesses defeat it. Independent archives hold their own copies, on their own servers, with their own timestamps.

7. Check it yourself

Every snapshot has a certificate page (the ✓ proof link next to each version) showing its root hash, both timestamp proofs, and the network evidence. From there you can download a proof bundle — a small zip containing everything needed to verify the snapshot with free, open-source tools, with step-by-step instructions included. No account, no permission, and no trust in this website required: the math checks out on your own computer, or it doesn't.

The one-sentence summary

These pages are preserved exactly as they appeared, with their fingerprints locked into public records that nobody — including the person running this archive — can rewrite.