You know that feeling when you predicted something — a stock crash, a product pivot, a scientific breakthrough — and nobody believes you? You said it at dinner. You mentioned it in a group chat. But when it happens, everyone acts surprised, and you’re left with nothing but the bitter taste of “I told you so” with no receipts.
We’ve all been there. And the solutions people have built for this problem are, frankly, insulting.
“Just post it on Twitter.” Great, so everyone can steal it.
“Use a blockchain timestamp.” Okay, but now the content is public immediately — defeating the entire purpose of keeping a secret.
“Hash it and publish the hash.” Sure, if you want to do five steps of manual cryptography every time you have a thought worth preserving.
A developer named Seb Mellen built something that cuts through all of this. It’s called prove-yourself, and the Hacker News crowd immediately dismissed it as “OpenTimestamps with extra steps.”
They’re wrong. They’re spectacularly, fundamentally wrong. And the reason they’re wrong reveals something important about how we think about trust.
Seb’s original startup was built on a naive but beautiful idea: get scientists to timestamp their work on a blockchain, couple it with a protective license, and stop IP theft. A golden age of science, ushered in by cryptography. It was optimistic. It didn’t work.
But buried in that failed startup was a mechanism that actually solves something real.
Here’s what prove-yourself does: you write a statement — a prediction, a claim, a hash of a file. You sign it with a local cryptographic key. You tie that key to your identity on GitHub, Twitter, wherever. Then you encrypt the whole thing to a future round of drand, a decentralized randomness beacon. You anchor the commitment in an Ethereum transaction and on IPFS, so there’s a public record it existed by that date.
After the reveal time, anyone can decrypt and verify it. Without you. Without any server. Without any central authority.
The blockchain timestamp isn’t the innovation. The timelock encryption is.
Here’s why that distinction matters. With a traditional blockchain timestamp, you’re proving that something existed at a certain time. But the content is either public — defeating the purpose — or you’re trusting that whoever holds the encrypted version won’t manipulate it before the reveal.
With timelock encryption, the decryption key literally does not exist until the chosen future round of drand fires. Not “it exists but is hidden.” Not “it exists but is locked behind a server.” It does not exist. The key is generated by a decentralized network at a predetermined time, and nobody — not the creator, not the server operator, not a court order — can produce it early.
This is the difference between hiding a key under a mat and having the key not exist until the exact moment you need it.
That’s not “OpenTimestamps with extra steps.” That’s a fundamentally different trust model.
Think about what this actually enables. A scientist who suspects they’ve found something but isn’t ready to publish. A trader who has a thesis about a market move but doesn’t want to move the market. A startup founder who wants to prove they had an idea before a competitor launched it.
All of them can now make a cryptographic commitment that’s verifiable, time-stamped, and completely secret until they choose to reveal it — and the reveal happens whether or not they’re still around to press the button.
The HN commenter who dismissed this as a rehash is looking at the surface — blockchain, timestamp, verification — and missing the architecture underneath. It’s like looking at a car and saying “this is just a carriage with an engine.” Technically adjacent. Fundamentally misleading.
The real question isn’t whether this is novel. The real question is why we’ve accepted for so long that proving you knew something requires either trusting a third party or revealing it prematurely.
Seb’s failed startup was built on the premise that scientists would adopt blockchain timestamps to protect their IP. They didn’t. But the tool that emerged from that failure does something more subtle and more useful than IP protection. It gives anyone — not just scientists — a way to make a claim with cryptographic certainty, on their own terms, without asking permission.
No company owns it. No server needs to stay up. No authority needs to vouch for you.
The next time you have a prediction you’re afraid to share, or an idea you’re afraid to lose credit for, remember this: the technology to prove you were right without showing your hand already exists. It just took a failed startup and a developer who refused to let the good parts die to bring it to the surface.
Trust isn’t about finding someone honest enough to hold your secret. It’s about building systems where no one needs to.
FAQ
Q: Isn't this just OpenTimestamps with extra steps?
A: No. OpenTimestamps proves something existed at a time — but the content is either public or you're trusting whoever holds the encrypted version. Timelock encryption means the decryption key mathematically does not exist until a future round of drand fires. Nobody can reveal it early, including the creator. Different trust model entirely.
Q: What would I actually use this for?
A: Any situation where you need to prove you knew something before a specific date but can't reveal it yet: scientific claims before publication, market predictions, priority disputes over ideas, whistleblower documentation. You get cryptographic proof of prior knowledge without exposing the knowledge itself.
Q: Doesn't relying on drand and Ethereum just replace one trust assumption with another?
A: Technically yes — you're trusting the drand beacon and Ethereum consensus. But you're trading trust in a single operator (who could be coerced or compromised) for trust in decentralized networks that would need to be collectively corrupted. It's not trustless in the philosophical sense, but it's dramatically harder to manipulate than any server-based alternative.