Hacker News new | past | comments | ask | show | jobs | submit login

Just commenting to double-down on the need for cryptographic timestamping - especially in the current era of generative AI.




How does that work exactly? Does it all still hinge on trusting a know Time Stamp Authority, or is there some way of time stamping in a trustless manner?


I'm so sad roughtime never got popular. It can be used to piggyback a "proof of known hash at time" mechanism, without blockchain waste.

https://www.imperialviolet.org/2016/09/19/roughtime.html

https://int08h.com/post/to-catch-a-lying-timeserver/

https://blog.cloudflare.com/roughtime/

https://news.ycombinator.com/item?id=12599705


You can publish the hash in some durable medium, like the classified section of a newspaper.

This proves you generated it before this time.

You can also include in the hash the close of the stock market and all the sports scores from the previous day. That proves you generated it after that time.


My mind immediately went to adversarial fixing of all sports games and the stock market in order to create a collision.

Sports sports are an interesting source of entropy.


If you are looking to prove that something happened after a certain timestamp, you can use a randomness beacon[0]. Every <interval>, the beacon outputs a long random number. Include the timestamped random number into your artifact.

You are relying upon the authority of the beacon to be random, but good practice is to utilize multiple independent beacons.

[0] https://csrc.nist.gov/projects/interoperable-randomness-beac...


This is the one thing blockchains are truly good for.


But there has to be economic incentives to maintain the data, and only Bitcoin can even to begin to make that claim, and even it is only 16 years old.

Still, Open Timestamps does exactly this, and had been running for over 8 years now.


Yeah it definitely could be, though you may similarly find yourself in a spot of trusting a limited number of nodes that guarantee the chain was never tampered with.


For something like this there’s ways to minimize how much you need to trust nodes such as regularly publishing hashes to 3rd parties like HN.

Not so useful if something was edited a few minutes after posting, but it makes it more difficult for a new administration to suddenly edit a bunch of old data.


> there’s ways to minimize how much you need to trust nodes such as regularly publishing hashes to 3rd parties like HN.

But you could do the same thing with any hashes, right? There is no need for a blockchain in the middle.


What happens as websites disappear? With a blockchain in 2090 you can point to a website post in 2060 as support that your hashes on data posted in 2030 are still valid. That’s useful when preventing people from rewriting history is the goal.

There’s also a size advantage. You can keep a diff on the archive for each hash being posted instead of the full index for every time you post a hash.


You make use of several independent authorities for each timestamped document.

The chance is exceedingly low that the PKI infrastructure of all the authorities becomes compromised.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: