1
$\begingroup$

It is possible to hash public info (e.g. new york times headline) into an object to verify that it existed after a given date. Is it possible to do the obverse, namely verify that an object must have existed before a given date?

$\endgroup$
4
  • 1
    $\begingroup$ Trusted Timestamping Authorities can be used assert existence of an object at a specific point in time, which should solve your problem? $\endgroup$
    – SEJPM
    May 17, 2017 at 11:54
  • $\begingroup$ The first sentence in the question seems to suggest that if we hash a NYT headline, the hash did not exist before the publication date. It could well be that there are already precedents of the contrary; there sure is for newspapers that make short headlines; it is easy to make a hash-of-plausible-future-headlines generator good enough to hit one; and influencing the NYT to get a particular headline does not seem infeasible. Hashing the date with the headlne only partially solves the issue. I agree on SEJPM about how the second sentence can be answered positively. $\endgroup$
    – fgrieu
    May 17, 2017 at 12:35
  • 1
    $\begingroup$ The question is unclear: Do you mean that you want to use some sort of algorithm today to prove that e.g. some object existed in 1985? Or do you want to create some verifiable token now and be able to check it at some future date? That works. $\endgroup$
    – tylo
    May 17, 2017 at 16:12
  • $\begingroup$ You can also publish a simple hash today and reveal the content in the future. Just post it to Twitter to show you had the Wikileak or Exploit first :) $\endgroup$
    – eckes
    May 18, 2017 at 6:24

3 Answers 3

2
$\begingroup$

You can do all sorts of verification with a time stamped hash and signed hash, but it can't be a NYT headline. Just as MD5 hashing is now considered insecure and ineffective, the block width of a NYT headline is insufficient. Actually, the input entropy of the textual headline is only a few tens of bits. "Trump Convicted" does not hold a lot of entropy, which means that another headline at another time might hash to exactly the same value.

You also can't do what all good kidnapper's do and take a photo of the paper's front page alongside your object. This isn't just for the obvious reason that it's digital, but noise and the avalanche effect will ensure that you'll never manage to exactly replicate the image for future verification.

The thing to do is to hash sufficient cohesive public data. Something like the FTSE 250 list of share prices (at the time of hashing and signing) contains enough entropy to make collisions impossible. There's over 8000 bits of pure entropy available there by my calculation. That's sufficient for any possible cryptographic use. The Million Dollar elliptic curve purports to use public lottery results but that must be very cumbersome to collate. The music industry uses almost imperceptible fluctuations in the mains frequency to watermark tracks. A frequency vs. time curve could be hashed and keyed alongside the object. Object dating could be achieved for the timespan of available records.

So for example, to check if your object existed at any particular time you'd simply hash and key against all the records you have available. A hash match at a particular time would prove the existence of that object at that time. You might be able to check with a 15 minute granularity for FTSE250 share prices. Again the most important thing is to hash against sufficient entropy to preclude manufactured collisions. Clearly you can't use this technique if you don't have reliable digital entropy for the time period under consideration.

$\endgroup$
3
  • $\begingroup$ that's all fine for verifying that something existed AFTER a given date. I am looking for something that will prove that an object had to have existed BEFORE a given date. $\endgroup$ May 18, 2017 at 23:23
  • $\begingroup$ @jeremy_rutman I've added an example of hash dating at any time. $\endgroup$
    – Paul Uszak
    May 18, 2017 at 23:42
  • $\begingroup$ sorry that still doesn't do it. You've gone into details about how to verify something existed AFTER a time T - and these will all work for any time after T. That doesn't help at all if I'm trying to verify that something happened BEFORE a given time T . See the answers below involving blockchain . Not sure how to explain this any more clearly. $\endgroup$ May 20, 2017 at 10:15
1
$\begingroup$

It sounds to me like one trick would be to have the object published into some stream where independent and adversarial parties routinely timestamp and sign such documents. Blockchains are a good example.

Another, fuzzier example comes from the 2016 US election: the Wikileaks dump of emails stolen from Hillary Clinton adviser Rob Podesta's GMail account. GMail implements DKIM (DomainKeys Identified Mail), an anti-spam measure whereby an email carrier signs its outgoing emails with a key associated with its domain, so that the servers to which they forward the messages can verify that they came from the claimed domain.

A portion of the emails in the dump contain such signatures, inserted by Google or other mail providers as part of their routine operations. Since email messages are routinely timestamped as well, and email service operators do not generally engage in retroactive forgery, the signatures on the timestamped messages provide very good evidence that the messages existed at least as early as those timestamps. If you have email chains with multiple participants served by independent email providers, and where respondents routinely quote the messages that they reply to, you then end up with timestamped signatures of the same text from independent parties, which makes the case stronger.

$\endgroup$
0
$\begingroup$

Ah, got it. I just need to use a blockchain - that's exactly the right hammer for this nail.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.