Hacker News new | past | comments | ask | show | jobs | submit login
How can a time traveler encrypt info so it's only decryptable after a given date (worldbuilding.stackexchange.com)
50 points by azeemba 5 months ago | hide | past | favorite | 67 comments



There's this star called R Scuti:

https://en.wikipedia.org/wiki/R_Scuti

* The light from it is variable,

* This variation is continuous, not a rare event,

* It's 4,000 light years away, so the butterfly effect is irrelevant because the light is already on its way,

* It's chaotic and unpredictable.

Basically a reliable stream of random data. I guess the only problem is that you need to know what observations of its light levels were, throughout the 80s, if the records even exist.

Edit: "The AAVSO International Database currently contains over 110,000 observations of this erratic star spanning nearly a century of activity." There you go, that should provide a thousand measurements to base a key on in any given year.


> * It's 4,000 light years away, so the butterfly effect is irrelevant because the light is already on its way,

The butterfly effect might still apply to the earth's atmosphere and the scientists/engineers maintaining snd operating the observatories, though. But this could likely be compensated for by cutting off some of the last digits of the measured values.


"It's 4,000 light years away, so the butterfly effect is irrelevant"

Wouldn't time travel imply that you could go to the source 4000 years ago, so for this scheme you'd need to get the data at time of creation, not at arrival.

Further, once you start having time traveling all bets are off anyway. You could go back to stop the star forming. But by that point the hypothetical question becomes impossible to answer


This reminds me of the Joe Grand video titled:

“ I hacked time to recover $3 million from a Bitcoin software wallet”

Just put a program in the computer that generates a ‘random’ key based on time of day and second…and it does this every day at pre defined hours…and maybe some way that the time and date can’t be modified. Then, only when that day and time comes will the right key be generated and the files decrypted. Program does not know when that date is it just bases the key generation on time and date.


You're talking about a secure hardware element to ensure that the algorithm can't be exfiltrated and the dates brute forced against it. If you have a secure element that knows what time it really is, so you can't lie to it, just have it withhold the key until the specified date.


Yes, and no, assuming secure hardware element is not possible. An encrypted log file can be made that keeps track of time. If a leap is detected than the program will shred the data. The user can be warned of this and decide not to risk brute force.


Without secure hardware, what stops me from keeping a backup and restoring it every time the data is "shredded"? If you have no hardware element there's nothing that can stop me from, say, spinning up 10000 VMs to make 10000 simultaneous attempts.


If the NSA knew how useful the info was, they would build a thousand clones to run the program on, connect them to clocks running zero, on, two, three, etc. years fast, and after a year know everything they’re supposed to know after 1000 years, after two years everything they’re supposed to know after 2000 years, etc.

If you’re a time traveler, you could likely get a secure enclave built a zillion years in the future and tested for a further zillion years. That likely would avoid the cloning, but still wouldn’t prevent the NSA from gambling that what you’ll tell them twenty years from now is more useful than what you’ll tell them tomorrow (hoping that they’ll be able to fill in intermediate steps if needed.

For example, if I told somebody from the 1930s how to sequence DNA, they probably could infer what DNA looks like and that DNA is something worthy of attention.

If you’re a time traveler, the simple solution is to just send them the info on the date you want them to know it, though.


The team behind the awesome drand project (https://drand.love) has proposed a timelock encryption scheme based on an existing threshold network (the League of Entropy).

They even provided a fully operational, web-based, open-source implementation: https://timevault.drand.love


Make it only decrypted by brute force, taking into account the history of the algorithms and the computational level. There is no clue, you just have to use the knowledge of the time to try to solve it. You can even create the program and its minimum requirements to make sure they have a direction.


That would only let you plan the average time it takes the secrets to be decrypted, it could actually take anywhere from 0 to 2*t to actually decrypt them by brute force


Availability of computational power is unknowable. You have encouraged all governments to create a Manhattan Project to brute force future space magic. There would be nearly unlimited funding to cryptographers and hardware design processes.


Well, I don't know, I just give a seed. Basically they are using a unicorn called a "space traveler", I don't know anything about it. But if you come from the future, you know everything xD.


How does that ensure it only gets decrypted after a certain date? Brute force can decrypt it on the first try, or the trillionth try, with equal chance.


It's statically ensured, the same way I can trust if you generate a Bitcoin address right now it won't have a few thousand bitcoins in it.


Statistically? With that analogy, you'd design a crypto algorithm that, on average, gets cracked on the desired date to guess the wallet with a few thousand bitcoins. You still have the problem that (1) guessing it right on that desired date, (2) guessing it right today, and (3) guessing it right on a day 42 times the amount of time as the desired date, all have exactly equal probability. You don't get to pick from infinite timelines to make your plot work, you get 1 try, and need to be somewhat sure it'll work on that date. It might not be possible. Unless you time travel and just hand them the information.


No, it's not like that: generate a random Bitcoin address and it's practically guaranteed that it will be empty. But in this case, if the time traveler picks a random key for their secret, it's equally likely that it gets brute forced on the first attempt as it is the trillionth attempt. There's no statistical guarantee except that on average it will take a certain number of attempts.


Sure but the compute difference between decades isn't as dramatic as decrypting vs not decrypting Bitcoin.


This doesn’t help because you have a time machine.

You automate the activation of your Time Machine and have the Time Machine go back in time 40min and try to crack it, if it’s not cracked it goes back another 40minuted and tries again until it cracks the code. The code might take 80 years of time in the time loop but for you it’s instantaneous.


Position in a coordinate system of epicentre of earthquakes, perhaps those above a certain threshold depending on how much entropy you needed.


the comment in the response about using stars going nova between the present and the travelled-from time was probably the best answer, but it requires accepting the paradox of the possibility of time travel while still ruling out faster than light travel- as you could get the key earlier than the set date by travelling light years toward where the stars were sampled from.

If you are protecting the secret from non-time travelers, you're probably fine, but if you are protecting it from other time travellers, your compute or proof of work has to be more expensive than their time-travel propulsion energy units.

We have this same problem with device attestation, where if the key exists in the same "universe" (hardware substrate) as the ciphertext, you're effectively fighting time travellers because they can run, re-run, use side channels, insert a breakpoint anywhere along the way. The current solutions are in the domains of physical tamper proofing, "white box" cryptography and other obfuscation schemes, and ultimately fully homomorphic encryption (FHE).

FHE would be the necessary solution against time travellers in this case because no matter where in time they were, they would have to be present at the instant the compute operation completes- and then you're into a race condition against Planck time to see who can grab the result of the computation first when it finishes. (assuming they can't infer or predict the result earlier than the completion of the FHE computation, that earlier decryption rounds don't reduce the time it takes to brute force the rest of it in a parallel stopped-timeline, and that Planck time still means anything in a universe where time isn't unidirectional and scalar)

Writing wise, there would have to be a hitch in their time travel scheme where to work it needed randomness and a lack of precision for some important quantum uncertainty reason, and the race to be present for the FHE decryption before the time travellers resolved their precision problem would drive the plot.

don't be discouraged though, a cursory reading of scientific discovery shows that impossible is mostly a convention, and cryptography historically reduces to a gentleman's agreement.


Use an encryption key that is a headline from a newspaper for that day. This assumes the time travel works like the Marvel movies and not like Back to the Future.


Headlines are short and dictionary words, would be broken in next to no time.


In order for the butterfly effect to not disturb the results of any random or pseudorandom key, the key must be derived in such a way that the distance of the key generation event is greater than 1/2 the light-time of the key delivery delay, so that in no case can any information from the key holders reference reach the key source prior to key generation.

Other than that the key must be based in a purely deterministic algorithm to be immune to butterfly factors, which can always be brute forced by accelerating the civilization to fractional relativistic velocities or bringing it close to an extremely massive object. (Leaving the keygen at rest) Of course other ways to brute force such a key generator are probably more practical, but a relativistic attack is the steel man for the impossibility of reliability.

Stealth might be the best bet, an undersea bouy that pops up and broadcasts the key at the specified time or something like that. But then again, a fast moving civilisation would get the key “early” from their frame, even for astronomical events.

I posit that there is no absolutely reliable delay, but stealth is probably the best practical hedge.


The top answer addresses the butterfly effect, by using the time and position of novas, which are unpredictable.


It wouldn’t likely apply to 1980s tech, but a “fast moving” civilisation (or one capable of locally warping spacetime to a meaningful extent) could still receive the key in less than the allotted time.

But if we’re going to stay unrediculous, the nova -should- work , but might not if the many-worlds interpretation (which is misleadingly named) interpretation of QM turns out to be right.

The time traveler effectively brings the nova into the light-cone of the past world, altering the relationship in the MWI.

Stealth might still be the most reliable.


I don't see how the novas (which happened long, long ago) would be stopped by anything happening in any version of the 1980s.


I guess the idea would be to send FTL probes into space in all directions, so they can see the light heading our way from novas and report back.

However, accomplishing this seems considerably more difficult than finding an underwater buoy.


Which would be hilarious when they found out the key unlocked tide pods lol.


How about this: take as input something random from beyond the time horizon (e.g., brightness from stars 300 light years away). Then (if we assume the input is sufficiently random) they should be fine.


This is really one of those questions where you want X, you started down the path of answering the question but got stuck, and now are asking how to get unstuck when you should be asking for X.

He's asking "how to decrypt only after a certain date" when in reality he should be asking "how to make information available after a certain date".


The more generic although potentially harder question is without the time traveler element what could you do today to make sure something could not be decrypted before a certain date. This seems potentially useful but also every bit as tricky to answer especially since we don't know what the future entails unlike a time traveller.


For me it is easy if you are an old man you can simply know what algorithm was used at each time and at the computational level of each time, therefore you can make the only way to decipher the cave is by brute force and that it is probable and easy to resolved in each era through advances and knowledge.


This is the only sane answer in the comments so far


This kind of answer is ok but since the purpose of "sending stuff back in time" is technological uplift, you can only predict the first key. Everything else could be accelerated. (maybe that's fine but doesn't fit the spirit of the question)


I don't quite get notion of sending something back in time but before the actual date you want the thing to work... But anyway, you really can't make the crypto device a tech level superior of the time or the crypto device itself gives information on the future.


Maybe their mode of time travel can only do fixed points - for example, in 11.22.63, there is a single portal that can only travel between ~2016 and ~1960 (roughly, I don't remember exactly).


You want 10 things happening in 10 different times but only have a disposable single use time machine?


Have you watched the movie "tenet"? That's basically the main premise of the movie, to a degree.


Not going to result in an equal sharing of information, but if you can drop doodads at different time and locations, make a scavenger hunt for the key.

First key is stored at the center of Olympus Mons on Mars.


From my reading of the question, the premise is: You drop everything at once, and you can't further interact in the future (such as waiting for the date to give the password). They want to know how (if) something, by its very nature, could only possibly be decrypted after a certain date.


Isn’t that exactly what drand is meant to provide?

See drand, the distributed randomness beacon protocol at https://drand.love/


Lost texts are recovered, and artifacts are found all the time. You could use an old artifact's decrypted text, or location, as a source of a passkey.



A basket of specified (closing) stock prices on day X.


Something like that would be highly sensitive to the butterfly effect (especially considering all the tech stocks and the time traveller leaving behind technological information).

This is why I’m guessing events off-planet is the highest voted answer. Human changes won’t really affect it (or are much less likely too)


Ignoring the butterfly effect issue, I would have said just pick the title of a given newspaper on a given date as the password.


Depending on how time travel works that might have a huge hole.

1. You pick a hash of the headline from the January 14th 2030 of the Springfield Shopper as your key and encrypt some data.

2. You use your time machine to send the encrypted data to someone in 2024 with instructions telling them how to recover the key, expecting that they will have to wait until January 14th 2030.

3. Some rich person decides they do not want to wait. They notice that January 14th is National Hot Pastrami Sandwich Day [1].

4. They buy the Springfield Shopper and institute a policy that every year on January 14th the headline will be "Happy National Hot Pastrami Sandwich Day!".

5. If we are in the kind of universe where time travel to the past can affect the time that the time traveler came from their actions in #4 make it so "Happy National Hot Pastrami Sandwich Day!" is the headline you used.

6. In 2024 they try to decrypt the data using a hash of "Happy National Hot Pastrami Sandwich Day!" and it works.

[1] https://www.daysoftheyear.com/days/hot-pastrami-sandwich-day...


That's probably crackable with enough effort though. A huge effort, but it might be practical for a team of governments if they knew the person was a time traveler.


You could add more entropy by including not just the headline but an assortment of text from other pages, e.g. market price of whatever stock is listed on page X, the 10th sentence on page Y, etc.


If you can ignore the butterfly effect, I'd use the last digit of the closing price of 100 different stocks.

Stock prices are hard to predict in the first place[1]. And the last digit will be the hardest digit to predict.

---

[1] You could get very rich if you knew how. Lots of people want to be rich, yet basically nobody is doing it successfully.


Even if you could: your actions based on that knowledge would alter the price again, such that any method you devise to predict stock prices will a) have to factor that into its calculations and b) by definition will only work for a single user.


or the concatenation of the lottery numbers for day A, B, C, D, E, F, G, H

A bit disappointed by the article: the top upvoted answers there are literally a repetition of what ChatGPT says to do :|


Alerting the course of human history by gifting baubles from the future. Lottery numbers will not be stable.


Transmit the key via radio wave, from a point source located x light years from earth. The key will reach earth after x years.


Would earthquakes be susceptible to butterfly effect? Historic earhquake data might solve the issue.


Human activity (e.g. oil production, underground nuclear testing) does impact seismological events to some degree so butterfly effects are possibly there too.

If he can leave a bunch of computers in the past and there is a strong enough incentive to keep them running indefinitely... could he have software on those specific machines that will produce an output every so many years?

Would there be a plausible defense against the stewards of the machines being able to reproduce the computers state and fast-forwarding it on a faster computer?


Measurement of earthquakes is susceptible. If there's extra reasons to pay attention to and record earthquakes, that could easily influence the development of measurement technology, which would change the historical record of earthquake data.


Why not use a private key based on a bunch of lottery draws that happened just before the date?


I suppose novas are less likely to be affected by the time travler's bufferfly effect.


Depends by which time-travel rules we have to play. In the "whatever happened, happened" case, the butterfly effect is not an issue.


Well then that means you can change nothing and not time travel so all. Or you are saying “big events” are fixed? I do not understand how the universe would make that designation. Pause your grandfather on the street for one second, and your genetic heritage is completely different as a different sperm was the progenitor of your parent.


No you can still time travel, but yes you cannot change things. It has to be self consistent.

For example, if you plan to go back and kill your grandfather, you'd perhaps be run over by a car two blocks before you get to him, but if you had read past newspapers carefully before your trip, you would have noticed a short article at the end of the newspaper about some mysterious John Doe who wore funny clothes and got run over by a car just on the day you planned to kill your grandfather, which ended up being you, the time traveler.

Sean Carroll discussed different time travel logics using popular Hollywood movies on his podcast once. Here is essentially the same thing in writing: https://arstechnica.com/features/2023/11/the-ars-guide-to-ti...


In that case, there is no point in the time travel at all. If you do not already live in a universe where a future Oracle gave technological secrets to elevate humanity, you know the plan fails.


If you can send things to the past, just deliver your one time pad on the date that you want things to be decryptable. Or, depending on how exactly the time travel works, just use the encryption scheme that you know gets broken on exactly your target date.


market close data: tickers, their closing prices, and volumes for a given market or markets. For more, first paragraph of a story for each ticker.


I guess you missed the comment about the butterfly affect.




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

Search: