What can a time traveler use to generate an encryption key to encrypt information so it's only decryptable after a given time period

Question

I have a time traveler basically dropping a few computers with a sort of technological uplift program into the 1980s. These computers have information about future science and technology, along with what guides and advice for how to most efficiently go from 1980s era technology to future technological level. Every computer has identical information and they are spread out across the world with the intent that all this information become publicly accessible to everyone.

Our time traveler also wants to make sure that certain information is not immediately available despite encouraging disbursing of data to everyone. They want some information to only become accessible at certain time periods. To do this, I'm imagining they encrypt this data and then provide instructions on how to generate the key required to decrypt it, but with the intent that the information required to generate the key would not be available until the desired time period.

Since the butterfly effect is in full force, I imagine they would most likely use something like astrological observations and measurements, things that happened long ago and thus can't really be changed by recent human technological development, for generating this key.

My question is how well can they reliable create such a key, and what measurements should they use? They want to ensure no one can access this information until time X, meaning they need a large key to prevent any attempt to brute-force the encryption. They also need to be able to adjust for error bands to ensure that those generating the key will always generate the exact same key without getting bits wrong because their sensors weren't perfectly calibrated etc. Ideally they would have a means where they could generate a key for any time X by simply loading historical recordings from that time and generating a key, so that means it not be dependent on rare astrological conditions.

Answer 1

Novas

About ten novas occur every year in the Milky Way. Many occur near the Galactic Center, but sufficiently many occur nowhere near it.

With the current scare about quantum computers the time traveller will want a 256 bit key, of course.

Each nova has a well-defined position in the sky, given as right ascension H₁H₂ hours M₁M₂ minutes S₁S₂ seconds (0 h to 24 h) and declination d₁d₂ degrees m₁m₂ minutes s₁s₂ seconds (−90° to +90°). The time traveller gives the following rule to derive the key for the period year n to year n + 4:

1. Take the first 16 novas which occur in the given five year period farther away than five degrees from the Galactic Center, and which are brighter than the 9th magnitude or so, sufficiently bright to be easily observable with binoculars.

^{You may get a few which are borderline around the 9th magnitude; this only means that you may have to do the calculations several more times, with them included or not, until you find the correct key. It is not a big deal.}

2. For each of the 16 novas write the decimal number H₁H₂M₁M₂S₁S₂d₁d₂m₁m₂s₁s₂ and represent in binary as a low-endian 64 bit integer.

3. Concatenate the 16 binary numbers to get a string of 128 octets.

4. Take the SHA-256 checksum of the binary string.

5. This will give you 256 bits. Use the key to decrypt the chunk of the Melontopaedia for the period.

(I am grateful to @SurpriseDog for suggesting to concatenate the numbers before taking the SHA-256 checksum.)

Novas cannot be predicted in advance. There are more than 16 bits of randomness in their coordinates. Astronomers will have zero difficulty in measuring their coordinates to the second. (They usually measure them with much greater precision.) The proposed rule effectively specifies a completely unpredictable 256-bit number.

(For the nit-pickers: the years are taken from 0:00:00 January 1st UTC to just before 24:00:00 December 31st UTC. The RA and declination are relative to the J2000 epoch; note that the J2000 epoch was defined in 1976, so it was already usable in the 1980s.)

Happy technolifting!

Answer 2

Use timers

A slight frame challenge, as I would opt not to use the users to decrypt anything.

Relying on the people themselves to unlock it is very dangerous. They could miss it thanks to fate, like a power outage the moment you want to measure gravitational waves or a nova explosion. They could also find ways to get the information early if they break in with some unforseen method.

The answer is refreshingly simple. There aren't any computers. There are just contraptions that are tamper proof. The contraptions have their own power source and sensors to detect any intrusion. It'll initiate a melt down if people try to physically open it, or see through it in any way. That might not discourage someone or some government from trying, with a small nuclear crater as a result. It is then a simple risk reward formula. The risk to have it melt down and lose all incredibly useful future data is too high.

At the right interval the contraption releases the data itself. All data has been decoded by the contraption and send via an easy method. Think a computer of non-future components, send it out over a simple radio band, or even just printed out on a piece of paper.

The least amount of interaction you offer, the least amount of interfering you can have. The scenarios are already set by you. As no one can interact with the contraption, no one can guess an encryption or force their way through. Any attempt to try so is discouraged disproportionately. Smart people will realise that the only way to access the information safely is to wait for the next message.

Just add a manual so people know what it is and be very clear what will happen if you try to tamper with it, without revealing how it would do it.

Answer 3

Low Tech solution...

In Flann O'Brian's 'The Dalkey Archive', the eccentric genius de Selby reveals a plot to simultaneously make the air lifeless throughout the Earth using some substance he called DMP. How was he going to spread this all over the surface of the Earth to be triggered on the same day? He was going to use the Post Office.

If the time-traveller is not going to turn up in person on the day, and reveal their message, then they could use some similar low-tech solution. They could lodge documents with several lawyers for fail-safe, with instructions that they should not be opened until a certain date. That could give the key to a public, encrypted document. Or it might just give directions to where the document was hidden.

If you have an encrypted document, this would be a nice plot twist. Computer scientists might spend 20 years trying to brute-force crack the documents, only to find that the key had been written down all the time. And when they decrypt it, the first words should be...

Mairzy doats and dozy doats and liddle lamzy divey

Answer 4

Provide a security token. That is, a tamper-proof hardware package that contains the secret. In your case, the token would contain a long-life power supply (maybe an RTG) and a clock; and it would not release the key until the clock has counted a sufficient number of ticks.

In the security business, it's generally recognized that there is no such thing as a lock that can't be broken, but they can do a pretty good job these days of creating containers that can't be opened without destroying their contents. That's especially true when the content is digital information.

Put a label on it that says, "If you are patient, you will be rewarded. If you are greedy and you try to open this before __date__, then you'll get nothing."

Answer 5

Combine the times of a number of events happening

So you want to have a long key, and room for error. You can’t do this with time of an event alone, let’s assume you bring AES, or something like it back in time and bundle it with it so that with a key, you can decrypt it, as encryption of the 1980s tended to not be very secure. With a 128 bit key, it would require accuracy down to roughly 10^-20 nanoseconds to generate a key. That’s not gonna happen.

So we need a way to take lower accuracy measurements, and combine them to create a long and hard to fake key.

One possible way to do this is the last few binary digits of the unix seconds at the time an event occurs.

So for example, one part of it could be 111011100000 which is the last 12 bits of the Unix timestamp of GW170817. Then you can have a bunch of other specific events picked out and communicated, with the rough time and location to observe to find them. That’s because 12 bits of Unix timestamp only covers just over an hour, meaning you can tell people what day it happens, without actually giving anything away. With just 22 events using this, you can get a 256 bit key, standard for encryption nowadays. More events, or using more of the Unix timestamp would increase the security of it, without making it much if any harder to measure.

Answer 6

The answer from @Topcode is a good one.

First of all, AES (Rijndael) wasn't invented until 1998. Before that, 3DES was the major encryption algorithm in use. So, since neither the ISO number, winner of the AES algorithm competition or date of publication would be related to most things even considering the butterfly effect, it would be sufficient to give clues to look for: "Use the algorithm which will supersede 3DES". It is always obvious that 3DES would be replaced sooner or later, so that's not much of a giveaway. Instructions for how to decode timestamps as bits into the key could be given as clues.

Typical clues for timestamped or localized events that would be obvious when they occur, but impossible to guess in advance, could be

• "Two timestamps on a Terrible Tuesday 2001" => The time of Twin Towers.
• "Timestamp of bolide in Russia" => Chelyabinsk meteor.
• "Date of perihelion for the discovery of July 23, 1995" => Hale-Bopp meteor.
• "Timestamp of shake in 2009" => Thailand tsunami.
• "From ashes to ashes, but where" => location of Eyjafjallajökull eruption.

Different timestamps can be described as "seconds/minutes/hours/days from date/time X", where X isn't necessarily Unix Epoch, but something that gives a reasonable number of unpredictable bits.

Naturally, there will be several failed attempts because of interpreting timestamps as UTC, local time, guessing wrong dates before the correct date arrives etc., but those mistakes are "few" compared to brute-forcing an entire timestamp. Either way, the development of non-technology-accelerated computer performance will naturally be limited, giving automatic protection against brute-force attempts.

Also, as time passes, the number of remaining bits to discover decrease, which of course would make it possible that someone actually brute-force the last set of bits rather than really waiting for it. In fact, there might be confusion because when the brute-forced timestamp finally happens, there is indeed nothing peculiar happening - because of the butterfly effect.

Answer 7

Digits of Pi

If you want the material released "when people are ready for it", use something that is hard to compute. A typical example would be

Lesson 1 is encrypted using digits 10,000 through 10,999 of pi.
Lesson 2 is encrypted using digits 100,000 through 100,999 of pi.
Lesson 3 is encrypted using digits 1,000,000 through 1,000,999 of pi.

And so on.

Being able to compute this requires both good hardware and good algorithms. Both have been improving steadily over the decades since computers arrived.

Unfortunately, people can calculate pi in secret and get a head start on the others, this might be against the wishes of the time traveller.

Still, given the speed of development, everybody should catch up within a year or two.

Answer 8

I am afraid astrological events are very (extremely even) predictable on the scale of millenia or more - chaos will eventualy win, but not on the timescale we are talking about.

Bigger danger for the traveller's goal is that his manuals will be discarded as one of (sadly too many) crackpot writings, if he openly mentions astrology (and time travel, on top of that).

Otherewise, timing of events would be the best bet - astronomical events are the best, probably the orbital elements of hirtherto undiscovered comets, they might provide handful of bits of information each. (Super)Novae are even better, since they are causally disconnected from recent history (unless the time travel brings works the "parallel universe way" and the rest of our observable universe might be lightly different).

Your traveller might be tempted to use major climate events (memorable hurricanes, floods etc.), but due to butterflies they are unlikely to occur the same way second time. But, given the traveller is into astrology, would he even know this? Might be a good plot element.

Geological events (volcanoes, earthquakes) might be more reliable, but they are still causally within reach, so chaos still might rule.

Overall, I would not rely on the cryptography that much - too much data makes it easier to crack the encryption. Instead, encode geographical location of dead drops, each one providing the next batch of knowledge.

There is also the method outlined in Robert L. Forward's Draggon's Egg, where the decryption bits come from the immediately preceding discovery -e.g. you find the next piece of quantum gravity theory if you measure some fundamental constant up to X digits (meaning you must have advanced accelerators first). You know the description of FTL travel is located in a secure box on the second planet of Proxima Centauri, etc.

Answer 9

Anything which is easily looked up after a given date but unknown before then. The birth name of Pope Sarah the Third, the official exact time of the first manned landing on a moon of Saturn, etc.

There are some extreme examples at the end of Robert Forward's book Dragon's Egg, where the nonhumans leave a set of scientific solutions in places where the humans would have to essentially solve the problem themselves before getting the exact/improved answer, as incentives to work toward finding those solutions.

Answer 10

Frame challenge.

Any specific prediction about the future that comes true (specific encryption keys that depend on currently unknowable values from the future) will inevitably be turned into a religion and any information attained in the process will be converted to some ideology or other. The Law of Unintended Consequences and the Law of Human Nature vs Science. High Priests of the past controlled the people because of their ability to predict eclipses, for instance, or the seasons.

When the Moon is in the Seventh House, ALL WILL BE REVEALED.

Astrology wins over Astronomy - using the stars in the heavens to predict human events and human fortunes. Foretelling the future by reading tea leaves.

Answer 11

Second frame challenge.

If you really think about it, according to the 'reversibility of time' notion in Physics that states, basically, that in any physics equation, time can be negative as well as positive, the one thing that is NOT covered by this is 'information'.

That is, maybe physical stuff can be sent back in time, in effect 'undoing' what forward time did. That is, chemical reactions go in reverse, undoing the change, gravity reverses, so things fall back up to where they started, stars un-burn their fuel, reverting to their former state, and so on, planets travel backwards in their orbits. Like a movie shown in reverse.

However, there are no provisions in these equations for sending information back in time, because, well, how do you 'undo' information? Physical stuff may be able to be 'undone', but not so much 'life'. If you send an encryption key back in time, then it reverts to - what? It certainly cannot exist as it is in the present, and there is no precedent for that encryption key - it did not 'come from' anything.

Answer 12

I suggest something simple. Imagine you want to decrypt a file by 1992.

"XOR this XXX file with the digital track of the winner of Eurovision 1992".

The result of a XOR of the data (contained in a 5 inch disk) with the CD of the eurovision winner will be an HQX file containing a Mac PowerPC executable that would decrypt the data (maybe asking for some extra info in the process, such as "please insert this program in a Mac with CD unit and introduce the Nirvana Nevermind CD").

PowerPC chips were sold from 1992, so there is little risk reverse engineering can get the algorithm. Also, nobody would expect in the eighties an HQX containing code different from 68k.

Answer 13

Frame challenge: knowledge alone won’t likely help

I’m not convinced that knowledge alone sent into the past would have any significant positive impact. Consider how many ideas never take off because they’re ahead of their time. Knowledge is only useful when the civilization has sufficient technology level to use the knowledge. For example knowledge about the theoretical underpinnings of modern computers wouldn’t help first century Romans much as they’d lack the ability to fabricate a computer (among other things).

So you’d first need to establish that in the 80s we’d have gotten much farther ahead if only we’d known what we know today. That seems unlikely TBH because our technology level was quite a bit lower then (at least in areas you’re probably interested in exploring) and technological advancement takes time, and not just because we haven’t figured things out yet. It takes time to develop collective capabilities to put theory into practice as a civilization.

In the 80s our capability for manufacturing high density integrated circuits was significantly cruder than it is today. Memory restrictions on computers meant programs couldn’t be nearly as complicated, and you couldn’t just run computations longer as a workaround because computer speeds were atrocious compared to today. Modern AI was completely out even though neural networks go back at least to the 70s I believe. The list goes on and on.

The limiting factor in general wasn’t as much that people couldn’t figure out theory as it was that our technology level always limits what is possible and technological advancement is often quite slow. I’m not convinced you can realistically speed this process up with a few choice “knowledge bombs” sent into the past.

Answer 14

Calculate the key by obtaining a hash from files of a particular software release in the future (like specific versions of Windows, Doom PC, etc).

For example, take some files from a Windows 95 release and use them to calculate a hash that can be used as a key.

If I gave you the computer in 1980, you don’t have a way to know what these files from Windows 95 will look like until they are released. And if the hash is big enough, you cannot brute force the key in an acceptable period.

The only problem with this method is the butterfly effect: if some people know what you're looking for, it could influence decisions and give you something different.