4
$\begingroup$

Is it possible, with modern, or easily foreseeable technology, to build a compact digital data storage device that cannot be erased without removing the data store? The storage in question to be used to blackbox data on a spacecraft.

Answers should use examples of any and all of:

  • currently prototype systems
  • existing data storage
  • and/or those used historically

The aim is to have a storage system from which data, once recorded, cannot be removed without completely removing the storage core of the system, that is to say that the recording medium is physically and indelibly write only. Data cannot be overwritten or erased by the system or by outside interference and should last as long as possible at Normal Temperature and Pressure (20°C and 1 atmosphere) and 45% relative humidity. I had considered these little babies but they can not only be overwritten they can also be annealed at high temperature to erase them completely. Assume that anyone who is going to be trying to read these systems will know the encoding system used and have the technology to read them, I'm looking for an indelible record, I do realise that on any long enough time scale encoding will shift and data will become unreadable.

$\endgroup$
  • $\begingroup$ Comments are not for extended discussion; this conversation has been moved to chat. $\endgroup$ – L.Dutch Jul 30 '18 at 15:55
  • 2
    $\begingroup$ "I had considered these little babies but they can not only be overwritten they can also be annealed at high temperature to erase them completely." Literally anything can be annealed at a high enough temperature. Which is why you're getting answers assuming you're looking for something completely indestructible. $\endgroup$ – Shufflepants Jul 30 '18 at 20:29
  • 1
    $\begingroup$ You say "last as long as possible at Normal Temperature and Pressure (20°C and 1 atmosphere) and 45% relative humidity" and then say "they can also be annealed at high temperature to erase them completely." That's contradictory. $\endgroup$ – RonJohn Jul 30 '18 at 21:15
  • 1
    $\begingroup$ It's 1970's technology called PROM (programmable read only memory). It has mostly been replaced with EPROM of different flavors, allowing BIOS updates and such like without having to physically replace integrated circuits. $\endgroup$ – pojo-guy Jul 31 '18 at 1:34
  • 1
    $\begingroup$ Slightly tangential, but unless you have infinite storage (counting regular replacement as infinite for this purpose), then you probably don't want it to be write-once. You don't know when something interesting is going to happen until it does, and you really want the events leading up to it. In practice, this means that you have to record everything all the time, whether it's interesting yet or not. Real-world blackboxes get around this by overwriting old data, so the (actually bright-colored) unit can be quite buried, well-protected, and consequently hard to get to. $\endgroup$ – AaronD Jul 31 '18 at 6:04

10 Answers 10

3
$\begingroup$

I think the simple answer is unfortunately No. Given time and the correct circumstances simply entropy will render everything obsolete and unreadable. If heat or some other external force doesn't get there first.

However there are ways to circumvent a lot of this. Diamonds Burn, this is true, but one would assume if you were to wrap it in a Carbon-Titanium case, which also protected the circuit board for reading the data from it, then the case would protect the diamond from harmful light, it would also protect it from wear and damage over time. Although the case could be breached, this act would be very obvious if it were to occur

Yes there is the risk that as technology progresses then the format may become unreadable, however the outside of the case could be laser etched with details on however to construct the equipment to access the data. That depends on how you plan on accessing that data and how quickly you would want to access it though.

If you didn't want to access a large amount of data all at once, then in theory you could layer very slim strips of Carbon Steel (stainless) and tungsten (chosen because of the comparative difference in visual light and dark metal and ability to withstand damage and heat), and if layered correctly could be "read" like a bar-code, bar-codes are just 1s and 0s after all in a sequence. If this happened then you could scan snippets of data one after another and build up the bigger picture, yes it would require the basic human understanding on binary to last, but it's likely that it will for the foreseeable future. Bar-code technology is so prevalent these days that its unlikely that humanity will go away from it as well.

But this is only useful for very short sentences and it would take a very very long time to "record" this information, but they would still be "readable" after many centuries and be resistant to standard temperature ranges across the globe and then some.

I appreciate this wouldn't be "compact" but invariably the more compact something is, the more susceptible it is to outside damage. It wouldn't really be digital other than being readable in binary...

So for now the questions remain... how much data? and how fast does it need to be read?

$\endgroup$
  • 1
    $\begingroup$ A bimetallic stack has some distinct possibilities, if laid down in sections only a few tens of atoms thick and welded in place it could be data dense and indelible unless physically trashed. $\endgroup$ – Ash Jul 30 '18 at 13:32
  • $\begingroup$ @ash, yeah that was my thinking, but again it wouldn't have a high read rate, and and would be quite expensive and time consuming to "write", but definitely would need someone to come along and take a lot of time and effort to physically destroy. and being that the potential metals i suggested have melting points above 2000 degrees its safe to say that even extreme temperature ranges short of falling into a volcano or have thermite dropped onto it, it should be safe $\endgroup$ – Blade Wraith Jul 30 '18 at 13:59
  • $\begingroup$ If the metal was in premade foil sections it needn't take too long or too much energy to lay them down and fuse them. A continuous pass reader, similar to what they use for digitising tree ring and ice core data could extract a lot of data quickly. $\endgroup$ – Ash Jul 31 '18 at 11:58
14
$\begingroup$

We can't make anything indestructible, that is a silly idea and I am going to skip this. We can make something 100% tamper proof though. I am going to follow on AlexP's comment.

The basic idea is to take something like a cd. Use a material that has some sort of noise to it, like granite but at a microscopic scale. Read the entire natural pattern of the material and store it securely at home. Use a laser to etch information into it. The laser will burn holes and damage the material as it writes it's message destroying the patters in the 1s and preserving it in the 0s. This disc cannot be faked. It is most likely that even if they wanted to, there is no manufacturing process that can recreate a random pastern created by a natural process. This would be the difference between getting a piece of wood, and trying to 3d print wood that has exactly the same grain pattern.

Even if they could they would not know it. Assuming that the device wrote the real data, some of the pattern is permanently gone. If there is any place on the disk where the machine printed a 1, and you need a 0 there is no way to do it. You simply don't know what a 0 looks like because that spot has been destroyed. Assuming that any new data would need some 0s where there are 1s forgery becomes impossible. Even if you think that you could cleverly only add 1s to change the data, make the system write out checksums. It will make this impossible.

$\endgroup$
  • $\begingroup$ Where did you read indestructible? That would be silly but as I didn't say it you might want to start again. $\endgroup$ – Ash Jul 30 '18 at 14:28
  • $\begingroup$ @Ash does the rest of the answer answer your question? $\endgroup$ – Andrey Jul 30 '18 at 14:32
  • 1
    $\begingroup$ @Ash we can simply make the motor that moves the robin not spin both ways. Does that solve it? $\endgroup$ – Andrey Jul 30 '18 at 14:35
  • 8
    $\begingroup$ @Ash honestly once someone is tampering with a black box enough to somehow force it to rewind and write garbage, something way outside of it's functionality, it's really no different that filling the device with lava. Then we are back to indestructible $\endgroup$ – Andrey Jul 30 '18 at 14:40
  • 1
    $\begingroup$ @dn3s as i mentioned in my answer, part of the security is that the writing process destroys some of the medium making cloning impossible as you simply lack the data $\endgroup$ – Andrey Jul 31 '18 at 14:58
9
$\begingroup$

Laser etched crystal should give you a sturdy storage medium which only allows tamper-evident destructive overwrites of data.

When strong lasers hit crystals they melt, chip, ablate, crack, and just generally deform the crystalline structure, creating large and easily detectable defects. Unless technology exists in your world to put individual atoms and molecules back into place, then any deformation is one way: either you completely wipe away an entire layer of crystal and destroy all written data, or you write new data on top of old data and leave behind obvious evidence of tampering.

$\endgroup$
  • $\begingroup$ What if you re-create the whole crystal with new data? $\endgroup$ – Michael Jul 30 '18 at 15:30
  • 1
    $\begingroup$ @Michael: That's no different from removing the entire storage system and replacing it, which the OP seemed ok with: "data, once recorded, cannot be removed without completely removing the storage core of the system" $\endgroup$ – Giter Jul 30 '18 at 15:41
  • 1
    $\begingroup$ Holographic etching allows for three dimensional data storage too - this will allow the entire crystal to be used, leading to orders of magnitude more data that can be stored. $\endgroup$ – Baldrickk Jul 30 '18 at 16:11
5
$\begingroup$

Since you say in a comment that you're willing to settle for a combination of tamper-evident and write-only unless tampered with, as opposed to strictly tamper-proof (which indeed is a much harder problem to solve)...

You could easily model this after aircraft black boxes (flight data recorders and cockpit voice recorders).

Basically, make a dead-simple recorder that meets whatever criteria you have, and make it reliable. For a long time in aviation, this was a simple recorder to magnetic or even physical storage (in the form of engraving onto a slowly moving aluminium foil), though recently manufacturers have moved to fully digital storage. Magnetic tape recording has the advantage that the tape can be made as an endless loop of basically arbitrary length, allowing recording of a known amount of data which is automatically overwritten when needed; digital storage systems would need to implement this in some other manner, but the principle can remain the same.

With the recorder in place, define a dead-simple interface to provide the data to be recorded. Don't allow any readout; for example, in the case of magnetic tape storage, you could accomplish this by physically having no playback/read head. The simplest would probably be for a n-track recorder to have n distinct analog inputs recording onto their own tracks on the storage medium.

As long as there's power, the recorder runs and records whatever is presented on the inputs.

Now apply standard tamper-evidence measures to the whole device. Seal the interior in epoxy (but make sure that it won't overheat during use), use one-way screws, apply some glitter nail polish to edges and document the resulting pattern, and whatever else might make the device more tamper-evident. Your ideas are probably as good as mine.

Next, apply tamper-evidence measures to the connections to whatever sensors feed data into the device.

If you want to go low-tech, add an old-style impact (for example, dot matrix) printer to the system and have it regularly print checksums of stored digital data along with some kind of timestamp. If there's tampering, the paper record and the digital record won't match, while the paper record doesn't need to contain all of the raw data.

None of this will prevent tampering, but it will make it a whole lot more difficult to tamper with the recorded data without there being some indications that tampering occured.

$\endgroup$
  • $\begingroup$ I did wonder if someone would remember the first generation flight data recorders I thought that might be usable but I really wasn't sure.. $\endgroup$ – Ash Jul 30 '18 at 14:17
  • 1
    $\begingroup$ even better: prepend the previous checksum to the block of data used for the checksum. that way tampering with one block of data will require not just faking that one printout but every subsequent printout as well. a primitive blockchain! $\endgroup$ – dn3s Jul 31 '18 at 7:03
  • 1
    $\begingroup$ @dn3s Sounds more like a primitive single-branch variation of a Merkle tree than a blockchain to me, but I guess more people are familiar with the latter these days. $\endgroup$ – a CVn Aug 1 '18 at 6:20
  • $\begingroup$ @MichaelKjörling oh yes that fits better! the blockchain comment was just supposed to be a reference to the idea of the append-only ledger functionality it provides using hashes, kinda a superficial comparison. i guess one could also say it's like a git repo! $\endgroup$ – dn3s Aug 1 '18 at 16:49
2
$\begingroup$

I don't believe this is possible, for one specific reason:

Any writable medium could, theoretically, be overwritten. Assuming you find something indestructable, all someone needs to do is overwrite whatever is already one there.

To avoid this, you'd not only need a resilient medium, you'd also need something that undergoes a transformation along all or part of its capacity that makes it unwritable.

So for example, maybe its only writeable for a certain period of time. Or maybe written sections are also made unwritable by the act of writing them.

But even that is problematic because it would require encoding the entire volume. You couldn't use a system that has any empty space it its encoding. E.g. even if someone can't engrave over etchings on a stone tablet, they could etch all the untouched stone in between your lines. Or if said another way: if a system uses 1s and 0s, the zeros can't be "unwritten" or blank, because they could simply be written over, corrupting the entire message.

That, in itself is a tall order. Its possible to brick write-only systems, so you're looking for an exotic setup that actually changes substantially at some post-write point, and which also has the other characteristics, specified.

$\endgroup$
  • 2
    $\begingroup$ Thanks that tells me where the challenge lies, I need a medium that is fundamentally not the same going in as it is coming out. $\endgroup$ – Ash Jul 30 '18 at 13:34
  • $\begingroup$ Do you think an edge cut tape of some sort could work then? $\endgroup$ – Ash Jul 30 '18 at 13:41
  • $\begingroup$ I'm not sure what you mean. Could you link? Are you talking about a medium that is cut after recording? If that's the case, that might solve the problem. If the tape is cut after recording and isolated from the recording mechanism, maybe housed only with a "reader" compartment, it would be tamper-able, but only if you open or remove the compartment...which seems to meet the criteria of the question. $\endgroup$ – user49466 Jul 30 '18 at 17:46
  • 1
    $\begingroup$ A tape that is incised along one edge, the cut is deep for a 1 and shallow for a 0 if you're encoding binary but the whole edge is used when writing data and for a write only device it only spools in one direction. $\endgroup$ – Ash Jul 30 '18 at 17:48
  • 1
    $\begingroup$ It could if it could over-write a written section which is why you'd have unidirectional spooling of the writer, and the reader for that matter. $\endgroup$ – Ash Jul 30 '18 at 17:57
1
$\begingroup$

Would a 2-column punch card or tape made of a suitable plastic not achieve this? Many suitable plastics will resist biodegradation for 400-500 years and are easily formed.

A segment is a specified dimension, let us say 5mm long, the two columns are 5mm wide each, and a hole is punched in the centre of one of these, 2.5mm in diameter.

A hole in one column denotes 1's, the other denotes 0's. No holes are an unused block. Holes in both columns is invalid and therefore the data is considered destroyed.

The device used to punch the holes in plastic would also introduce a contaminant to the edges of the hole - perhaps just a different colour of plastic - this is to help prevent the holes being simply filled in.

Once punched in, there is no valid way to "rewrite" the data without reforming the card or tape to re-fill the holes anew and begin again, by which time you may as well have replaced the media altogether, which is a risk with any type of media.

The device used to read the media would use an optical sensor as well as a means to detect the physical hole, so as to confirm the presence of the contaminant in the edges.

$\endgroup$
  • $\begingroup$ Yeah that'll work. $\endgroup$ – Ash Jul 31 '18 at 17:47
0
$\begingroup$

Nothing is totally indestructible. Pretty much anything made of matter can be destroyed by a concentrated beam of positrons.

But WORM drives are possible and doable. Put the thing in a case that's reasonably durable. Send it info and it either records the data without overwriting old data, or it fails to do so. You can't overwrite old data because it refuses to do so and even fails to give you instructions for how to do it.

So if you break into the case you can do stuff.

And if you throw it into the nearest sun you can probably destroy it. Or depending on how your spacecraft engine works, its exhaust might destroy it. But the challenge is only the physical one of creating a black box that survives a lot.

$\endgroup$
  • $\begingroup$ Where did you read indestructible? That would be silly but as I didn't say it you might want to start again. $\endgroup$ – Ash Jul 30 '18 at 14:28
  • 2
    $\begingroup$ The original question considered a solution that was not adequate because it was safe only up to 1000 degrees. It sounded like you wanted something that was very tough. You can mostly prevent data from being overwritten just by having a system that refuses to overwrite data. A mechanical malfunction could result in overwrite, but to make it happen you'd have to break into the black box and do sophisticated things. I $\endgroup$ – J Thomas Jul 30 '18 at 15:49
0
$\begingroup$

Guarantee that any tampering leads to the total destruction of the data.

Put the storage device in a box with a bomb. Set the bomb to go off if the data is overwritten, or if the box is opened.

You've successfully reduced the possibilities to "untouched" or "destroyed", and destroyed is always going to be an option.

$\endgroup$
0
$\begingroup$

If you’re doing interstellar travel, you’re already dealing with made up physics, so I will assume you aren’t. The answer, then, is simple: put your storage medium on a celestial body, say, the moon, and attach a transmitting station that accepts authenticated signals for new writes to capacity and never accepts overwrites (always open in append mode). It also will, if sent an authenticated read request, disgorge its data. The ship will send whatever it needs recorded to the monitoring station via radio waves to the transmitting station.

Short of landing on the moon and getting your hands dirty, there’s no way to tamper with the device non-destructively.

$\endgroup$
-1
$\begingroup$

I believe There is a way, though it was only done as part of research as of yet.

There is a way to store data on a diamond.
Since its one of the hardest materials in nature, it can survive quite a lot, so it may be your solution.

$\endgroup$
  • 7
    $\begingroup$ Diamonds burn, they're also surprisingly fragile, as we often remind people here, hardness is not toughness, diamonds can't absorb much energy before fracturing. $\endgroup$ – Separatrix Jul 30 '18 at 11:23
  • $\begingroup$ Additionally the write system could easily be used to scramble the data; as the article notes if you expose the diamond to any light at all the data encoding is ruined. $\endgroup$ – Ash Jul 30 '18 at 11:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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