How would you design a satellite to last 5 billion years?

[Seret]

Actually having had a think about it creating a physically durable object might not be good enough. What you'd want is a self-replicating system with flawless error-checking.

Five billion years is an insane amount of time though. Nothing lasts that long without changing.

[lajoswinkler]

Actually having had a think about it creating a physically durable object might not be good enough. What you'd want is a self-replicating system with flawless error-checking.

Five billion years is an insane amount of time though. Nothing lasts that long without changing.

That's impossible. Self-replicating system would have moving parts and there isn't anything like "flawless" in any technology.

Durable materials can survive 5 billion years if they're shielded from fluids, temperature fluctuations and ionizing radiation. A simple crystal of sodium chloride can do it and it's not an example of durability. As long as the constituents do not creep, stretch or shrink, and there are no chemical reactions, the material stays unchanged for an indefinite amount of time. Even concrete might fail because it's in a state of longterm solid-state chemical reaction. Its composition is very complex.

Diamonds could survive. They aren't a stable allotrope, but the time it takes for them to deteriorate even a bit is huge compared to 5 billion years.

Dense data storage might be done by inscribing stuff onto graphene sheets but not by single atoms because that tends to move around. Microscopic, but in the realm of bacteria, not individual atoms.

5 billion years is not enough to experience weird stuff we would see with 10⁷⁰, approx. time needed for quantum tunneling to complete turning any solid arrangement of particles into mush.

5 billion is in the realm of practical thinking, not heat death, nucleon decay, Boltzman brain and quantum tunneling realms.

Edited January 15, 2014 by lajoswinkler

[K^2]

That's impossible. Self-replicating system would have moving parts and there isn't anything like "flawless" in any technology.

You can do error checks with any quality you need. Definitely enough to prevent mutations of a self-replicating system over a 5-billion year span. The only reason life forms mutate is because it's actually beneficial for spreading. If your goal is to preserve information in a self-replicating system, you can build a system that does it.

[Seret]

That's impossible. Self-replicating system would have moving parts and there isn't anything like "flawless" in any technology.

Nope, nothing is flawless, but you don't need to be a flawless replicator if you can detect the errors. You'd want massive redundancy, and continuous self-replication with the errors discarded. At each generation the error-checking validates the integrity of the data. The main problem is that all this would consume energy, so you'd need an appropriately long-lived energy source. Realistically that means a star, I suppose.

[Pyre]

I would design a synthetic lifeform and throw it at a passing comet. It then spreads to other bodies in the outer solar system. Each colony records and stores data, then transmits it to it's neighbors. The ultimate in "cloud" storage.

[lajoswinkler]

You can do error checks with any quality you need. Definitely enough to prevent mutations of a self-replicating system over a 5-billion year span. The only reason life forms mutate is because it's actually beneficial for spreading. If your goal is to preserve information in a self-replicating system, you can build a system that does it.

Errors during DNA replication arise not because that's useful, but because that's unavoidable, and the nature uses it in a passive way. It was not evolution that made up the laws of chemical kinetics which are the basics behind replication and in the end, life.

Explaining the details would require to shove an entire semester of molecular biology into one post. I can't do that.

I don't see how such flawless system could exist. Cellular mechanisms are incredibly complex and precise, much more than anything we've ever made, relative to the magnitude of the complexity of the task, and yet even with holy-crap-amaze-wow redundancy systems, errors arise simply because it's work with moving molecules.

Nope, nothing is flawless, but you don't need to be a flawless replicator if you can detect the errors. You'd want massive redundancy, and continuous self-replication with the errors discarded. At each generation the error-checking validates the integrity of the data. The main problem is that all this would consume energy, so you'd need an appropriately long-lived energy source. Realistically that means a star, I suppose.

If one tiny error is made, it would not only tag the system as flawed, but it could also create a huge mess, depending on where the error is and how important it is.

The fact that there's an error checker doesn't mean no errors can escape. The error checker can make an error, too.

[Seret]

The error checker can make an error, too.

Indeed, frankly I think the whole concept of this thread is absurd.

What I had in mind was more akin to the way the integrity of digital data is safeguarded. Copying data, then taking a hash is a pretty bombproof way of ensuring the data is intact, especially if you had numerous systems doing so in parallel which had to agree. The likelihood of getting a false positive on a cryptographic hash is very, very low.

What kind of system you'd run this algorithm on, I have no idea.

[Thiel]

Anther problem with the self replicating system is that it need a power source.

Personally I'd go for engraved gold disks buried a couple of km below tthe lunar surface in the deepest polar crater I could find and seal it in several metres of lead.

[theend3r]

Power wouldn't be much of a problem even with current level of thechnology. Solar until it's unusable and then thorium reactor. Not only is it everywhere but for one facility? Would last forever. And yes, there can be errors and the error checker can make an error, too but that's okay, right? Those arguments don't matter if the keepers are intelligent. Make multiple AIs take care of it. The only difference between that and people would be that robots are easier to mantain - repairs and energy is all they need.

[Nuke]

Errors during DNA replication arise not because that's useful, but because that's unavoidable, and the nature uses it in a passive way. It was not evolution that made up the laws of chemical kinetics which are the basics behind replication and in the end, life.

Explaining the details would require to shove an entire semester of molecular biology into one post. I can't do that.

I don't see how such flawless system could exist. Cellular mechanisms are incredibly complex and precise, much more than anything we've ever made, relative to the magnitude of the complexity of the task, and yet even with holy-crap-amaze-wow redundancy systems, errors arise simply because it's work with moving molecules.

If one tiny error is made, it would not only tag the system as flawed, but it could also create a huge mess, depending on where the error is and how important it is.

The fact that there's an error checker doesn't mean no errors can escape. The error checker can make an error, too.

cosmic rays and other cosmic radiation are pretty much the universe's random number generator. ive seen some hacks on various blogs (such as hackaday) that exploit this noise to generate random numbers for a computer.

the possibility of genetically engineering a life form to extract energy from cosmic noise so that it can keep a dna string replicating is a good idea, im thinking a very basic unicellular life form that only breeds. data loss would be significant but if half the data was recoverable it would just take multiple satellites and multiple strains to increase the odds for the whole archive. the problem is keeping the data from getting switched on and causing the life form to die out. it would need to be engineered to thrive on nothing but cosmic radiation in very extreme conditions, which im not even sure is possible.

Edited January 15, 2014 by Nuke

[K^2]

Errors during DNA replication arise not because that's useful, but because that's unavoidable, and the nature uses it in a passive way. It was not evolution that made up the laws of chemical kinetics which are the basics behind replication and in the end, life.

Errors are natural. Flaws in error detection are "intentional". We have just the right amount of error correction for diverse yet stable system. Take a look at how many things have to go wrong for cancerous cells to develop. Note that some of these things are left intentionally loose because survival of the organism is more important than the perfect preservation of data.

If one tiny error is made, it would not only tag the system as flawed, but it could also create a huge mess, depending on where the error is and how important it is.

The fact that there's an error checker doesn't mean no errors can escape. The error checker can make an error, too.

It's called error-correcting codes. Ever had a CD player? Audio CDs use these codes to let the player skip over scratches and smudges. The idea is that given n bits of data you want to hold on to, you encode them with n+m codes, such that if any m bits are missing, you can recover the data.

And if you are clever about it, error checker will not make errors, because it checks its own operation. That's the basic idea behind CRCs, for example.

[Capt. Hunt]

Well, the first thing that comes to mind is to not put it into an orbit of anything in the solar system, I just skimmed the first couple pages, so I might have missed something, but I did see a couple of suggestions to place it in LEO or Lunar orbit, I wouldn't go with either. Orbits decay, maintaining any orbit would be a huge draw on propellant over the 5-10 billion year lifespan we're talking about, no matter what fuel source we use. Even a Lagrangian point orbit wouldn't be stable enough. The simple solution is to fling it out of the solar system all together. Of course, without a big heavy telescope you'd have to give up on recording the annual pictures of the Earth.

[check]

I have no idea how you would build a system that would be able to take and store pictures over the course of 5 billion years. Even if the system was 99.999999% reliable every year and you built a billion of them for redundancy, the odds of any one of them still functioning after 5 billion years is virtually zero.

[Kilmeister]

I don't see a way anything could continually operate for that amount of time. As far as leaving a long term data archive it may be possible by going very low tech. Build some form of physical data storage mechanism and put it in orbit somewhere in the outer solar system. And so someone in the far future could detect you could coat one side with a reflective metal (Gold?) and spin it. But then again the whole thing depends on some future species coming along and finding a tiny rock floating around far away from some insignificant white dwarf and actually inspecting it.

[chainchomp]

Early in this thread there was a lot of talk about the magnitude of storage. Considering the following two exerpts from wikipedia.

"Earlier studies from the University of California, Berkeley, estimated that by the end of 1999, the sum of human-produced information (including all audio, video recordings, and text/books) was about 12 exabytes of data.[26]"

"Research from University of Southern California estimates that the amount of data stored in the world by 2007 was 295 exabytes and the amount of information shared on two-way communications technology, such as cell phones in 2007 as 65 exabytes.[28][29]"

- coupled with the 1 yottabyte capacity of that funky new NSA server farm they're building that is only meant for the next 100 years of data collection.

and using the high likelihood of humanity producing transmittable records for the next 100,000 years... it is wholly impractical to assume anything other than a sneaky combination of the brilliant ideas generated so far in this thread, and probably stuff we haven't thought of yet.

From some flakey general knowledge and lots of reading this thread, I would consider a vast series self-repairing factories that produce and repair data-gathering probes, resource mining AI, archival maintenance(from redundancy) AI, and swarm R&D AI - all using gathered material.

These factories would be based inside asteroids which are transported from the inner metalliferous asteroid belts to the ice outer asteroid belt+ in order to seem out of place enough to warrant inspection and mitigate some of the complications of stellar evolution upon something as close as the Earth's SOI and neighbours.

Utilising the compounding unbridled research capabilities of the R&D AI, eventually a matrioshka brain (or similar analogue) would produce some Clarke Tech that would be able to store (almost an extra magnitude above) yottabytes of data in a completely isolated environment, yet somehow has selective manipulability.

So essentially, give aggressive AI control over it's own destiny, and attempt to nail a note to it with the gameplan scrawled on it and a caption saying "please remember us and our achievements in case someone drops by, much love, your squishier, soon to be extinct famly".

That's what I'd do, and a great reason to keep me away from AI-related decision making.

[nhnifong]

Perhaps as a long term sort of passive camera, the satellite could be in a long period orbit, synchronous with the earth. And it had an internal part that was spinning. Once an orbit, a pair of windows would align, exposing a piece of film that was also passively aligned by a slowly spinning part. But honestly, I don't think it will last 5 billion years.

As many have said, the sun is expected to grow to a size which engulfs the earth, putting in in lunar orbit or underground there is not an option. I think a long period comet-like orbit with a perigee outside of mars is probably safe. Making it relatively inclined should decrease it's probability of a collision or a close encounter.

The only power source that will last long enough is the Sun itself. Perhaps the solar panels would have to be shielded, and then only exposed at perigee when sensing and error correction is performed.

[DerekL1963]

Re: shielding an object against physical degradation from impact and radiation damage;

shielding shielding shielding. to prevent data loss to collisions, have more than one. you can also reduce the data resolution to compensate for any high energy ray damage that you cant shield against.

Since your shielding material is subject to the same damage you are shielding the data inside from... over time any "Hey, You! Look Inside For Cool Stuff" labeling you put on the outside will itself be degraded into uselessness. How do make labels that survive as well as the contents?

[Psycix]

Self replicating machines are the key here. You can't expect any sensitive data storage to last for 5 billion years, so even with redundancy I would look into manufacturing and re-writing storage over time.

Another interesting idea might be to send the data to a celestial cluster 2.5 billion lightyears away, have it return due to a complex gravity lensing effect and receive the transmission another 2.5 billion years later.

[lajoswinkler]

Re: shielding an object against physical degradation from impact and radiation damage;

Since your shielding material is subject to the same damage you are shielding the data inside from... over time any "Hey, You! Look Inside For Cool Stuff" labeling you put on the outside will itself be degraded into uselessness. How do make labels that survive as well as the contents?

Labelling is indeed a problem. This movie has tackled some of the problems regarding longterm labeling and I recommend it. It doesn't deal with billions but thousands of years, though it still raises important questions.

Of course, we could resort to huge monolithic labels, but that might not survive 5 billion years of asteroid impacts. Any material would gradually errode to dust.

But we can be smarter than that. We could contaminate the area with some unusual isotope (doesn't have to be a radioisotope) which doesn't exist anywhere else on the Moon. That would attract some interest from an intelligent species and they would investigate it.

Even better - by manipulating the ratios of several rare isotopes we could encode a simple message in binary system. For example you could put the ratios in a prime number order.

I'm presenting a hypothetical cryptography and it indeed can become very complex. An intelligent species or their computers (or both if they're merged) should be able to notice that the composition is unnatural and then crack the code.

[Nuke]

i dont think labeling is an issue. if your probe looks significantly unnatural enough, it will warrant investigation from anyone who happens to spot it. though it may collect enough debris to look like an asteroid.

a bigger issue is once they are inside the vault containing all the storage media, how are they supposed to read it. you would need to describe your data formats mathematically (and this is the easy part). then you need to describe our language in the same way. our languages with phonetic alphabets might be a limitation here, since they are just a way to write down the sounds made by spoken language. if an alien comes along that does not have ears, they may have an issue understanding this. if we use a language like chinese (with an ideographic alphabet), we can assign a number to a character and then convey the meaning of the character with math (and that sounds really hard). you just need to describe one language and then provide a rosetta stone for all the other languages (same text written in all languages), to open the door to understanding all human language.

Edited January 16, 2014 by Nuke

[lajoswinkler]

I was talking about lunar burial sites. Any probe orbit would decay after 5 billion years so that's out of the question...

[Nuke]

regardless if you make it as unnatural looking as possible, it will draw attention.

[randomness5555]

I can't really see anything much lasting that long, some stars wouldn't. Perhaps engraving giant letters into the face of a planet with no atmosphere that is also far from any star, that way they would (in theory) never wear away. The problem would be how to store enough data...

[cypher_00]

A lot of people seem to miss the point.. He wants the data to last even after our Sun dies. This means putting something on the Moon is useless. The Sun will grow so large near the end of it's life that it will eat up Earth and our Moon.

So the data needs to be ejected out of our solar system before that time.

[Xorth Tanovar]

No Lunar storage, no solar system ejection needed.

Use Pluto as the data repository. Far enough out where where our sun's death would most likely not have a big effect. No hazard-ridden cometary orbit that could collide with an inner planet or graze the sun. Given it's distance, it would probably be colder than hell, thus slowing any chemical-based decay of recording medium. Bigger than any man-made satellite, so it's easier to spot. Seed it with the oddball radioisotope or even paint it hot pink to attract attention.

Edited January 17, 2014 by Xorth Tanovar