1000 year data storage for autonomous robotic facility

[Bernhard Kuemel]

Sorry for repost, I posted to sci.electronics before, which does not exist.

Hi!

I'm planning a robotic facility [3] that needs to maintain hardware
(exchange defective parts) autonomously for up to 1000 years. One of the
problems is to maintain firmware and operating systems for this period.
What methods do you think are suitable?

Top priority is it must work about 1000 years. Price is not a big issue,
if necessary.

I thought about this:

ROMs/PROMs, replacing them when checksum fails.

ROM/PROM masters, being copied once a year to flash ROM.

1000 flash ROMs, refreshing once a year from the ones that still have a
valid checksum.

Non electronic masters:

Microfilm/microfiche
HD-Rosetta (ion beam engraved nickel disc)
glass CD/DVD
Paper [2]
punched cards

The drawback of the non electronic masters is their reader system which
can fail mechanically/optically (dust, gears, ...) and requires
electronic components/firmware themselves.

Is it possible to make robots or their spare parts that suffer only
minor degradation when kept as spare parts for 1000 years at good
storage conditions? semiconductors, inductors, (non electrolytic)
capacitors, circuit boards, plastic/metal structures, CCD/CMOS cameras,
actuators, solar cells, thermo couples, etc. Batteries are probably
difficult.


Thanks, Bernhard


[1]
http://www.norsam.com/rosetta.html
http://www.norsam.com/nanorosettawp.html
http://en.wikipedia.org/wiki/HD-Rosetta

[2]
something like http://ronja.twibright.com/optar/

[3]
A cold store to keep humans frozen (vitrified) in LN2 until mind
uploading (
https://en.wikipedia.org/wiki/Mind_uploading#Serial_sectioning ) becomes
possible.

 

[Uwe Hercksen]

Is it possible to make robots or their spare parts that suffer only
minor degradation when kept as spare parts for 1000 years at good
storage conditions? semiconductors, inductors, (non electrolytic)
capacitors, circuit boards, plastic/metal structures, CCD/CMOS cameras,
actuators, solar cells, thermo couples, etc. Batteries are probably
difficult.

Hello,

I think it is too optimistic to believe that all the needed parts will
survive 1000 years, even if stored as spare parts.
All metallic contact surfaces will oxidize. Plastic parts will degrade.
Isolations will fail.

A building that should provide good storage conditions would not survive
this long time.

Bye

 

[Spehro Pefhany]

Sorry for repost, I posted to sci.electronics before, which does not exist.

Hi!

I'm planning a robotic facility [3] that needs to maintain hardware
(exchange defective parts) autonomously for up to 1000 years. One of the
problems is to maintain firmware and operating systems for this period.
What methods do you think are suitable?

Cryogenic storage of parts that can stand thermal cycling and cold
temperatures (should be most of them) should reduce most mechanisms of
failure. For example, liquid nitrogen temperatures (77K) would reduce
aging, in theory, by 2^20 or about 10^6:1, so the items would age
maybe the equivalent of 8 hours in 1000 years. Of course the
warranties will expire in that length of time, and probably all the
companies that made them, and quite possibly many of the countries in
which the companies were located will also be gone.

Also keeping liquid nitrogen around continously for 1000 years on
earth is non-trivial.

Top priority is it must work about 1000 years. Price is not a big issue,
if necessary.

I thought about this:

ROMs/PROMs, replacing them when checksum fails.

ROM/PROM masters, being copied once a year to flash ROM.

1000 flash ROMs, refreshing once a year from the ones that still have a
valid checksum.

Non electronic masters:

Microfilm/microfiche
HD-Rosetta (ion beam engraved nickel disc)
glass CD/DVD
Paper [2]
punched cards

The drawback of the non electronic masters is their reader system which
can fail mechanically/optically (dust, gears, ...) and requires
electronic components/firmware themselves.

Is it possible to make robots or their spare parts that suffer only
minor degradation when kept as spare parts for 1000 years at good
storage conditions? semiconductors, inductors, (non electrolytic)
capacitors, circuit boards, plastic/metal structures, CCD/CMOS cameras,
actuators, solar cells, thermo couples, etc. Batteries are probably
difficult.


Thanks, Bernhard


[1]
http://www.norsam.com/rosetta.html
http://www.norsam.com/nanorosettawp.html
http://en.wikipedia.org/wiki/HD-Rosetta

[2]
something like http://ronja.twibright.com/optar/

[3]
A cold store to keep humans frozen (vitrified) in LN2 until mind
uploading (
https://en.wikipedia.org/wiki/Mind_uploading#Serial_sectioning ) becomes
possible.

 

[hanson]

I'm planning a robotic facility [3] that needs to maintain
hardware (exchange defective parts) autonomously
for up to 1000 years. One of the problems is to maintain
firmware and operating systems for this period.
What methods do you think are suitable?
Top priority is it must work about 1000 years.
Price is not a big issue.
hanson wrote:

Bernie, Bernie, Bernie, you should think about
that much, much more... That "1000 year" thing
was tried during your Great Grandpa's days. It
didn't work too well. Collapsed after a dozend
years.

 

[Joerg]

Hello,

I think it is too optimistic to believe that all the needed parts will
survive 1000 years, even if stored as spare parts.
All metallic contact surfaces will oxidize. Plastic parts will degrade.
Isolations will fail.

A building that should provide good storage conditions would not survive
this long time.

Consult the Romans, they knew how to do it

As others have suggested, a threat analysis is in order and then
consider storage under inert gas. And write the instructions on good
paper, like this:

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

 

[Bernhard Kuemel]

I'm planning a robotic facility [3] that needs to maintain hardware
(exchange defective parts) autonomously for up to 1000 years. One of the
problems is to maintain firmware and operating systems for this period.
What methods do you think are suitable?

Cryogenic storage of parts that can stand thermal cycling and cold
temperatures (should be most of them) should reduce most mechanisms of
failure. For example, liquid nitrogen temperatures (77K) would reduce
aging, in theory, by 2^20 or about 10^6:1, so the items would age
maybe the equivalent of 8 hours in 1000 years.

I'm worried that components or assemblies (circuit boards with
components) crack or break with large temperature changes due to
different thermal expansion coefficients.

Also keeping liquid nitrogen around continously for 1000 years on
earth is non-trivial.

Absolutely. That's what this is all about. Linde machines probably won't
last long enough. Peltier coolers don't cool deep enough. I'm waiting
for adiabatic demagnetization of gadolinium alloy coolers.

Bernhard

 

[Bernhard Kuemel]

Inert gas, sealed capsule, under antartic ice, lots of redundant copies, most
media would probably last 1000 years.

This is about media being used during these 1000 years as a source of
firmware and operating systems to keep the robotic facility functional.

 

[[email protected]]

Consult the Romans, they knew how to do it



As others have suggested, a threat analysis is in order and then

consider storage under inert gas. And write the instructions on good

paper, like this:



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



--

Regards, Joerg



http://www.analogconsultants.com/

Romans ! ?

What about the Egyptians, Indians or even the Chinese? Don't their cultures and inventions go back even farther? Rome is in ruins today. The Great Pyramids are still standing. Perhaps Bernhard needs a more durable material like stone.

 

[Bernhard Kuemel]

Hello,

I think it is too optimistic to believe that all the needed parts will
survive 1000 years, even if stored as spare parts.
All metallic contact surfaces will oxidize.

http://www.norsam.com/nanorosettawp.html claims about nickel data
plates: "Nickel has a much longer life span than microfilm up to
thousands of years." Something like nickel or gold might solve this
problem. Lots of precious metals might attract thieves, though.

Plastic parts will degrade.

Polyethyleneterephtalate (PET) microfilm is thought to last up to 700
years. So maybe Teflon can last 1000 years.

Isolations will fail.

There are teflon insulated cables.

A building that should provide good storage conditions would not survive
this long time.

Egyptian pyramids are over 4000 years old. They may not be in sufficient
condition, but I still have hope that with improved
technology/possibilities we can build something that lasts 1000 years.
But the daily temperature changes in the Australian desert certainly
require very durable building materials. Glass ceramics come to my mind.
http://en.wikipedia.org/wiki/Glass-ceramic

Bernhard

 

[Joerg]

http://www.norsam.com/nanorosettawp.html claims about nickel data
plates: "Nickel has a much longer life span than microfilm up to
thousands of years." Something like nickel or gold might solve this
problem. Lots of precious metals might attract thieves, though.

It's so thin, not worth the theft. To reduce the temptation cover it
with something so it isn't shiny. Vandalism is much more of a concern.

Polyethyleneterephtalate (PET) microfilm is thought to last up to 700
years. So maybe Teflon can last 1000 years.

I'd trust metal on glass and similar structures much more. A lot of
times adverse effects in modern materials only become discovered decades
after the fat. Like the spontaneous combustion that happened in old
movie archives.

There are teflon insulated cables.

Even here it may be best to stay with the true and tried. I have a
transformer that's probably almost 100 years old, the secondary is
copper insulated with cloth. Looks like new. If something shows next to
no decay over a hundred years chances are it may last another 900 years.
With modern plastics we often don't really know.

Egyptian pyramids are over 4000 years old. They may not be in sufficient
condition, but I still have hope that with improved
technology/possibilities we can build something that lasts 1000 years.
But the daily temperature changes in the Australian desert certainly
require very durable building materials. Glass ceramics come to my mind.
http://en.wikipedia.org/wiki/Glass-ceramic

Unless there is a continuous stream of money from pharaohs, king or
admission-paying tourists even pyramids fall apart:

http://philipcoppens.com/nap_art3.html

The other problem is theft of building materials from the actual
buildings. AFAIK it already started with the Ottomans raiding the
pyramids for limestone. Guess they were not willing to pay full retail
pricing ...

BTW, you might want to build the robot with through-hole parts. That
reduces the chance that someone turns it on in 998 years, doesn't work,
and a hair crack is discovered in a big ceramic cap.

I wonder what an extended 1000-year warranty would cost

 

[Tim Williams]

It's likely that whatever scheme you offer, would also prevent access
to the preserved data, thus making regular verification difficult. How
do you know that the data is still there without taking a big risk in
trying to read and verify it? You can bury your time capsule, but you
still couldn't prove that the data is still there.

But maybe you could have some fun anyway.
http://www.smbc-comics.com/index.php?db=comics&id=1915#comic

Tim

 

[Jasen Betts]

Sorry for repost, I posted to sci.electronics before, which does not exist.

Hi!

I'm planning a robotic facility [3] that needs to maintain hardware
(exchange defective parts) autonomously for up to 1000 years. One of the
problems is to maintain firmware and operating systems for this period.
What methods do you think are suitable?

Top priority is it must work about 1000 years. Price is not a big issue,
if necessary.

for the master copy: core rope memory.
you said price not a big issue, that'll test how not-big it really is.

 

[Spehro Pefhany]

I'm planning a robotic facility [3] that needs to maintain hardware
(exchange defective parts) autonomously for up to 1000 years. One of the
problems is to maintain firmware and operating systems for this period.
What methods do you think are suitable?

Cryogenic storage of parts that can stand thermal cycling and cold
temperatures (should be most of them) should reduce most mechanisms of
failure. For example, liquid nitrogen temperatures (77K) would reduce
aging, in theory, by 2^20 or about 10^6:1, so the items would age
maybe the equivalent of 8 hours in 1000 years.

I'm worried that components or assemblies (circuit boards with
components) crack or break with large temperature changes due to
different thermal expansion coefficients.

Probably would not end well with BGA lead-free boards, or QFN boards.

Through hole and gull wing SMT parts with leaded solder seem to do
fine, even with thermal shock. Cycling over the military temperature
range (-55~105°C) is a similar change, however it's not centered at
room temperature, so the stress could be greater.

Absolutely. That's what this is all about. Linde machines probably won't
last long enough. Peltier coolers don't cool deep enough. I'm waiting
for adiabatic demagnetization of gadolinium alloy coolers.

Bernhard

I wonder if you could make something almost passive that would cool
to, say, dry ice temperatures using radiative cooling into a desert
night sky.


Best regards,
Spehro Pefhany

 

[Robert Baer]

Sorry for repost, I posted to sci.electronics before, which does not exist.

Hi!

I'm planning a robotic facility [3] that needs to maintain hardware
(exchange defective parts) autonomously for up to 1000 years. One of the
problems is to maintain firmware and operating systems for this period.
What methods do you think are suitable?

Top priority is it must work about 1000 years. Price is not a big issue,
if necessary.

I thought about this:

ROMs/PROMs, replacing them when checksum fails.

* add - 3 in "parallel" using majority logic for output a and checking.

ROM/PROM masters, being copied once a year to flash ROM.

* see above.

1000 flash ROMs, refreshing once a year from the ones that still have a
valid checksum.

* see above.

** Semiconductor storage is useless in a RAD environment.

Non electronic masters:

Microfilm/microfiche

* Degrades - maybe not as fast as the old acetate movie films, but 100
year life is not realistic (but may be better in a RAD environment).

HD-Rosetta (ion beam engraved nickel disc)

* of ideas mentioned,this seems the best. now,how does it get read?

glass CD/DVD

* maybe good enough for 20-10 years.

Paper [2]

* NOPE! Leather,if kept in a dry environment is at least an order of
magnitude better (eg: Dead Sea scrolls). Proven technology. Proven
characteristics. Good enough for a few thousand years.

punched cards

* See above.

The drawback of the non electronic masters is their reader system which
can fail mechanically/optically (dust, gears, ...) and requires
electronic components/firmware themselves.

Is it possible to make robots or their spare parts that suffer only
minor degradation when kept as spare parts for 1000 years at good
storage conditions? semiconductors, inductors, (non electrolytic)
capacitors, circuit boards, plastic/metal structures, CCD/CMOS cameras,
actuators, solar cells, thermo couples, etc. Batteries are probably
difficult.

* Stay away from anything electronic..mechanical parts if not used
nominally do not wear out and reasonably tolerate a RAD environment.

Thanks, Bernhard


[1]
http://www.norsam.com/rosetta.html
http://www.norsam.com/nanorosettawp.html
http://en.wikipedia.org/wiki/HD-Rosetta

[2]
something like http://ronja.twibright.com/optar/

[3]
A cold store to keep humans frozen (vitrified) in LN2 until mind
uploading (
https://en.wikipedia.org/wiki/Mind_uploading#Serial_sectioning ) becomes
possible.

 

[[email protected]]

I asked a co-worker who was fluent in Chinese about this once. If I
remember right, he said that he could read written Chinese as old as a
couple of thousand years and it would pretty much make sense. He also
said that he could understand written Japanese, but if he tried to
speak it, someone fluent in spoken Japanese wouldn't understand him;
the pronunciation would be wrong.

Also Latin, which was the reason that the legal profession adopted it
as their favored language. Since it's a dead language, it's unlikely
to change.

I figured it was a customer lock-in tactic, like EBCDIC and Word .doc .

1000 years from now, it will look as strange as Olde English.

Pic related: http://xkcd.com/771/ (sfw)

Matt Roberds

 

[[email protected]]

Sorry for repost, I posted to sci.electronics before, which does not exist.

This might sound weird, but consider posting in rec.arts.sf.science
where the topic has been discussed by fairly tech-savvy people re:
blunders in SF stories about data that lasts for very long times to
sustain widely dispersed long-lived cultures.

Hi!

I'm planning a robotic facility [3] that needs to maintain hardware
(exchange defective parts) autonomously for up to 1000 years. One of the
problems is to maintain firmware and operating systems for this period.
What methods do you think are suitable?

First, when you say "autonomous" do you mean absolutely no external
resources (parts, materials, software etc.) are presumed available? If
so, that implies no external sources of corruption can get in- the
facility is must be water/gas/light/(unknown)-tight. What will the
facility use for power? I'm thinking not just storage medium
degradation but data corruption (viruses, language changes re:
obsolescence and so on) due to outside influences as well as say stray
radiation from a plutonium power source or similar.

Threat assessment has been mentioned. Suppose someone side-channel
attacks you by modulating the sunlight falling on your solar
collectors with a virus...

Top priority is it must work about 1000 years. Price is not a big issue,
if necessary.

Consider corruption of data during read/write cycles in anything
from quipu through punch cards and electronic storage media to (I
think) Speff's laser-engraved porcelain. It's the "telephone game"
problem with just one player.

I thought about this:

ROMs/PROMs, replacing them when checksum fails.

ROM/PROM masters, being copied once a year to flash ROM.

1000 flash ROMs, refreshing once a year from the ones that still have a
valid checksum.

Already answered with bit flipping due to cosmic rays, metal
migration, glass creep, and so on. Not good for a few decades much
less centuries. Then there was radioactive ceramic used to encapsulate
memory chips. Oops!

Non electronic masters:

Microfilm/microfiche
HD-Rosetta (ion beam engraved nickel disc)
glass CD/DVD
Paper [2]
punched cards

The drawback of the non electronic masters is their reader system which
can fail mechanically/optically (dust, gears, ...) and requires
electronic components/firmware themselves.
Yep.

Is it possible to make robots or their spare parts that suffer only
minor degradation when kept as spare parts for 1000 years at good
storage conditions? semiconductors, inductors, (non electrolytic)
capacitors, circuit boards, plastic/metal structures, CCD/CMOS cameras,
actuators, solar cells, thermo couples, etc. Batteries are probably
difficult.

Well, think like a customer-service tech (one with brains). Your end-
users, the robots- what tech are they built with? What conditions do
they have to survive for a thousand years- temp-controlled low-
pressure argon atmosphere?

What do your robots use for eyes and ears? If your primary archive
is a big slab of granite/basalt/osmiridium engraved with literal
zeroes and ones corruption is very difficult to envision barring
geological events that would threaten the integrity of the facility,
but can the robots read it?


Mark L. Fergerson

 

[Fred Abse]

Also Latin, which was the reason that the legal profession adopted it as
their favored language. Since it's a dead language, it's unlikely to
change. Therefore legal definitions, interpretations, judgments, and laws
will not change over the years.

Latin isn't a dead language. It is still the official language of the
Vatican City state, spoken daily.

 

[Joerg]

Nope. The dead sea scrolls, like many other ancient documents, were
written on parchment paper and on papyrus. And they were just tucked
into a cave without much other protection.

Ummm... parchment *is* leather. Well, hide, at least. It's an animal
product, not paper.
[/QUOTE]

True, it is an animal product. But some of the dead sea scrolls were
written on papyrus and that is not animal-based.

 

[benj]

Sorry for repost, I posted to sci.electronics before, which does not
exist.

This might sound weird, but consider posting in rec.arts.sf.science
where the topic has been discussed by fairly tech-savvy people re:
blunders in SF stories about data that lasts for very long times to
sustain widely dispersed long-lived cultures.

Hi!

I'm planning a robotic facility [3] that needs to maintain hardware
(exchange defective parts) autonomously for up to 1000 years. One of
the problems is to maintain firmware and operating systems for this
period. What methods do you think are suitable?

I recall some SF story I read a long time ago with a similar line. It was
about a beehive style human culture maintained by self-repairing machines
(much as described here) And then the machines began to stop. Naturally
hilarity ensues...

I would point out that there is a difference here between maintaining
digital data for 1000 years (hard enough to do) and maintaining FIRMWARE
for 1000 years.

In the first case you store the data (say in a refractory cubic
transparent crystal that was built up crystal layer by crystal layer
using defects to store data. The data is only in the CENTER of the object
so even if the outside is dirty and degraded for whatever reason, The
damage can be polished off and the data read. But reading is entirely up
to the skilz of the future generation.

Firmware, on the other hand implies that you are constantly reading and
using the data. This is a whole different kettle of worms. Now the life
expectance and reliability of the READER becomes involved! I found it
interesting that the millennium clock does not display time, only keeps
it! When a human enters the cave (whatever) THEY have to pull the weight
that runs the display. And then the display shuts off until the next
person visits. That greatly reduces all display wear etc. But for
firmware running a robotic facility reading would have to be more or less
regularly.

 

[Jasen Betts]

One can also store bulk binary on nickel sheet by writing code blocks
in hexadecimal, with embedded error correcting codes.

or more compact symbologies like QR or matrix codes.