Computing for preppers
October 9, 2016 12:18 PM Subscribe
Reading more dire climate change posts made me idly wonder: what could one build or buy today that would still be usable as a general-purpose computing device in 100 years? What about 1000 years?
Searching for "sustainable computing" and such gives me stuff that generally assumes the infrastructure is around to replace components when they fail, and so focuses on modularity, (ecological) cost of manufacture/disposal, etc.
But suppose manufacturing new stuff becomes hard, owing to some form of systemic collapse, or people just lose access to manufactured goods? Let's say that these future people will continue having a way to supply this device with low voltage power (from a simple generator or something) and maybe patch bits together with new copper wiring, but aside from that they have to rely on components/systems which already exist.
What tradeoffs would have to be made? Have people tried it?
Searching for "sustainable computing" and such gives me stuff that generally assumes the infrastructure is around to replace components when they fail, and so focuses on modularity, (ecological) cost of manufacture/disposal, etc.
But suppose manufacturing new stuff becomes hard, owing to some form of systemic collapse, or people just lose access to manufactured goods? Let's say that these future people will continue having a way to supply this device with low voltage power (from a simple generator or something) and maybe patch bits together with new copper wiring, but aside from that they have to rely on components/systems which already exist.
What tradeoffs would have to be made? Have people tried it?
Also depends on what you mean by "computing" - ie what tasks do you want to automate for the next 100 or 1000 years? Also, would you accept mechanical computing as a thing? Most common calculations (compound interest, correlation coefficients, etc) can be done with (relatively) simple mechanical devices.
Otoh, if you want large data sets, complex equations (fluids, etc), crypto...?
posted by Mogur at 12:45 PM on October 9, 2016
Otoh, if you want large data sets, complex equations (fluids, etc), crypto...?
posted by Mogur at 12:45 PM on October 9, 2016
I feel almost silly asking this, but what does computing mean, without Internet?
Serious scientific and engineering pursuits notwithstanding (for that you'd be best off just sharpening your calculus and other math skills), I'm thinking back to what I did before the Internet was always-on. Word processing. Spreadsheets. Reading. Games. None of those things really require computers.
posted by rokusan at 12:56 PM on October 9, 2016 [1 favorite]
Serious scientific and engineering pursuits notwithstanding (for that you'd be best off just sharpening your calculus and other math skills), I'm thinking back to what I did before the Internet was always-on. Word processing. Spreadsheets. Reading. Games. None of those things really require computers.
posted by rokusan at 12:56 PM on October 9, 2016 [1 favorite]
I used to own a pocket calculator which had a solar cell mounted on it. It didn't need batteries.
posted by Chocolate Pickle at 1:01 PM on October 9, 2016 [1 favorite]
posted by Chocolate Pickle at 1:01 PM on October 9, 2016 [1 favorite]
I still have (and use) a solar-powered calculator my mom bought me when I was going into 8th grade. That was 36 years ago.
posted by BlahLaLa at 1:14 PM on October 9, 2016 [2 favorites]
posted by BlahLaLa at 1:14 PM on October 9, 2016 [2 favorites]
You'll be interested in the Long Now Foundation.
Here's a start to thinking about the computers. Discrete transistors can be made by hand.
Also look up the TRS-80 pocket computer.
Engineering has a rule where a 10°C increase in temperature cuts the life in half. If a processor running at 70°C lasts 5 years (that's a pretty conservative number), making something similar but much slower so that it runs at 30°C maximum will last at least 80 years. Displays would need to be discrete LED segments (like the TRS-80) or VFD, which both will last much longer than monitors. Keyboards would need discrete repairable mechanical switches, though storing enough capacitive-touch sensor ICs would be less work. Capacitors would need to be a type without electrolyte or you would have to know how to make more; there are kinds that will last basically forever, they just cost more and take up more space.
Presumably you'd want something like Wikipedia stored. Flash chips are only rated to keep their data for 20 years. Anybody who used CD-Rs knows they don't last for long. Tape drives might work so long as you have enough spares, but ones simple enough to repair would take quite a few to store Wikipedia. Punch cards would be too slow to search (and, at that point, you might as well just print it out if you can't search). Pressed CDs should last 'forever', so if you can design a circuit using simple components that's able to read CDs, or design a less-dense version of the same technology and use derated laser components, you could build readers that would last.
posted by flimflam at 1:27 PM on October 9, 2016
Here's a start to thinking about the computers. Discrete transistors can be made by hand.
Also look up the TRS-80 pocket computer.
Engineering has a rule where a 10°C increase in temperature cuts the life in half. If a processor running at 70°C lasts 5 years (that's a pretty conservative number), making something similar but much slower so that it runs at 30°C maximum will last at least 80 years. Displays would need to be discrete LED segments (like the TRS-80) or VFD, which both will last much longer than monitors. Keyboards would need discrete repairable mechanical switches, though storing enough capacitive-touch sensor ICs would be less work. Capacitors would need to be a type without electrolyte or you would have to know how to make more; there are kinds that will last basically forever, they just cost more and take up more space.
Presumably you'd want something like Wikipedia stored. Flash chips are only rated to keep their data for 20 years. Anybody who used CD-Rs knows they don't last for long. Tape drives might work so long as you have enough spares, but ones simple enough to repair would take quite a few to store Wikipedia. Punch cards would be too slow to search (and, at that point, you might as well just print it out if you can't search). Pressed CDs should last 'forever', so if you can design a circuit using simple components that's able to read CDs, or design a less-dense version of the same technology and use derated laser components, you could build readers that would last.
posted by flimflam at 1:27 PM on October 9, 2016
rokusan: There is always packet radio, which requires no shared infrastructure. If you could create some autarkic social units which could maintain PCs and transceivers of the kind HAM radio enthusiasts use today, and could agree on some rough "rules of the [spectrum] road" (which HAMs basically do already, within the de jure bounds set by the ITU), they could have an "internet" for a very long time.
posted by phrontist at 1:29 PM on October 9, 2016
posted by phrontist at 1:29 PM on October 9, 2016
rokusan: And there are plenty of fun/edifying/important things you could do with computers even if you don't have a computer network. Computer-aided drafting, artistic rendering, music composition; hypertext note-taking, space-efficient document storage.
posted by phrontist at 1:33 PM on October 9, 2016
posted by phrontist at 1:33 PM on October 9, 2016
I wonder if radiation hardened CPUs/memory would have a longer operational life in low-radiation environments because radiation-fault-tolerance and thermal/mechanical-fault tolerance are basically the same.
posted by phrontist at 1:36 PM on October 9, 2016
posted by phrontist at 1:36 PM on October 9, 2016
Right, phronitist, but those are just more examples like mine -- the computer isn't necessary for them, and the huge work of maintaining a computer for drafting or music-making or note-taking is almost certainly not worth the effort, when you can do it the slow, old-fashioned way with rudimentary tools that will last 1000 years.
(I do like the idea of bootstrapping a packet radio replacement-net, but that's getting out there.)
posted by rokusan at 2:34 PM on October 9, 2016
(I do like the idea of bootstrapping a packet radio replacement-net, but that's getting out there.)
posted by rokusan at 2:34 PM on October 9, 2016
You need to be clear on what the conditions are of this thought experiment, i.e. What are you prepping for?
It sounds like you're looking for something that's still going to work in a scenario where there is a complete lack of infrastructure, and all that's available is what can be provided on a individual human or small group scale. At which point the question isn't what type of computing is possible, but what would be relevant and useful.
I think the answer is: none. The question isn't about what tech will last unserviced or is the most robust or whatever.
The underlying need that a modern car solves for is getting from A to B faster and hauling more than can be done by a single person. A horse or donkey also satisfies this need.
The underlying need for email is fast (maybe asynchronous) communication at a distance. So maybe something in between smoke signals or semaphore or a letter carried by someone.
So with slides rules, the abacus, pen and paper calculations, etc.
Is the underlying need for simulating protein folding relevant in a world where there are no conferences, no peers, no labs, no manufactories or government research grants...?
posted by danny the boy at 3:08 PM on October 9, 2016
It sounds like you're looking for something that's still going to work in a scenario where there is a complete lack of infrastructure, and all that's available is what can be provided on a individual human or small group scale. At which point the question isn't what type of computing is possible, but what would be relevant and useful.
I think the answer is: none. The question isn't about what tech will last unserviced or is the most robust or whatever.
The underlying need that a modern car solves for is getting from A to B faster and hauling more than can be done by a single person. A horse or donkey also satisfies this need.
The underlying need for email is fast (maybe asynchronous) communication at a distance. So maybe something in between smoke signals or semaphore or a letter carried by someone.
So with slides rules, the abacus, pen and paper calculations, etc.
Is the underlying need for simulating protein folding relevant in a world where there are no conferences, no peers, no labs, no manufactories or government research grants...?
posted by danny the boy at 3:08 PM on October 9, 2016
I know of two places that have Commodore 64s in active use. That's over 30 years, probably close to a third of the way to your 100 years hypothetical. My impression is that if things are running well within temperature spec, the only solid state electronic components that are really likely to fail are capacitors.
So, yeah: Post-apocalypse, I'm thinking the hackers will be using Apple ][s and C64s. Maybe with hand-made capacitors.
posted by straw at 4:26 PM on October 9, 2016
So, yeah: Post-apocalypse, I'm thinking the hackers will be using Apple ][s and C64s. Maybe with hand-made capacitors.
posted by straw at 4:26 PM on October 9, 2016
A place I used to work was running PS/2 in 2012. I have a client now who's still running Windows 98. And of course, half the world still uses XP. So old stuff can last a while.
Backward compatibility is a problem. Future software updates won't be compatible with old hardware, so you'd have to have all your software locally installed, and make sure everything has the functionality you'll need for 100 years, since you won't be updating it.
One of the biggest things, if your computer is going to be networked, is antivirus software. You'll probably want to prompt for confirmation when you access the network, and limit (or even prohibit) what programs can be installed.
You'd also be wise to incorporate redundancy. Don't just back up your drive. Buy several different computers, and rotate everything every few years.
Ultimately, though, like others have said, there's not much you'd actually need a computer for. Most things could be done more reliably by other tech. The computer is just more convenient in our current world.
posted by kevinbelt at 6:31 PM on October 9, 2016
Backward compatibility is a problem. Future software updates won't be compatible with old hardware, so you'd have to have all your software locally installed, and make sure everything has the functionality you'll need for 100 years, since you won't be updating it.
One of the biggest things, if your computer is going to be networked, is antivirus software. You'll probably want to prompt for confirmation when you access the network, and limit (or even prohibit) what programs can be installed.
You'd also be wise to incorporate redundancy. Don't just back up your drive. Buy several different computers, and rotate everything every few years.
Ultimately, though, like others have said, there's not much you'd actually need a computer for. Most things could be done more reliably by other tech. The computer is just more convenient in our current world.
posted by kevinbelt at 6:31 PM on October 9, 2016
Best answer: I think a personal pocket library filled with searchable information about the world (physics, chemistry, biology, etc) and technology (medicine, engineering, agriculture, etc) and entertainment could offer considerable value in a post-apocalyptic future. Paper libraries are massive which makes it a giant task to keep them protected from the elements, and they can't really be moved. That's really a data-storage and data-access issue, though computers are how we currently do it.
Archival data retention is a fairly researched field at least. Perhaps a reading device of sufficient usefulness could be made from discrete transistors? (Discrete transistors are likely to last a lifetime, and could potentially be replaced with hand-made transistors or old transistors if enough knowledge remains in the world to diagnose the fault, so maybe it needs two readers - each to provide access to knowledge on how to fix the other).
Perhaps it might be better to sacrifice the search indexing and have a purely optical reader and store the library in microscopic engraving instead. In the microscopic world, you can get a long way with just a droplet of water and a good squint.
posted by anonymisc at 7:12 PM on October 9, 2016
Archival data retention is a fairly researched field at least. Perhaps a reading device of sufficient usefulness could be made from discrete transistors? (Discrete transistors are likely to last a lifetime, and could potentially be replaced with hand-made transistors or old transistors if enough knowledge remains in the world to diagnose the fault, so maybe it needs two readers - each to provide access to knowledge on how to fix the other).
Perhaps it might be better to sacrifice the search indexing and have a purely optical reader and store the library in microscopic engraving instead. In the microscopic world, you can get a long way with just a droplet of water and a good squint.
posted by anonymisc at 7:12 PM on October 9, 2016
Presumably for many computing applications you'd also need a long lasting display monitor that won't burn out, too, which could also be tough.
posted by Pryde at 8:38 PM on October 9, 2016
posted by Pryde at 8:38 PM on October 9, 2016
Best answer: You're talking about how long you could keep consumer tech going after the end of Western Civ, right? Like, could you download a whole bunch of survivalist texts onto a thousand identical Kindles, and then, I dunno, store them in nitrogen or in mineral oil or whatever and decant them as you needed them. How long could you keep it going? Could your post-apocalyptic grandchildren be rolling their own solar panels according to instructions read off of of tablets you hoarded down in the salt mines? Could you salvage enough copper wire to roll your own electrical generation? Could you make your own nickel-iron batteries? Consumer computer tech is designed to fail. I have had a bunch of old (digital) music tech fail on me recently due to leaky capacitors. I have to add the parenthetical because my old analog musical tech that is older than me and still going strong.
I've thought a lot about this, but I don't have any easy answers.
posted by BrunoLatourFanclub at 9:11 PM on October 9, 2016
I've thought a lot about this, but I don't have any easy answers.
posted by BrunoLatourFanclub at 9:11 PM on October 9, 2016
Response by poster: As I say it was an idle question but let's say these devices are worthwhile if only for cultural reasons so people can still play snake or perform sequenced music or devise new functional programming languages or as suggested up thread determine the most awesome chilli recipe.
Would a C64 outlive a raspberry pi?
posted by doiheartwentyone at 12:27 AM on October 10, 2016
Would a C64 outlive a raspberry pi?
posted by doiheartwentyone at 12:27 AM on October 10, 2016
The reasons electronics fail are fairly well understood, but for consumer goods some causes of failure are not worth the cost to better protect against. So at a guess I think the idea for a lifetime-plus general computer would be to use as many of the known failure-prevention methods as is feasible. Eg use modern radiation-hardened electronics, and run them at as low a voltage and hz as you can (so it runs cool), with larger dry capacitors instead of compact electrolytic ones. Use conformal coatings, slap it in a hermetically sealed argon-filled vibration-damped case with over-engineered passive heat dissipation and thermal throttling, move all batteries/memory-backup to the outside of the case, avoid cables and if you can't avoid them then use finger-screw connectors for every wire strand, and so on. Store in a temperature-stable cool location free of vibration. Use infrequently. It could(should?) last lifetimes.
I think a really robust display that works indefinitely would be possible with discrete under-powered LEDs and hardened transistors, though that would be pretty blocky and ugly. The displays of microwave ovens from the 80s still seem to work fine and that's just consumer-grade stuff, so I'm guessing there probably are some off-the-shelf displays that score well in longevity.
E-ink isn't very robust over the long term, but apparently TFT is pretty good. This paper (which I didn't bother to properly read) seems to put the MTBF (mean time before failure) of an aircraft LCD display at 84 years, and I think our hypothetical device may be a less hostile environment than an airframe (less pressure variation, vibration, time in use, etc)
I think you could do a pretty good job with off-the-shelf parts and processes developed for critical systems needing longterm reliability, but it would take a fair bit of effort and research, and no doubt a lot of it is very specialized and hard to obtain. Kickstarter anyone? :)
posted by anonymisc at 3:16 AM on October 10, 2016
I think a really robust display that works indefinitely would be possible with discrete under-powered LEDs and hardened transistors, though that would be pretty blocky and ugly. The displays of microwave ovens from the 80s still seem to work fine and that's just consumer-grade stuff, so I'm guessing there probably are some off-the-shelf displays that score well in longevity.
E-ink isn't very robust over the long term, but apparently TFT is pretty good. This paper (which I didn't bother to properly read) seems to put the MTBF (mean time before failure) of an aircraft LCD display at 84 years, and I think our hypothetical device may be a less hostile environment than an airframe (less pressure variation, vibration, time in use, etc)
I think you could do a pretty good job with off-the-shelf parts and processes developed for critical systems needing longterm reliability, but it would take a fair bit of effort and research, and no doubt a lot of it is very specialized and hard to obtain. Kickstarter anyone? :)
posted by anonymisc at 3:16 AM on October 10, 2016
Serious scientific and engineering pursuits notwithstanding (for that you'd be best off just sharpening your calculus and other math skills)
Computing reliable numerical results in reasonable times is a whole other discipline from calculus: back before electronics, a computer was a person. The Manhattan project used hundreds of them organized in groups and equipped with paper, pencil and slide rules.
If you are rebuilding the human civilization from scratch, be sure to stock a good computational methods textbook, a set of slide rules and a maybe some mechanical calculators (something like this).
posted by Dr Dracator at 6:15 AM on October 10, 2016
Computing reliable numerical results in reasonable times is a whole other discipline from calculus: back before electronics, a computer was a person. The Manhattan project used hundreds of them organized in groups and equipped with paper, pencil and slide rules.
If you are rebuilding the human civilization from scratch, be sure to stock a good computational methods textbook, a set of slide rules and a maybe some mechanical calculators (something like this).
posted by Dr Dracator at 6:15 AM on October 10, 2016
I've probably told this before, but my father was working on a military project as a draftsman in the late 50s and needed to calculate a long spiral with increasing pitch. He needed the values for some hyperbolic trig functions to more decimal places than could be found in tables. He ended up going to a university math department, where they were in handwritten in a compendium, no doubt calculated by some long suffering grad students.
posted by 445supermag at 8:32 PM on October 10, 2016
posted by 445supermag at 8:32 PM on October 10, 2016
This thread is closed to new comments.
posted by dilaudid at 12:26 PM on October 9, 2016 [7 favorites]