There is something to be said about using proven technologies. But there is also something to be said about using still supported technologies.
Really, are there producers of 8" floppy media anymore? Are we relying on "new-but-old" stock for occasional replacement needs for these systems? Sounds like that wrench that had to be FedEx'ed between silo bases for Minuteman maintenance:
I seem to remember someone telling me how there are a couple of cottage industries that center around the fact that the US Government has to buy goods that are US-made.
I think ceiling fans are one example. (i.e. There's relatively few US-based ceiling fan manufacturers. For one of them, I think the large chunk of their business is the US Government.)
At one of my jobs, we used CPU (pre-Intel design), made by a defense manufacturer, in our control systems. When mfr. decided to discontinue CPU manufacturing, we bought all the inventory from last few production runs. AFAIK, we are still running those control systems using the same CPUs and hardware for now 30+ years.
Cool, a web page linked in a serious discussion that has copyright 1997 on it. The page was created about half way between the heydey of 8" floppies (late 70s) and today.
What's really sad is I remember 1997 well; I was still in college at the time, and I laughed any time I saw 8" floppies because they were so antiquated. I remember laughing about them in 1991 even because I came across some in that year. I remember also watching WarGames somewhere around that time (high school) and laughing about the 8" floppies in that move. And now more time has passed since those times than there was between the heyday of the 8" floppies and when I laughed about them. I feel really old... but I also really question WTF is going on with our government that they're still using the things. Disk drives don't last forever.
If the NSA can run its own semiconductor fab, then I'm sure the US Gov't can pony up a few million annually to manufacture floppy media for something as critical as its nuclear arsenal.
What about the level of hardware failure with floppy disks?
CF cards seem pretty suitable. They're simple and well established; there's likely to be a supply of them for a long time, and it seems feasible to have a batch custom made if that's necessary 30 years down the line.
I can verify that a floppy disk can resist a long time.
I have 3" 1/4 disks from early 90's that keep working fine, plus this last year, I saw a bunch of old 80's 8" floppy disks working on a old Z80 computer.
There is more to this than "out dated" technology. There is tremendous espionage and malicious activity protection by using outdated technology. The schematics of these systems were never online, making and introducing malware is also prohibitively difficult due to the lack of domain knowledge and expertise on the technology.
There are huge upsides to using outdated technology when it comes to nuclear warheards.
If you're using Windows 10 or MacOS, people know how to find their way around in the absence of documentation. How many people could figure out the command line interface to a 1970s era IBM Series/1 without docs?
It is interesting to learn just how much "technical" debt USA has
911 that is falling apart
Nuclear weapons debacle (and a lot of other systems too)
The 7 trillion infrastructure problem
Everyone needs to shut the hell up about nuclear obsolescence.
If we can just keep the politicians from noticing the need to "modernize" these terrible weapons, then in a few short decades we will have achieved a de facto global disarmament. Everyone can continue manning their silos filled with stale dud warheads, fingers hovering over buttons that no longer work. We'll have all the stratigic stabilizing influences of mutually-assured-destruction, with none of the actual threat-of-destruction.
But engineers and journalists need to shut up and not point out that the emperor's got no nukes.
It's too late. The Obama administration has proposed a one trillion dollar budget to modernize the nuclear arsenal. The inclusion of low-yield nuclear devices is considered especially unnerving as it is believed their use would be easier to justify and could lean to a full nuclear exchange. http://mobile.nytimes.com/2016/01/12/science/as-us-modernize...
Obama also unfortunately tends to act in good faith and with intent, so snark aside if he requested the funding for them he is probably OK with tactical nukes.
No, people that agree with you about nuclear disarmament need to stfu if they want the de facto win you suggest.
Personally I don't think you can keep the stability benefits of MAD if secretly none of the buttons work. Eventually there will be a leak. You are literally suggesting security through obscurity.
But irrespective of my disposition this I still think this is the wrong way to get a "win". If you want disarmament then you should go through the political process and vote for people. Getting "wins" outside of the proper political process is just a bad precedent in general.
But what if one side, realizing its weapons are getting stale, decide that the other side's weapons are staler, and thinks it can launch a first strike with relative impunity?
It reminds me this story, which was told to me as a true story, but I wouldn't bet too much on that.
A soviet bomber gets lost and finds itself over western germany. It has to land on a military airport, which pilots jovially invite the soviet pilots for a drink. While they are drinking, intelligence officers take pictures of the plane, and are laughing at antiquated soviet technology, using bulbs while the US was well into the age of transistors. It's only back in the US that the intelligence officers realised that the soviet were using bulbs not for the lack of transistors, but because they can resist a nuclear explosion...
As far as I remember that story about "vacuum tube hindsight" is from a documentary on History Channel(?) about Soviet pilot Viktor Belenko [1] defecting by landing his Mig 25 in Japan, thereby letting the west get their hands on the most advanced soviet plane at the time [2].
Fun story, but to my knowledge most military computer systems have always been radiation hardened in the nuclear age. Enrico Fermi correctly predicted EMP prior to the first nuclear tests all the way back in 1945, and made sure that the test equipment was shielded. Although, we did explore high altitude nuclear explosions (produce huge amounts of EMP) in the late 50s to early 60s and managed to accidentally fry some of our own early satellites in the process.
They still sell them today. They're called Nuclear Event Detection (NED) chips. It detects a spike in voltage from the air and drops all power to everything, acting as an instantaneous power interrupt for everything you care about.
I used to test them for Lockheed all the time on stuff that flies on stealth aircraft. There is a very specific sequence of events with very tight specs they have to pass, the whole thing is over in 250 ms, and that includes the shutdown duration.
All it's measuring is for a voltage spike. The ones we had had a real input pin and a test input pin. You hook up an O'scope to the output, supply a one-shot signal to the test pin, and check to make sure the output did what it was supposed to in the allotted time.
Large radio antennas can create electromagnetic pulses like what you'd get from a nuke blast. I lived not too far from a site that tested them[1][2]. Just before it was decommissioned there were a lot of news articles written about how they did the testing. I can't find any photos now, but they're out there somewhere.
Would dropping power help? I thought the way the nuclear blast damages electronic circuits was by inducting large electric currants which would burn the components. If that's the case it doesn't really change anything whether the circuit was active or not at the time of the blast.
You can quite easily shield all your components from outside pulses. However, if you're processing signals from the air (like in the case of Radar systems, the ones I worked on), your stuff could quite easily get fried. Like an amp could get overloaded, a processor could get fried, a line could get burned, all kinds of things could go wrong. But the best thing you can do is have everything off. If you're close enough to a nuke that stuff that's off gets fried, your electronics are probably pretty low on your list of priorities.
There are EMP shielding techniques for solid state systems. The reality is the Soviets were very behind technologically in many areas compared to the west. A good example is the bulbous nature of their spacecraft. They simply didn't have the metalworking to make square cutouts and square windows that didn't eventually fail, so everything was round like a boat or a submarine design. Its fairly ahistoric to pretend this was all part of some master plan. It was just the best they could do and playing up the old technology as being magically superior is pretty silly.
The other story you hear is how the Soviets used pencils in space while NASA use spacepens. Pencils break off graphite which could get into controls and pressurized ink cartridges in spacepens are cheap and easy to manufacture. There's no downside to using a spacepen, but lots to using pencils.
Meanwhile, the few live confrontations between US fighters and Soviet made MiGs resulted in dead MiG pilots. Its clear that the higher tech and more innovative approach was superior and continues to be today. US missile defense systems keeping Russian and North Korean aggression from expanding for example.
"Meanwhile, the few live confrontations between US fighters and Soviet made MiGs resulted in dead MiG pilots. Its clear that the higher tech and more innovative approach was superior and continues to be today. US missile defense systems keeping Russian and North Korean aggression from expanding for example."
USSR (Russia) and USA both export junk versions of their technology to their clients, so I can't agree with your conclusion. Actual direct confrontations between the USSR and USA (MiGs in Korea and Vietnam) were on a much more even footing.
I think there are strengths to both approaches. The low-tech and rugged designs of many USSR designs have stood the test of time in many cases, most famously in the AK rifles and Soyuz capsules.
Treasury's master business file, which contains all tax data on individual business income taxpayers, likewise is written in that same assembly language code, which was first used in the 1950s, and maintained on the old-school IBM mainframe.
https://callenq.com is the free market bringing 2016 technology to the IRS to receive prompt customer service.
The article describes the IT systems as "creaky" and referred to the "jaw dropping" GAO report that inspired this article was described as "sobering". So, it would appear that the problem the article referred to is the obsolete systems.
At one time, government engineers were in equal standing with contractor engineers. As defense spending dwindled down after the Vietnam war, defense contractors began to lobby for more control of defense workloads. On a contract, the government engineers would maintain the product and defense contractors did the development. Money set aside for documentation was devoured by the contractors so that only the contractors had the knowledge base of the product. Future weapon system development is the money maker for contractors, so to have them bid on upgrading a weapon system such as this, the cost would be very inflated. So you have this situation where the government cannot organically do the job either because of loss of expertise or documentation of the system. Or it is cost prohibitive using a defense contractor. The expertise for government oversight for contracting is a major problem. As an example, the DMV database system for the state of California. When you look at all the companies today using the cloud and creating huge databases practically overnight, how can projects like this fail. Another example, the Obama Care website. The government at every level, just does not have the expertise to insure that contractors perform to the letter of the contract. And then you have all this lobbying crap by the contractors to contend with. It really comes down to ETHICS. Contractors are not out to give you the most bang for your buck, they want the most bucks for smallest bang. After spending 28 years working with defense contractor engineers, I was not impressed. Unfortunately, I was not impressed with the ethics or the adherence to security concerns of my coworkers either, so I walked away. I would not be so concern with the 8 inch floppies, but more concern with the electronics that use them.
> The government at every level, just does not have the expertise to insure that contractors perform to the letter of the contract.
Ideally, they would have even more expertise to make judgement calls, so that they could rely (in some sense) on the spirit of the contract and not just the letter.
I guess comparisons to public sector and military procurement in other countries (and times) might be instructive.
I think they have just as much experience and technical know-how as their contractors, but have neither the funds or the go-ahead to implement this stuff themselves.
For example, Obama signed an executive order requiring the DoD to use electronic records management systems to store all of their classified data.
The tricky part is that the way it is written, they have to buy this, they're not allowed to write the system that holds their own data.
This is what I observed at McClellan AFB during my time there. No matter how hard I fought to do the work as a government engineer, management continuously turned the work over to contractors.
As to why? I was taken aside and given the most ridiculous reason. If the project fails or overruns, the manager just gets more money to give to the contractor. Being screwed by the contractor was acceptable. But if done organically and there were problems, the manager gets a ton of grief. But this is the government, what terrible fate will come down on the manager? I also noticed that a lot of these managers retired from government service ended up working for the contractors, and the process recycles itself.
This came back to bite McClellan AFB in the arse later. Come BRAC time, the contractors made it an easy choice to close the base by taking the work back their sites at a much reduced cost.
Old storage media (floppies, CDs) have a big advantage over new ones (SD cards, USB sticks, hard drives) - they don't have firmware. Most people didn't notice how over the last decade or so we've moved from having a computer read out data directly from a removable medium to having two computers talking together over a low-level, high-privilege protocol. While there are many benefits to this evolution, it also opens up many serious attack vectors.
Along the same lines, you can visually inspect them. Now, perhaps there's a way to hide some sort of wafer-thin wireless reader and transmitter in a floppy disk that can somehow harvest enough electrical energy to be useful, but that's a long shot.
As for SD cards, ones that are more than just storage (e.g. with wireless technology embedded) already exist.
Even if you could hide a wireless reader and transmitter in a floppy disk, you would still be able to pick it up on X-ray. It would be very difficult to differentiate an SD card with a malicious payload even if one X-rayed it.
I think the answer is no. You have to have a magnetic field to induce changes on a floppy (don't get them near a magnet). An x-ray would require a curved flow to generate a magnetic field (think electromagnet) so the generally direct x-ray flux should be harmless. Of course the equipment and power requirements may make it difficult to avoid magnetic fields. Not an x-ray expert.
My father worked for a very long time on traffic signal systems, and a lot of their old technology was a simple EEPROM chip which they wrote a program for the signals, burned to a chip that costed a pittance, and that was it. I remember him telling me when they changed to a new system based on (i think) PCMCIA interfaces and the complexity of deployment and cost were magnitudes greater for no real benefit.
If its an embedded system and the cost and work of supporting it is less than the cost and deployment of a new system, whats the point in updating it?
>If its an embedded system and the cost and work of supporting it is less than the cost and deployment of a new system, whats the point in updating it?
The problem is support and maintenance. Things wear out eventually, and if you can't get the parts to fix them any more, then your whole system fails.
This really shouldn't be a problem with EEPROMs, however; I'm pretty sure you can still buy those, and the programmers for them too. But 8" floppy disks, and the drives to read them, are another matter.
That's true but not why they use them. You can easily restrict the interfaces of many forms of media where firmware is hardcoded in ASIC as a state machine or restricted at interface points. The military also has systems available to do exactly that from Rockwell-Collins, General Dynamics, etc.
Nah, this is a scary example of critical capabilities depending on legacy hardware people might not dven make that they're afraid to replace. Sandia at least modernized some of our nuclear system. They need to deal with this, too.
What's wrong with 8" floppy disks? They seem easy to manufacture (same as a VHS tape but cut differently), are pretty reliable if you keep them away from magnets (Which is common sense), and don't really have a shelf life problem until the plastic gets brittle. Hell, if they degrade, you can even retrieve the undamaged parts with no fuss.
The problem is trusting such a critical function to ancient hardware that's bound to wear out at some point, and for which you can't buy a replacement. Disk drives don't last forever; they're fairly complex machines with moving parts.
Pages 15 to 19 are what the CNBC article picked up on:
This system coordinates the operational functions of the nation’s nuclear forces. This system is running on an IBM Series/1 Computer—a 1970s computing system—and uses 8-inch floppy disks - from page 19
The IBM Series/1 isn't that ancient (well for me anyway) and was comparable to 16 bit Data General (I trained on Data General Nova 3's, 4's and Eclipse S/130's back in the mid-80's) and DEC mini's of the day:
As long as you keep up with the usual preventative maintenance (filter changing etc), these old boxes just keep on working. Sometime in the early 2000's a buddy of mine still had a preventative maintenance contract on a DG Nova 3 [0] attached some astronomy kit. John asked if I fancied a day out to see this machine and go along with him on his PM visit. I hadn't seen an old Nova or Eclipse machine since the early 90's, so why the hell not? When I asked why such an old machine (probably manufactured in '75 or '76) hadn't been replaced with a faster smaller box (e.g. top end PC server type of thing), the reason was at the time nothing could come close to the data channel speeds this old box could sustain when collecting data from their astronomy thing.
If anyone from St Andrews astronomy department knows about this old machine I'd love to know what happened to it.
I wonder what the cost is to maintain uber legacy systems like that? Like at what point does it teeter over to being the most expensive thing like it? (while remaining less capable)
I'd rather have floppy disks and air-gapped systems for our nuclear arsenal than have them all connected to an insecure network (which is pretty much any network these days).
I take the government's actions very seriously. I left the country, wife and kids and all, because I got tired of it.
I respect Jon Oliver's work. I think he makes discussions about things like this more approachable to more people than any article cnbc will every publish.
In the words of Joel Spolsky, "code doesn't rust."
Who cares how old the code or the medium is? If it's tested and working, and it's used for nuclear effing weapons, I'm happy to leave it in my pile of tech debt.
80 comments
[ 4.3 ms ] story [ 117 ms ] threadReally, are there producers of 8" floppy media anymore? Are we relying on "new-but-old" stock for occasional replacement needs for these systems? Sounds like that wrench that had to be FedEx'ed between silo bases for Minuteman maintenance:
http://abcnews.go.com/Politics/us-wrench-nuclear-bases/story...
It would appear so, and in the US, yet: http://www.athana.com/html/diskette.html#8
I think ceiling fans are one example. (i.e. There's relatively few US-based ceiling fan manufacturers. For one of them, I think the large chunk of their business is the US Government.)
I have a feeling this may be an example of that.
That site and I are about the same age.
https://en.wikipedia.org/wiki/RCA_1802
http://gaisler.com/index.php/products/components/gr712rc
Off-the-shelf for government even more so:
http://www.rockwellcollins.com/~/media/Files/Unsecure/Produc...
https://www.cs.utexas.edu/users/jared/ssp-hase-submission.pd...
All the folks who are whining about how we should be using USB sticks should Google "stuxnet."
CF cards seem pretty suitable. They're simple and well established; there's likely to be a supply of them for a long time, and it seems feasible to have a batch custom made if that's necessary 30 years down the line.
I have 3" 1/4 disks from early 90's that keep working fine, plus this last year, I saw a bunch of old 80's 8" floppy disks working on a old Z80 computer.
There are huge upsides to using outdated technology when it comes to nuclear warheards.
If we can just keep the politicians from noticing the need to "modernize" these terrible weapons, then in a few short decades we will have achieved a de facto global disarmament. Everyone can continue manning their silos filled with stale dud warheads, fingers hovering over buttons that no longer work. We'll have all the stratigic stabilizing influences of mutually-assured-destruction, with none of the actual threat-of-destruction.
But engineers and journalists need to shut up and not point out that the emperor's got no nukes.
Personally I don't think you can keep the stability benefits of MAD if secretly none of the buttons work. Eventually there will be a leak. You are literally suggesting security through obscurity.
But irrespective of my disposition this I still think this is the wrong way to get a "win". If you want disarmament then you should go through the political process and vote for people. Getting "wins" outside of the proper political process is just a bad precedent in general.
A soviet bomber gets lost and finds itself over western germany. It has to land on a military airport, which pilots jovially invite the soviet pilots for a drink. While they are drinking, intelligence officers take pictures of the plane, and are laughing at antiquated soviet technology, using bulbs while the US was well into the age of transistors. It's only back in the US that the intelligence officers realised that the soviet were using bulbs not for the lack of transistors, but because they can resist a nuclear explosion...
[1] https://en.wikipedia.org/wiki/Viktor_Belenko
[2] https://en.wikipedia.org/wiki/Mikoyan-Gurevich_MiG-25#Wester...
I used to test them for Lockheed all the time on stuff that flies on stealth aircraft. There is a very specific sequence of events with very tight specs they have to pass, the whole thing is over in 250 ms, and that includes the shutdown duration.
It's now a wildlife refuge..
[1]: http://www.globalsecurity.org/military/facility/woodbridge.h... [2]: http://eservice.pwcgov.org/library/digitallibrary/hsdw/A_Fol...
The other story you hear is how the Soviets used pencils in space while NASA use spacepens. Pencils break off graphite which could get into controls and pressurized ink cartridges in spacepens are cheap and easy to manufacture. There's no downside to using a spacepen, but lots to using pencils.
Meanwhile, the few live confrontations between US fighters and Soviet made MiGs resulted in dead MiG pilots. Its clear that the higher tech and more innovative approach was superior and continues to be today. US missile defense systems keeping Russian and North Korean aggression from expanding for example.
USSR (Russia) and USA both export junk versions of their technology to their clients, so I can't agree with your conclusion. Actual direct confrontations between the USSR and USA (MiGs in Korea and Vietnam) were on a much more even footing.
I think there are strengths to both approaches. The low-tech and rugged designs of many USSR designs have stood the test of time in many cases, most famously in the AK rifles and Soyuz capsules.
https://callenq.com is the free market bringing 2016 technology to the IRS to receive prompt customer service.
Ideally, they would have even more expertise to make judgement calls, so that they could rely (in some sense) on the spirit of the contract and not just the letter.
I guess comparisons to public sector and military procurement in other countries (and times) might be instructive.
For example, Obama signed an executive order requiring the DoD to use electronic records management systems to store all of their classified data.
The tricky part is that the way it is written, they have to buy this, they're not allowed to write the system that holds their own data.
(See JITC 5015.2.)
As for SD cards, ones that are more than just storage (e.g. with wireless technology embedded) already exist.
If its an embedded system and the cost and work of supporting it is less than the cost and deployment of a new system, whats the point in updating it?
The problem is support and maintenance. Things wear out eventually, and if you can't get the parts to fix them any more, then your whole system fails.
This really shouldn't be a problem with EEPROMs, however; I'm pretty sure you can still buy those, and the programmers for them too. But 8" floppy disks, and the drives to read them, are another matter.
Nah, this is a scary example of critical capabilities depending on legacy hardware people might not dven make that they're afraid to replace. Sandia at least modernized some of our nuclear system. They need to deal with this, too.
https://www.youtube.com/watch?v=lTVgJ8Sb34E
Who wins by moving to SD?
http://www.gao.gov/assets/680/677454.pdf
Pages 15 to 19 are what the CNBC article picked up on:
This system coordinates the operational functions of the nation’s nuclear forces. This system is running on an IBM Series/1 Computer—a 1970s computing system—and uses 8-inch floppy disks - from page 19
The IBM Series/1 isn't that ancient (well for me anyway) and was comparable to 16 bit Data General (I trained on Data General Nova 3's, 4's and Eclipse S/130's back in the mid-80's) and DEC mini's of the day:
https://en.wikipedia.org/wiki/IBM_Series/1
As long as you keep up with the usual preventative maintenance (filter changing etc), these old boxes just keep on working. Sometime in the early 2000's a buddy of mine still had a preventative maintenance contract on a DG Nova 3 [0] attached some astronomy kit. John asked if I fancied a day out to see this machine and go along with him on his PM visit. I hadn't seen an old Nova or Eclipse machine since the early 90's, so why the hell not? When I asked why such an old machine (probably manufactured in '75 or '76) hadn't been replaced with a faster smaller box (e.g. top end PC server type of thing), the reason was at the time nothing could come close to the data channel speeds this old box could sustain when collecting data from their astronomy thing.
If anyone from St Andrews astronomy department knows about this old machine I'd love to know what happened to it.
[0]: https://en.wikipedia.org/wiki/Data_General_Nova
This shows me ignorance, not John Oliver, but the person who chooses to use it to satisfy some valid and real point.
I take the government's actions very seriously. I left the country, wife and kids and all, because I got tired of it.
I respect Jon Oliver's work. I think he makes discussions about things like this more approachable to more people than any article cnbc will every publish.
Who cares how old the code or the medium is? If it's tested and working, and it's used for nuclear effing weapons, I'm happy to leave it in my pile of tech debt.
They were surprised to see things like "an antique computer without a monitor."