Slashdot Mirror
Join the Hunt For the Government's Oldest Computer (muckrock.com)
v3rgEz writes: As the saying goes, 'If it ain't broke, don't fix it.' If a machine is doing its job, reliably and without error, then common sense dictates that you just shouldn't mess with it. This is doubly true for computers and quadruply true for government computers. This lends itself to an obvious question: what's the government computer most in need of an upgrade? MuckRock has launched a new FOIA project to find out, and has already started receiving some interesting results.
According to the FAA, just knowing what kinds of computers the FAA is using would endanger the security of national air traffic. That's pretty bad, both for this project and my confidence in our air traffic system.After all, despite my vague wording, the FAA found 11 pages of documents responsive to the same request. But! They refused to release any of their records to me, citing the blanket Exemption 3 because they deemed, "disclosure would be detrimental to the safety of persons traveling in air transportation."
He does say in the article that some agencies have confirmed mainframes from circa 1970, but doesn't say which specifically. It should be interesting to see how this project goes over the next few months.
I'd be curious if he has a way to differentiate active, in-use machines vs. old stuff that may just still be in the inventory roles. I helped a bit with my previous company's effort to clear out old storage and inventory and there was some pretty old equipment that had been in the closet for close to a decade, plus some stuff in their inventory records that had long since disappeared.
So I wonder if that Gateway Liberty 2000 is actually still sitting on someone's desk where the toil away with WordPerfect and Windows 3.1 every day, or if it got tossed/walked out of the building in the late 90s and no one bothered to update the records.
I get highly annoyed when i get an e-mail it takes me 15 minutes to find an answer to, i can imagine how the poor office drone who has to find all this frivolous information feels, especially when it's at the behest of somebody merely curious.
And that's assuming he can find it on his own, once he has to start bothering several different departments people tend to develop a deep resentment towards you...and trust me i know this from experience.
A couple of years ago I heard of a late-70's VAX still being used at a small power plant. To my knowledge it controlled some sort of HVAC systems. Another old system, one I've actually seen, was a mid-80's computer of unknown make/model used to control traffic lights in a small city. It's funny, or actually impressive, to see such old systems still in use. The old-school guys that keep them running tell nice stories about flea market and eBay scavenging.
-SR
When a department gets downsized for one reason or another at my government job, the department manger is supposed to turn in all the unused workstations for redeployment or recycling. They typically don't. A favorite hiding space is the utility closet inside a women restroom. All the field techs are male. Go figure.
"If a machine is doing its job, reliably and without error, then common sense dictates that you just shouldn't mess with it."
"what's the government computer most in need of an upgrade?"
You've just given a great reason why some hardware is still in use, it works.
Why turn around and conclude that it needs an upgrade?
"knowing which agencies are running hardware older than I am is important"
Sure, for a very loose definition of important.
What's with. the random periods. in your sentences?
A double space on the iPhone virtual keyboard becomes a period. Type too fast, random periods appear.
I've got boxes of vacuum tubes that could be available for the right price.
Just sayin'.
Have gnu, will travel.
The Central Logic Control Computer (CLC) at AN/FPQ-16 Perimeter Acquisition Radar Attack Characterization System (PARCS) was assembled at Los Alamos in 1968 and installed at Cavalier AFS sometime before 1972. It remains installed and operational - although a replacement computer system is now also installed. The CLC was probably the fastest computer in the world when it was built and may have still been the fastest computer in the world in the lat 1970s. It is compose of 40+ racks of water cooled equipment.
http://www.srmsc.org/par2010.html
https://en.wikipedia.org/wiki/AN/FPQ-16_PARCS
I hate that goddamn phrase. When the inevitable time comes when suddenly the old system *does* break, it's no longer under any support, nobody's left at the company who knows how it works, there's no budget for a modern replacement, and it has to be fixed in four hours or the company goes bankrupt. Been there, done that, ate the T-shirt after hours of working with no break for food.
People who say "if it ain't broke, don't fix it" are the same idiots who brag about uptime.
Pro tip: *every* system is broken. The trick is being able to repair or work around the broken parts without disruption, not to just seal it behind a wall and rediscover it years later when trying to track down what's still pinging.
I use Macs for work, Linux for education, and Windows for cardplaying.
Back in the 1980s my then-employer shared the cost with the FAA of updating IBM's 1401 emulator to the latest architecture (IIRC whatever was the half generation between 370 and 390), so I wouldn't count on that 360 being the oldest!
sPh
I've found that sometimes old systems leave footprints that last far beyond the computers themselves.
For example, a couple of years ago we had this working networked system that we wanted to upgrade one computer of. The issue was that the protocol used to talk between the systems was a custom network layer written on top of a serial protocol called DR-11W. The cards were rather hard to find, the hardware very finicky to get talking right, and finding good docs for our custom layer was a real challenge.
I eventually found out in researching it that DR-11W was in fact the serial printer port on the original PDP-11's back in the 70's. Neither machine was a PDP-11, but since every upgrade ever done was one computer at a time, we've had to maintain this PDP-11 printer port communications interface for the last 40 years. The protocol even required converting all floating-point values to IBM's old format even though neither side used that format! The conversion's not trivial either.
Our one vendor for these cards has since gone out of business. The story I heard is that they lost their building lease, and didn't feel like it was worth it to move. So it looks like next upgrade, the PDP-11 printer port networking may finally die.
The moral here is that just because the 40-year-old computer may be physically gone, it might not really be gone.
I used to run a pair hobbyist/enthusiast sites for fans of DEC's VAX and PDP-11 series of machines.
Shortly after 9/11, I got a phone call from someone at the Pentagon who was looking for certain parts so they could repair an older VAX that had been damaged in the attack. I was able to get them in touch with a third-party reseller who still had those bits in the back of a dusty warehouse.
It was surprising that they hadn't upgraded to Alpha (which had been out almost ten years) then; the telco where I worked had one big system that had gone through three company changes (DEC -> Compaq -> HP) and had been upgraded in-chassis from VAX to Alpha.
I think all large systems sold to the federal government are required to have service/support available for something like 5 to 10 years after final sale availability; can't find concrete details via Google.
There are lots of systems that were designed around embedded PDP-8s and PDP-11s. And given the numbers of Digital VAX sold and specialized software it would not be very surprising if some of these systems are still be used. There were probably over 1.5 million of these machines sold (about 300,000 PDP-8s, 700,000 PDP-11s, 500,000 VAX machines), so there's probably some happily humming away.
I'm sure the same is true for some earlier IBM 360/370, but they had a better upgrade path and were more expensive to start with. Most of those machines got replaced when they came off lease or the parts availability expired. But probably some of the software from the early 360's is still be used.
Those were the days when machines were rock solid (and weighed about as much). Unlike today, when electronics are designed to be replaced every two years or so.
Well, there are obviously computer systems on the old Voyager probes, which I believe are still in communications with NASA. The earlier Pioneer probes might still be too.
Although my bet is there is an old PDP-8 somewhere that is still in use.
random periods appear
"Scheduled monthlies" are bad enough; I can only imagine how rough "random" must be for those around you...
While we had fun telling visitors to the labs that most of systems running Air Traffic Control were "essentially from the 60s and 70s" that's only technically correct.
The truth is the original HOST mainframes were replaced in the 80s, and then again in the 90s, and now being phased out for ERAM (which is built out of COTS PC parts). In some cases software was brought forward during upgrades, so it may have been possible you'd of been running some assembly code back from the Apollo era but not really. There were still some oddities, though (like the "units" of some fields being in 64ths of a nautical mile), that made interfacing with the ARTCC systems fun.
There might be some ARTS IIA terminals left? Those are vintage early '80s, but I think they've all been replaced with at least IIE or IIIE ('90s) and are all slated to become some form of STARS (late '90s) by 2020 or so.
Maxim: People cannot follow directions.
Increases in truth directly with the length of time spent explaining them
They must have a 360 running.
"A 360" in the sense of "an IBM System/360", or "a 360" in the sense of "a machine compatible - except perhaps at the supervisor-mode level - with an IBM System/360"? If the latter, then that's not interesting; my laptop was purchased in 2015, but, as far as I know, its CPU is compatible, even at the ring 0 level, with the CPU from the original IBM Personal Computer (whether I could boot the original version of MS-DOS on it is another matter), and the most recent CPU design capable of running System/360 problem-state code was also announced in 2015.
Despite the FAA claiming security concerns,, it took me 10 minutes to find Lora
Presumably you meant "Loral".
was once bidding on a contract to replace IBM 9020Es,
Yes, in 1995. IBM replaced the 9020A and 9020D machines in the late 1980s, and then replaced the 9020E's in the 1990s.
so they are probably running 360-compatible gear.
What they're running now might well be "360-compatible", but that might be in the same way that the Haswell Core i7 in my laptop is "8088-compatible".
The DEC manual for the DR11-W is here.
Back in the 1980s my then-employer shared the cost with the FAA of updating IBM's 1401 emulator to the latest architecture (IIRC whatever was the half generation between 370 and 390), so I wouldn't count on that 360 being the oldest!
Where were 1401's involved? The 9020s were based on 360/50 and 360/65 models, not on 1401s; unless 1401s were used prior to the 9020s (was air traffic control computerized at all before the 9020s), and the 1401 software run in emulator mode or under a simulator (I'm not sure the models 50 and 65 had 1401 emulator microcode - 709x, yes, but not 1401, which may have been used in the smaller models 30 and 40), I don't see how 1401s would have been involved. (As I remember reading, at least some of the 360s had microcode changes to add instructions to the S/360 instruction set to process radar scan lines, suggesting that the code was written in 360 assembler or compiled into 360 machine code, rather than 1401 code.)
The VAX was introduced in 1977. It would not have been a tube machine.
Contribute to civilization: ari.aynrand.org/donate
They must have a 360 running.
"A 360" in the sense of "an IBM System/360", or "a 360" in the sense of "a machine compatible - except perhaps at the supervisor-mode level - with an IBM System/360"?
Basically, you just described CICS
While I may be wrong or simply out of date, my understanding is that CICS is an application running in what is essentially a virtual IBM/360. The virtualisation meant that CICS didn't have to be rewritten for the IBM.370 to take into account newfangled things like memory protection.
Protoplasm. Quiet Protoplasm. I like quiet protoplasm.
Reminds me of my first marriage. One night, after twenty years, I discovered that it was indeed broke. A light tap, and the whole thing shattered. The most solid things are often the most fragile. It is wiser to periodically evaluate your universe.
There are a lot of comments here that this FOIA request is a waste of time and money, all for "fun". But if you think about it, this information shouldn't have to be commanded or browbeaten out of millions of government employees. And if the government is operating the way it *should* be, then the gathering, processing and collating of this data should all be routed through a record so that -- voila -- anytime the government, itself, needs to know what it has on hand then it knows where exactly to look and in one place. This would be very handy if the government, itself, started to wonder if entire groups of machines and protocols needed immediate replacement or not. They could have a specialist go through all of this gathered inventory data and give them a realistic budget and timeline. With the data already on hand, that sort of project would be saved the time and effort it took to compile the data in the first place.
Now, on top of that, consider the four stated aims of the report:
* Find which departments are thrifty enough to keep machines in service long past their expiration.
-- how would this not serve as valuable information to the government if it's assessed and quantified in a way that's easy to understand and makes sense?
* Learn the technical dependencies that are holding back the effectiveness of public services.
-- again, how would this not be valuable information?
* Reveal security problems raised by departments that refuse to upgrade from unsupported, obsolete systems.
-- again!
* Critique the allocation and utilization of computer resources by the government
-- which doesn't offer as much usefulness as the other three points, really just generalizes the entire thing, but still there it is. How is this not valuable information?
"Stratigraphically the origin of agriculture and thermonuclear destruction will appear essentially simultaneous" -- Lee
They must have a 360 running.
"A 360" in the sense of "an IBM System/360", or "a 360" in the sense of "a machine compatible - except perhaps at the supervisor-mode level - with an IBM System/360"?
Basically, you just described CICS
While I may be wrong or simply out of date, my understanding is that CICS is an application running in what is essentially a virtual IBM/360. The virtualisation meant that CICS didn't have to be rewritten for the IBM.370 to take into account newfangled things like memory protection.
That wasn't newfangled; System/360 had protection even though all models other than the model 67 had no memory mapping, just physical memory access. Oh machines with storage protection (optional in the smaller models, and standard in the larger ones), a 2KB block of memory had a 4-bit "protection key" associated with it, and the processor status word had a "protection key" value giving the key of the current task (process). A block of memory could be stored into only if either the PSW protection key was 0 or was the same as the protection key on the block; with store-and-fetch protection (not available in some models), the block's storage key had an additional "fetch protection" bit which, if set, disallowed fetching from the block, as well as storing into the block, if the PSW protection key is non-zero and doesn't match the block's key.
And even on early versions of DOS/VS and OS/VS for System/370 models with "dynamic address translation" (i.e., an MMU), all tasks ran in the a single common virtual address space, so it wasn't that different from an S/360 in that regard.
CICS might have done its own multitasking of requests and responses within a DOS or OS task, and not protected individual "subtasks" from each other, but that wouldn't require virtualization, even on MVS where OS tasks ran in separate address spaces, so if you're referring to "The entire partition, or Multiple Virtual Storage (MVS) region, operated with the same memory protection key including the CICS kernel code." from the Wikipedia article, that doesn't require virtualization to run on later systems.
What I'm describing is just that the problem-state instruction set of an IBM z13 is, except for anything removed in the original S/360 to S/370 transition (such as ASCII mode), a superset of the original S/360 instruction set.
So either AC is making it up, or AC isn't that familiar with VAX and the computer (if it's real) was something even older?
That's sort of crazy, to think that somewhere in our government there might be a machine that's older than anybody working around it, that nobody understands even in terms of what function it's processing or even what inputs and outputs it's dealing with, and that if it broke maybe -- who knows -- an entire agency could collapse or a huge part of the infrastructure could fail, or a nuclear war could start or something.
Pretty weird even if it is speculative fiction.
"Stratigraphically the origin of agriculture and thermonuclear destruction will appear essentially simultaneous" -- Lee
Whether it's the appeal to "common sense" or the "if it ain't broke, don't fix it".
People seem to think that software is forever. It isn't. It usually rots from the outside in as OS's are updated, peripherals are replaced, and other external factors kick in. To keep these dinosaurs running require more and more desperate measures as time passes (then again, we're talking government, where expensive solutions are considered just another day). Parts that haven't been made in decades wear out, people who know how it worked die off.
That's not even taking into account stuff like the higher power consumption, lower computing speed, more physical space required for discrete boxes and so forth.
We still have a sparcstation 10 running something somewhere (and by that I mean noone knows what it does anymore and we can only pinpoint its location down to somewhere in a regional office)
Don't take it personal but it's very US-centric. Unless specifically stated otherwise, it's US-related. Most of the time. We don't actually have editors. We just have people who hold that title and get paid for it. I presume they're busy but I'm not actually sure what it is that they do.
However, this isn't going to be something you see change in the near future. It's the way it is, the way it will be, and the way it has always been. Don't let anyone fool you, Slashdot was never "good." If you don't believe me, read some of the old comments or summaries and compare them with today. No, Slashdot was never good. But, it is what it is.
Tromping in and thinking that you're going to effect meaningful change isn't going to do you any good. You'll just find it frustrating. You can't change it and it won't change for you. The momentum is strong, the drive of entropy is stronger. It's all and nothing, at the same time. Just when you think you've got it figured out, you'll find out you were wrong.
Pull up a chair and register an account, if you want. We've new Overlords. As someone said, they bought the site by mistake. I prefer to think they were drunk. So far, so good. They're pretty good people, albeit not the speediest. That's okay, we might not want the speediest. We're a bit risk/change averse. You should see what happens if they move a button. Heh...
As I said, it is what it is. You either like it or you don't. It's actually possible to hold both views at the same time. It's probably best to hold both views at the same time. Don't worry, there's nothing you can do right. Then again, there's not much you can do wrong. No matter what your motives, someone will find fault, almost invariably. As I said to the new owners, "You'll adjust."
"So long and thanks for all the fish."
It's possible there were some tubes in the power supply of older VAXen.
> And if the government is operating the way it *should* be, then the gathering, processing and collating of this data should all be routed through a record so that -- voila -- anytime the government, itself, needs to know what it has on hand then it knows where exactly to look and in one place.
Lol. That's quite the opposite of the US federal government.
In some ways, it's a good thing that the federal government isn't designed for efficiency, that speed and efficiency aren't anywhere on the priority list. As a most obvious example, it would be far quicker and more efficient to have Kim Jong Obama make all of the decisions unilaterally rather than have the whole country debate policy. Heck, it took more than a decade to decide to do Hilarycare, and more than 10 more years to implement it after it was passed. (Including renaming it "Obamacare" along the way.) A dictator would be more efficient, but then you end up with North Korea. So we don't really want efficient government ; we want fair government, we want transparent government, we want big policy decisions to be made carefully, thoughtfully - slowly. Efficiency isn't what we want in government.
I have some tubes of 6100 processor chips. The 6100 is a Intersil/Harris CMOS 12 bit processor that runs the PDP-8 instruction set. They are static CMOS so can be clocked from about 1 Mhz down to DC if you like.
People make PDP-8 machines using these processors. They are standard 40 pin DIP package with early 1980s date codes. There is a popular design that uses the 6120 processor which is the next gen version of the 6100, and there are turnkey designs for a hobby single board computer with that processor. People run PDP-8 software, like the FOCAL time sharing interpreter on these boards.
He needs to qualify his statement: he's clearly looking for digital computers. The Navy (and for all I know other services) has long used analog computers (yes, with gears) for controlling their big guns; we had them on my ship, a DDG retired in 1992. I believe the last of the analog gunfire control computers were retired from the US Navy in the early 2000s, but I could be wrong.
The government holds auctions on old or now unusable equipment from the nuclear reservation Hanford.
At one time I was going to bid $300 (US) on a Univac that took up an entire corner; if I'd of gotten it, it would of taken over the house (it's huge, with many pieces).
I got the bid on another batch of items -when picking them up I asked about the Univac and was told it was pulled as no bid came close to the value of the gold and other heavy metals it could be melted down for.
That was over 25 years ago so imagine it's part of an engagement ring at this time.
Very unlikely. The 'trailing edge' of Tubes in electronics was in RF power applications like large Radio Broadcast equipment. Mundane things like power supply rectifiers and regulators went all solid state by the early 60's for the most part.
The last tubes in consumer electronics were the high voltage rectifiers in TV sets and the CRT (obviously), and probably Guitar Amplifiers for aesthetic reasons (clipping on the output for classic rock guitar distortion)
I really don't know if NDAs involving high security clearance expire.
I was at an government auction with a friend when a SPARCstation came up for bid.. It was him against a contractor who had a customer for it, the final bid cost him $500, the contractor said it was a nice system even with the hard drives they way they were. This was news to the both of us, looking at the two drives installed inside they had taken a bad saw, sawing each hard drive half way through (to the spindle) ; as the system had come out of a secure area.
The computers that run the ICBM equipment in Cheyenne, Wyoming are well over 50 years old. If you're interested in seeing antique computers still in use, go to the 'Fort D.A. Russell Days' event, held annually. They give a tour of one of the missile silos, and will tell you all about how old the computers are, yet they keep using them, because they have no need to change.
Also, in a USDA office somewhere in Western Nebraska (Bridgeport, if I'm not mistaken), I once saw an old IBM server in use. I believe it was a PS/2. I noticed it had a 5 1/4 flippy drive, and asked someone about it, and she told me they use it to store mapping coordinates, and that there was only one person there who knew how to use it (it had a CRT from a PS/1 displaying a CLI), and that they could only use it to print results on an impact printer. That was about 3 years ago, and I'm guessing it's still in use today.
http://youtu.be/1Y1ya-yF35g
Does Voyager 2 on-board computer count? "Reason for not upgrading: cost, too expensive to send a technician to the edge of the solar system."
A while back we did a site visit to a bank mainframe room. They had one backup holdover of every mainframe model they had ever bought going back to day one, just in case they needed to rerun some old piece of software. You felt like you were traveling back in time as you went down the backup row.
He goes on to talk about thriftiness, security, "obsolete", and "holding back"... but he doesn't seem to understand the difference. Those indeed are problems, but that doesn't mean the oldest, non-broken, computer is one of them. Windows XP on a public desktop is old; a mainframe data storage retrieval system might be 20 years older. The Windows box is probably "[more] in need of an upgrade."
This seems like a fun project (finding the oldest government computer system), but I'm not sure the author fully understands what that aphorism truly suggests.
Hire a Linux system administrator, systems engineer,
Yes. Tubes of, I think, 16 chips per tube.
He should ask the IRS and see if they are off 8 track mag tape, yet. [This is no a a joke, they last I heard, they were trying to get off mag tape.]
An engineer who ran for Congress. http://herbrobinson.us
There is a utility for changing MAC address, hostid etc on those old sparcs for the purpose of such things as running legacy software locked to such details and those things are already running it so they are already pretending to be something other than what is on the chip.
We the People of the United States, in Order to form a more perfect Union, establish Justice ... do ordain and establish this Constitution for the United States of America.
If you prefer North Korea-style society, you are welcome to move there.
It is very US-centric. I just write about thing in my country and make occasional reference to "foreigners like Americans". Not many of them even notice, because they simply don't think of themselves as being foreigners.
Birds are not dinosaur descendants;birds are dinosaurs, for all useful meanings of "birds", "are" and "dinosaurs"